Home

Microdata User Guide National Longitudinal Survey of Children and

Contents

1. 0 cece eee eee 34 Examples of Variables Names scrin seuratil canna oh tie eae tedth erate Sh Asters 36 Coding Structure for NLSCY Variables 0 0 c eee cece eens 37 Coding of Open ended Questions 0 06 c cee 40 Naming Imputation cyereresienc Seles fees ge aes ae ee het erieaeetceds 41 Derived Variables oeo sscincstera sonare ace Sexes Be awe Da e na Eos s 41 Chapter 6 Weighting ware cttais we tenia hae atten Oe wale he aa reral Mae dhe anne aa arate 43 Estimation cams ee So tsat stan eae ed Seat eh 1G See ieee ck ones 43 The Longitudinal Sample or Cross sectional 0 0 c cece eee eee 44 Weighting Procedures for the Cross sectional and Longitudinal Samples 45 Cross Sectional Weights for Children Surveyed for the First Time in Cycle 3 48 Weighting of Children Sampled in 1994 and 1996 0000005 50 Chapter 7 NLSCY Concepts and Definitions 00 c cece eee ee 53 Cross Sectional and Longitudinal estimates 00 cece eee eee 53 NLSCY Units of Analyses sy codum nunnana ode hie bebe hous by Gane kore 55 PMK and SPOUSe 4 0 96 951 a D3 ad DE oe ea aai SOG on wi S ae e aoa 56 Family Derived Variables 0 0200 c cece eee eens 59 Socio Economic Derived Variables 0 0 cece eee ee eee eee eee eees 61 Chapter 8 Content and Validation of NLSCY 0 c cece eee eee 71 Validation of Scale Data arrsewna te ee Between
2. For each child under two the nature of the delivery general health of the child at birth and the use of specialized services following the birth were collected in this section The NLSCY also investigated the biological mother s pregnancy and delivery history topics such as the mother s breast feeding experiences and prenatal lifestyle Since birth weight is such an important variable caution was taken in editing this variable The records for children with very low birth weights lt 1 5 kilograms were examined to verify that the response was legitimate Other variables considered in the edit were the length of the baby at birth the number of days early of the delivery the conditions of the delivery e g multiple birth and special medical care and the health of the child at birth If there was nothing to corroborate the low birth weight it was set to not stated NLSCY Data Users Guide 126 2001 2002 There were a couple of derived variables created for this section that Gestational bear note Two variables were derived to indicate the gestational Age of the age of the child CMDCD06 gives the gestational age in days and Child CMDCDO07 indicates if the child was born prematurely gestational age 258 days or less in the normal range gestational age 259 to 293 days or late gestational age 294 days or later A variable was derived CMDCD08 to indicate if the child was of normal birth weight 2500 grams moderately low birth
3. The questions on number of days spent doing things with friends number of friends and getting along with friends parents teachers and siblings BRLCQ01 Q02 Q06 Q09 are based on those in the Ontario Child Health Study The objective of this scale is to measure certain parenting practices Specifically two scales were used The first was designed to measure the positive interaction hostility ineffectiveness and consistency of the parenting of the child The second scale was designed to measure parental practices that may or may not provoke aversion The questions from the Child s Questionnaire used to measure these aspects of parenting are identified in the following paragraphs As mentioned in Section 9 1 complete factor analyses were done on the parenting scales to evaluate the psychometric properties of these scales for the NLSCY population The make up of each factor obtained during these analyses was compared to that which had been indicated in the literature The results of these analyses are presented later in this section NLSCY Data Users Guide 93 2001 2002 Questions Questions Questions BPRC Q1 to BPRC Q18 on positive interaction hostility for the 0 11 or ineffectiveness and on coherence were provided by Dr M Boyle Age Groups of the Chedoke McMaster Hospital based on the work of Dr Ken Dodge Venderbilt University and an adaptation of the Parent Practices Scale of Strayhorn and Weidman For children ages 0 to
4. 70 8 and the numerator portion of the percentage 59 567 in determining the coefficient of variation 3 The numerator 59 567 does not appear in the left hand column the Numerator of Percentage column so it is necessary to use the figure closest to it namely 60 000 Similarly the percentage estimate does not appear as any of the column headings so it is necessary to use the figure closest to it 70 0 4 The figure at the intersection of the row and column used namely 5 0 is the coefficient of variation to be used 5 The approximate coefficient of variation of the percentage of low birth weight babies who were premature is estimated to be 5 0 Since the c v for the estimate falls in the 0 0 to 16 5 range this estimate is acceptable and the finding that 70 8 of low birth weight babies were born prematurely requires no warning note Using NLSCY data it is estimated that 6 1 45 690 753 203 of female babies were born prematurely while 4 9 38 395 791 149 of male babies were born prematurely How does the user determine the coefficient of variation of the difference between these two estimates 1 Using the c v table for the 0 to 3 age group in the same manner as described in example 2 gives the c v of the estimate for female babies as 10 3 and the c v of the estimate for male babies as 10 9 168 2001 2002 Example 4 Suppose now a user wants to compare the number of low birth Estimates of
5. As discussed earlier all of the information for the household collection except for the 10 11 year old and 12 13 year old self completed questionnaires was collected in a face to face or telephone interview using computer assisted interviewing CAI As such it was possible to build various edits and checks into the questionnaire for the various household CAI components in order to ensure high quality of the information collected Various types of computer generated edits were used to check data while the interviewer was completing the interview The NLSCY computer generated survey used the following lt Review Screens lt Range Edits lt Flow Patterns Edits lt Consistency Edits Review screens were created for important and complex information Example The selection procedures for the PMK a critical element of the survey were based on the household roster The household roster screen showed the demographic information for each household member and his her relationship to every other household member The collected information was displayed on the screen for the interviewer to confirm with the respondent before continuing the interview NLSCY Data Users Guide 24 2001 2002 Range Edits Range edits were used for continuous variables to confirm or correct unusual answers during collection Example For the question regarding the weight of a child at birth if a weight entered into the computer was either signific
6. Completed by PMK was asked to complete a series of questionnaires The Parent the PMK Questionnaire for this person and their spouse if applicable a Child s Questionnaire for each child selected in the survey and a computerized consent form about contacting the schools attended by the children The Parent The first part of this questionnaire was completed by both the PMK Questionnaire and his her spouse partner and was designed to gather socio economic and health data about these two individuals Topic areas included education labour force and income The second part of the Parent Questionnaire was completed by and for one of the parents only usually the PMK The purpose was to gather information about the child s family environment notably the mental health of the PMK and family functioning NLSCY Data Users Guide 15 2001 2002 Child s Questionnaire The Child s Questionnaire was completed for selected children in the household aged newborn to15 years Topic areas included health birth information temperament behaviour education activities literacy social relationships parenting and legal custody of the children For each child who attended school in 1997 98 the PMK also answered a computerized questionnaire in which his her consent was requested to a contact the child s teacher and the school principal and b administer a test of about 45 minutes measuring the child s mathematics computation and reading compr
7. Depression Scale PMK Introduction Symptoms of Depression Questions for the Depression Rating Scale The depression scale was administered to the PMK as part of the Parent Questionnaire Questions for this scale BDPPQ12A to BDPPQ12L are a shorter version of the depression rating scale CES D comprising 20 questions developed by L S Radloff of the Epidemiology Study Center of the National Institute of Mental Health in the United States This rating scale is used to measure the frequency of symptoms in the public at large The occurrence and severity of symptoms associated with depression during the previous week are measured The rating scale was reduced to 12 questions by Dr M Boyle of the Chedoke McMaster Hospital of McMaster University This rating scale is aimed at gathering information about the mental health of respondents with particular emphasis on symptoms of depression Several members of the NLSCY advisory group of experts pointed out that the best way of proceeding was to measure one particular aspect of the PMK s mental health instead of trying to measure overall mental health It was proposed that this section focus on depression for the following reasons depression is a prevalent condition it has been demonstrated that depression in a parent affects the children present research on this subject is generally based on demonstration groups and not on population samples and it is felt that introducing polici
8. Example of Using C V Tables to do a T Test Example Let us suppose we wish to test at 5 level of significance the hypothesis that there is no difference between the proportion of low birth weight female babies and that of low birth weight male babies From example 3 Section 12 1 1 the standard error of the difference between these two estimates was found to be 008 Hence X 0 061 0 049 0 012 0 0 008 0 008 Since t 1 5 is between 2 and 2 no conclusion at the 0 05 level of significance can be made regarding the difference in proportions of low birth weight male or female babies Coefficients of Variations for Quantitative Estimates Quantitative Estimates NLSCY Data Users Guide For quantitative estimates special tables would have to be produced to determine their sampling error Since most of the variables for the NLSCY are categorical in nature this has not been done As a general rule however the coefficient of variation of a quantitative total will be larger than the coefficient of variation of the corresponding category estimate If the corresponding category estimate is not releasable the quantitative estimate will not be either For example the coefficient of variation of the total number of days of special medical care received for low birth weight babies would be greater than the coefficient of variation of the corresponding proportion of babies who were of low birth weight Hence if th
9. Individuals with missing values were excluded from the analysis conducted for the purpose of constructing the factor After these exclusions The sub samples contained 1 680 and 1 709 individuals respectively for analysis purposes No imputation took place As a result of factor analysis one factor was identified the activities factor AACCS6 Items AACCQ6A AACCQEE loaded into the factor Cronbach s Alpha for Activities Cronbach s alpha coefficients raw values were calculated with SAS using the normalized weighted data Please note that in general Cronbach s alphas calculated with SAS are lower than those produced by the SPSS software package The Cronbach alpha for the activities score was 0 778 The item that affects the factor the most is AACCQ6B If it were removed from the analysis the Cronbach s alpha would drop to 0 705 The final activities score could not be calculated for 45 1 3 individuals due to missing values for the items comprising this factor My Parents and Me Scale BPRCbS07 and BPRCbS08 Parent The objective of the My Parents and Me scale is to measure the parent s perception of his her relationship with his her child This was asked only for children 12 or 13 years of age Belowis a description of the items that were included in the My Parents and Me section of the parent report questionnaire to measure family relations the analysis used to construct the scale and the results of these analy
10. Microdata User Guide National Longitudinal Survey of Children and Youth Cycle 3 September 1998 to June 1999 Eeg gess Statstoue Canada Table of Contents Chapter 1 Introduction i Senco Soeur wc Sinha oul eewerwbembarda howard ads 1 General Inionmalono 208 socscb qnicdoese sie dbdeemi coupe eeren 1 Chapter 2 Survey Methodology 00 cece eee eee eee 3 VCE Be hale stir tal a ten Rd py ek big Gate Shh ghee Onc bi tals State O 5 Chapter 3 Response ales sg eei 1 oc tee dai en eae Send eee eer tt 7 Chapter 4 Data Collection 222 4 setesd ol eceoc sev te ye eee et oie eee eee 11 Computer Assisted Interviewing ace need eset sae eeeas exw 11 Ho s old Collectif 6 o cc6 2 t nn Sore eset pies nun Bae eee E et 15 School Collection wie ceeGri atv elute eA Sih ete Wiad ste EA ste 20 Interview Training Supervision and Control 6 00 cece eee eee 22 Chapter 5 Data Processing sxsate seed Scots ee Sie e s dk g ev eee Se CE ee oe 24 Eding esai Kav eccew ne cence av da ee ETE TEET ENET os CEPE ae e cere eo 24 Data Capt re corte Se tle ea cele Oi EPE te oe tts Pett ee EEE RA 26 Minimum Completion Requirements 00 000 c cece eect eee 27 Head Office BCI est er a ates tae et ts E wane Sac ee SE ae eed ace ee 29 Consistency Editing e nent aawkcets extn ests epee eae Gash eased 31 Naming Convention and Coding Structure for NLSCY Variables 32 Acronym Names for Questionnaire Sections
11. Please note that in 13 Year general Cronbach s alphas calculated with SAS are lower than Olds those produced by the SPSS software package Cronbach s alphas for these factors are given in the table below CRONBACH S ALPHA VALUES FOR MY PARENTS AND ME SCALE 12 13 YEAR OLDS PARENT REPORT Factor Items that lowered Cronbach s cronbach s alpha the alpha if the item most if excluded is excluded Parental Rejection BPRCQ29M BPRCbS08 NLSCY Data Users Guide 107 2001 2002 Child Scales from Self completed Questionnaire Friends and Family self complete 10 13 Information Collection Peer Relations Sub scale Friends Scale BFFCS01 Questionnair e Items Analysis of the NLSCY Data NLSCY Data Users Guide 108 Friends and Family was one of the sections on the questionnaire completed by children in the 10 to 13 age group The objective was to determine how well the child felt he she was getting along with others The section collected information on numbers of close friends time spent with friends presence of someone the child can confide in and the quality of relationships with others such as parents peers and teachers This information is important in identifying the extent and quality of the child s social support network To allow for comparison the section includes questions which are also included on the Child s Questionnaire completed by the PMK There was one group of questions in th
12. Scales BAMCS01 BAMCS02 Factor Scores The objective of the About me scale is to measure the child s overall self esteem and perception of physical appearance Specifically two scales were used one was designed to measure overall self esteem and the other was designed to measure perceptions of physical appearance In Cycle 2 the factor scores were derived based on the factorial structure identified in Cycle 1 Below is a description of the items that were included on the questionnaire to measure these scales the analysis used to construct the scale and the results of these analyses all from Cycle 1 NLSCY Data Users Guide 117 2001 2002 Questionnaire Items Analysis of the NLSCY Data Missing Values Results In Cycle 1 questions AA1CQ01A to AA1CQ01D on overall self esteem were taken from the General Self Scale of the Marsh Self Description Questionnaire developed by H W Marsh Questions AA1CQ01E to AA1CQ01H on perceptions of physical appearance were taken from the Physical Appearance Scale of the Marsh Self Description Questionnaire developed by H W Marsh To construct the About me Scale for the NLSCY a factor analysis was conducted to test the theoretical construct In the factor analysis the items were multiplied by the child s normalized weight An individual s statistical weight is normalized by dividing his her weight AWTCWO01 by the average weight of all individuals Consequently the sum of the normal
13. To determine the prevalence of various risk and protective factors for children and youth gt To understand how these factors as well as life events influence children s development gt To make this information available for developing policies and programs that will help children and youth gt Collect information on a wide variety of topics biological social economic gt Collect information about the environment in which the child is growing up family peers school community gt Information comes from different sources parent child teacher and from direct measures PPVT math reading tests etc Cycle 4 data will be released All questions about the data set or its use should be directed to Lecily Hunter Project Manager NLSCY Special Surveys Division Statistics Canada 7 C8 Jean Talon Building Tunney s Pasture Ottawa Ontario K1A OT6 Telephone 613 951 0597 Facsimile 613 951 7333 Internet huntlec statcan ca Toll free 1 800 461 9050 The contact person for Human Resources Development Canada is Susan McKellar NLSCY Project Coordinator Applied Research Branch Human Resources Development Canada Place du Portage Phase II 165 H tel de Ville Hull Qu bec K1A 0J2 Telephone 819 953 8101 Facsimile 819 953 8868 Internet susan mckellar spg org 2001 2002 Chapter 2 Survey Methodology Definition of The NLSCY survey population consists of two sample groups t
14. included both married and common law partners Detailed socio economic information was collected about the spouse partner in order to describe the family situation of the child See the table below These numbers for the PMK and spouse partner are based on unweighted data NLSCY Data Users Guide 56 2001 2002 The following is the breakdown of the relationship of the spouse partner to the children For families where the PMK s The was did not have a spouse partner residing in the household 14 8 Spouse partner was the father 78 9 was the biological father 74 8 was the step adoptive or foster father 4 1 spouse partner was the mother biological step adoptive or foster 5 4 spouse partner was not a parent For several reasons the PMK and his her spouse could be different Change in people than those designated in the first and second cycles For this PMK reason a variable flagging the change in individual on the longitudinal Between file was created see CDMPcD27 for the PMK change and Cycles CDMScD28 for the change in spouse This new variable indicates whether there was any change in the PMK from one cycle to the other It is highly recommended that this variable be used when doing longitudinal analyses involving the characteristics of the parents Here is a breakdown of the consistency of the relationship between the NLSCY children and the PMK and his her spouse a new individual had no spouse
15. 2 to 15 year old child ee ooo Ea Table 11 There is a possibility that this name will not correspond to the questionnaire in the present cycle given that we keep the same names of variables in the data dictionary This usually happens when a number or section changes from one cycle to another For example cmdcbq31 corresponds to question 3 in the section on Working After Birth whereas in Cycle 2 it corresponded to question 31 in the Medical and Biological Information section NLSCY Data Users Guide 36 2001 2002 Variable Refers to Name CHLCbZ3 a flag that indicates an inconsistency in the child s height between the current and previous cycles a Cycle 2 variable i lt lt Coding Structure for NLSCY Variables Introduction Some standards have been developed for the coding structure of NLSCY variables in order to explain certain situations in a consistent fashion across all variables The following describes these various situations and the code used to describe the situation During a CAI interview the respondent may choose to refuse to provide an answer for a particular item The CAI system has a specific function key that the interviewer presses to indicate a refusal This information is recorded for the specific item refused and transmitted back to Head Office Refusal On the NLSCY data file an item which was refused is indicated by a code 8 For a variable that is one digit long the code
16. 23 months only questions APRCQ1 to APRCQ7 were asked Questions Questions BPRC Q19 to BPRC Q25 which measure parental for the 2 11 practices which may or may not cause aversion were provided by Years Age Dr M Boyle Groups Analysis of NLSCY Data The factor structure of each scale was determined based on data Factor from the first cycle The factor structure imposed on the scales Structure already used in the first cycle and repeatedly used in the second cycle of the survey was the result of analyses done based on data from the first cycle To conduct the analysis on the parenting scales for the NLSCY data a factor analysis was conducted on the scale for the 0 to 23 months age group and the two scales for the 2 to 11 age group separately New factor structures emerged which are described in the Results Section below NLSCY Data Users Guide 94 2001 2002 Once the factor structures were analysed and the items included in each factor were determined scores were calculated To produce the scores 1 was subtracted from each item so that the lowest possible score value would be 0 For each of the four factors a score of 0 indicates New Factor Structures the absence of positive interaction for the positive interaction factor the absence of hostile ineffective interaction for the hostile ineffective factor the absence of consistent parenting for the consistency factor the absence of punitive interaction or aversion produ
17. 6K 6Q 6V 6CC 6ll 6MM 6RR Physical aggression conduct Ta ABECQ6G 6X 6AA 6FF 6JJ disorder 9 6NN Indirect aggression ABECS1 ABECQ6J 6R 6Z 6LL 6TT 0 Property offence ABECS1 ABECQ6C 6E 6L 6T 6DD 1 6PP NLSCY Data Users Guide 90 2001 2002 Cronbach s Cronbach s alphas for these factors are given in the table below Alpha for 4 Normalized weighted data were used in the computations to 11 Year Olds CRONBACH S ALPHA FOR THE BEHAVIOUR SCALE FOR 4 TO 11 YEAR OLDS FACTOR CRONBACH ITEM THAT LOWERS CRONBACH S ALPHA CRONBACH S S ALPHA IF RAW ALPHA THE MOST IF THE ITEM IS IT IS EXCLUDED EXCLUDED Hyperactivity inattention 0 838 ABECQ6l 0 810 Prosocial behaviour 0 816 ABECQ6BB 0 789 ABECS07 Emotional disorder ABECQ6II Physical aggression 0 770 ABECQ6AA 0 716 conduct disorder ABECS09 ABECS10 ABECS11 The scores for these factors could not be computed in 338 647 324 358 814 and 310 cases respectively because of unreported values Motor and Social Development Objective for The Motor and Social Development Section of the Child s Motor and Questionnaire was completed for children in the 0 to 3 age group Social The objective was to measure motor social and cognitive Development development of young children A scale was used to assess these Section concepts BMSCQ01 to BMSCQ48 NLSCY Data Users Guide 91 2001 2002 Overview of the Motor and Social Develo
18. AN A ILE ABECB10 LIKEWISE WHEN THE SCALE IS NEW IN CYCLE 3 THE SCALE ANALYSIS WILL CITE CYCLE 3 VARIABLES Validation of Scale Data Scale Definition For some of the concepts deemed important to measure in the NLSCY it was decided that the concept would most appropriately be measured through the use of a scale A scale is simply a group of questions or items that measure a certain concept when the answers to the items are put together For example on the child s questionnaire it was determined that it was important to have an assessment of certain parenting behaviours The Parenting Scale that was employed was one that was proposed by Dr M Boyle at Chedoke McMaster Hospital based on work by Dr Ken Dodge Vanderbilt University which was an adaptation of Strayhorn and Weidman s Parent Practices Scale The scale is intended to measure three different constructs or factors related to parenting positive interaction hostile ineffective parenting and consistent parenting NLSCY Data Users Guide 72 2001 2002 Scales and For each factor measured by a scale a score is calculated The Calculations score for a particular factor can be used to give an ordering of individuals For example for the Parenting Scale for children with higher scores for the positive interaction factor the PMK reported having more positive encounters with the child e g laughed with them more praised them more etc The score for a part
19. Questionnaires Returned as Fog 6 o 175 _ Ps a7 1094 9 1 3 1024 There were 21 children that are 3 years old and 4 children that are 16 years old The 3 year olds were regrouped with the 4 year olds and the 16 year olds were regrouped with the 15 year olds NLSCY Data Users Guide 145 2001 2002 The response rate for the Teacher s Questionnaire appeared to be lower for four year olds The percentage of questionnaires returned and the school board consent rate were also lower for four year old students Response Rates by age Principal s questionnaire As in the case of the Teacher s Questionnaire when we had the permission of the parents and the school boards we mailed a questionnaire to the school principal In cases where more than one child was attending the same school the principal received only one questionnaire The response rate for the Principal s Questionnaire was 68 This means that we collected Principal s Questionnaires for 69 of the children whose parents and school boards gave permission or 63 of all children attending school Principal s questionnaire Response Rates by province Principal s Questionnaire Received a Questionnaire Frequency The province of the school is missing for 3 principals NLSCY Data Users Guide 146 2001 2002 Principal s Questionnaire Province Consent of Consent of School Board Children whose Principal Parent have return
20. and opposition were omitted and indirect aggression and some aspects of conduct disorder were added The following indicates the items that were included on the questionnaire to measure these various constructs of behaviour As discussed in Section 9 1 a complete factor analysis was carried out for the behaviour scale to assess the psychometric properties of this scale for the NLSCY population As part of this analysis the items that loaded into each construct or factor were compared to the expected result described below The results of this analysis are presented later on in this section Below are the theoretical constructs used for the factor analysis The actual scales that emerged from the analysis vary from these constructs 85 2001 2002 Two and three year olds Conduct disorder Items include BBECQ6G from the Ontario Child Health Study OCHS Hyperactivity Items include BBECQ6B Q6l Q6N Q6P Q6S and Q6W from the OCHS and ABECQ6HH from the Montreal Longitudinal Survey Emotional disorder Items include BBECQ6F Q6K Q6Q Q6V Q6CC Q6MM and Q6RR from the OCHS Anxiety Items include several of the OCHS emotional disorder questions BBECQ6F Q6Q Q6V and Q6CC Physical aggression Items include BBECQ6X from the Montreal Longitudinal Survey and BBECQ6G from the OCHS Inattention Items include BBECQ6P from the OCHS and ABECQ6EE Q6KK and Q6QQ from the Montreal Longitudinal Survey Prosocial behavio
21. can determine whether the number of smokers increased or decreased However this comparison conceals the fact that a number of young people quit smoking in the interim From this analysis it would therefore not be possible to verify whether a program designed to reduce the number of young people who smoke is effective Again using our example the cross sectional sample would make it possible to quantify each transition and therefore to calculate the gross flows NLSCY Data Users Guide 44 2001 2002 Weighting Procedures for the Cross sectional and Longitudinal Samples NLSCY Weighting Strategy Weighting of Longitudinal Sample Step One Adjustment Factor The NLSCY weighting strategy is based on a series of cascaded adjustments applied to a basic or initial weight Conceptually the basic weight of each child is approximately equal to the inverse of the child s probability of selection In the case of the selected households of the LFS in 1996 the basic weight was the sub weight calculated by this survey For the longitudinal children that is those sampled in 1994 the basic weight was determined using the weight calculated for Cycle 1 The final weight cross sectional or longitudinal was obtained by multiplying the basic weight by many adjustments This section explains the various corrections made to the basic weight and the procedures used to weight the cross sectional and longitudinal samples We will discuss the longi
22. grade 1 or less included those over 6 years of age the PPVT was administered to the child About Me 10 to 13 Self complete Questionnaires Section C Feelings and Behaviour 10 to 13 Self complete Questionnaires Section D My Parents and Me 10 11 questionnaire Section E 12 13 questionnaire Section G PU Puberty 10 to 13 Self complete Questionnaires Section F 12 13 year Health questions in Section H Smoking drinking and 10 11 questionnaire Section G 12 13 drugs questionnaire Section F Activities 10 11 questionnaire Section H 12 13 questionnaire Section E HT Health 12 13 Self complete Questionnaire Section H WK Work and Sources of 12 13 Self complete Questionnaire Section Mone DA Dating 12 13 Self complete Questionnaire Taken from questions in the Family and Friends and the Health Sections EP Principal s Education Child s Principal about the school and the resources available to the staff Teacher s Education Child s Teacher about the child and the classroom environment Reading test children in grade 2 and over MA Math computation test children in grade 2 and over NLSCY Data Users Guide 35 2001 2002 Examples of Variables Names In order to illustrate the naming convention used for variables included on the NLSCY data file the following examples are given Name Q anitemasked directly onthe questionnaire Variable Refers to Name ae a positive interaction score on the parenting scale for a
23. in Cycle 1 an average MSD score of 100 children who are 1 month old had an average MSD score of 100 and children 47 months old had an average MSD score of 100 Using a standardized score BMSCS02 makes it possible to compare scores of children across the 0 to 3 age group not controlling for age In the previous cycle the name was based on the cycle 1 child However since the number of children is not very big we decided to create new names in cycle 3 based on the combination of the scores from cycle 1 2 and 3 The standardized scores from cycle 1 and 2 will be recalculated based on these names NLSCY Data Users Guide 92 2001 2002 Relationships Objective Questions from the Ontario Child Health Study Parenting Scale Objective The Relationships Section of the Child s Questionnaire was completed for all children 4 years of age and older The objective was to provide information about the child s relationships with others Positive relationships with other children and adults may help to counteract other factors which place a child at risk The section collects information about how the child gets along with parents brothers and or sisters teachers friends and classmates with some variation by age of the child Parents knowledge of the names of the friends of 8 to 13 year olds is also investigated along with their perception of these other children s behaviours and whether their own child is shy or outgoing
24. in Cycle 2 NLSCY Data Users Guide 4 2001 2002 Cycle 3 Sample Size In total 38 035 children were sampled in Cycle 3 Of those sampled children 1 089 8 were out of scope either because the respondent had moved permanently outside of Canada or because the household did not contain a child who was eligible to complete the NLSCY The sample size of the 1 and 5 year old children for Cycle 3 was increased from that used in previous cycles This was due to the federal government s 1997 Speech of the Throne which outlined the intent to have measures of the early years and the commitment to report on the measures This enabled us to produce provincial estimates of readiness to learn for the 5 year old children Cycle 3 Children from the age group of 0 to 11 months were taken from the Cross LFS sample Unlike Cycle 2 no siblings of the longitudinal cohort sectional were selected Sample of Children 2086 households representing 2 123 children were added to the NLSCY sample in Cycle 3 Seven households were selected for the 0 11 Months LFS sample but were excluded from the NLSCY sample since these households were already in our sample 37 households had twins It should be noted that these children were sampled to ensure that they would be 0 11 months old at the time of the interview and that collection work for this component began in October 1998 and ended in July 1999 NLSCY Data Users Guide 5 2001 2002 NLSC
25. in the survey and also for periodically monitoring their interviewers and reviewing their completed documents Senior interviewers ensure that prompt follow up action is taken for refusal and other non response cases If necessary non response cases were transferred to the senior and reassigned The senior interviewers are in turn under the supervision of the LFS program managers located in Statistics Canada regional offices For the NLSCY a combination of classroom training and self study materials were prepared to ensure that interviewers had a proper understanding of survey concepts Self study e involved the interviewers reading the Interviewer s Manual prepared for the survey and completing home study exercises Classroom a program manager or a senior interviewer presented an overview of the survey went through a mock interview with the participants gave more specific training on administering the PPVT R and presented exercises to help interviewers minimize non responses In total 14 hours were devoted to these training activities for each interviewer NLSCY Data Users Guide 22 2001 2002 NLSCY Data Users Guide 23 2001 2002 Editing Introduction Computer Generated Edits Types of Computer Edits Review Screens Chapter 5 Data Processing The main output of the NLSCY is a clean master data file This section presents a brief summary of some of the processing steps involved in producing this file
26. interviewer recorded in words the answer provided by the PMK At Head Office these written descriptions were coded into industry and occupation codes to describe the nature of the work of the PMK Similar information was collected for the spouse partner and codes assigned to describe the nature of the work How they are coded The coding systems used were the 1980 Standard Occupational Classification codes SOC and the 1980 Standard Industrial Classification codes SIC Grouped versions of these codes are available on the data file CLFPDO7 and CLFPD08 for the PMK and CLFSD07 and CLFSD08 for the spouse partner NLSCY Data Users Guide 40 2001 2002 Naming Imputation Missing Variables Imputation For various reasons there are certain variables that may be missing for responding households on the NLSCY file This is usually referred to as item non response Earlier in the chapter the various codes that have been used to describe the reason for the item non response refusal don t know not stated are described For some variables on the NLSCY file however rather than using a special non response code imputation has been carried out Imputation is the process whereby missing or inconsistent items are filled in with plausible values For the NLSCY imputation was carried out for household income and PMK income Imputation flags have been included on the NLSCY file so that users will have information on the extent of imp
27. kindergarten questionnaire e a teacher questionnaire for students who had one teacher for the basic academic subjects e a different teacher questionnaire for students who had different teachers for the basic academic subjects NLSCY Data Users Guide 20 2001 2002 The Principal s Questionnaire The goal of the Principal s Questionnaire was to gather information on the school environment in order to assess how this may impact child development Consequently the Principal s Questionnaire collected information on school policies resources and educational climate rather than data about a specific child The Math The math portion of the skills test to be administered to the child was a Computation shortened version of the Mathematics Computation Test of the and Reading standardized Canadian Achievement Tests Second Edition CAT 2 Comprehensi The CAT 2 is a series of tests designed to measure achievement in on Test basic academic skills Some of the test s questions on reading comprehension are taken from the CAT 2 test and some are new questions developed for the NLSCY NLSCY Data Users Guide 21 2001 2002 Interview Training Supervision and Control Interviewers The NLSCY was conducted by Labour Force Survey interviewers All Training LFS interviewers are under the supervision of a staff of senior interviewers who are responsible for ensuring that interviewers are familiar with the concepts and procedures involved
28. l 91 0 64 1 i 91 2 66 9 91 2 66 2 59 6 89 0 59 5 88 0 592 5 f 94 1 913 63 0 There were 21 children that are 3 years old and 4 children that are 16 years old The 3 year olds were regrouped with the 4 year olds and the 16 year olds were regrouped with the 15 year olds 88 8 Response Rates As in the case of the Teacher s Questionnaire the response rate for by age the Principal s Questionnaire appeared to be slightly lower for four year olds The school board consent rate was also lower for four year old students NLSCY Data Users Guide 148 2001 2002 NLSCY Data Users Guide 149 2001 2002 Chapter 10 Guidelines for Tabulation Analysis and Release Introduction This section of the documentation outlines the guidelines to be adhered to by users tabulating analyzing publishing or otherwise releasing any data derived from the survey microdata file With the aid of these guidelines users of microdata should be able to produce the same figures as those produced by Statistics Canada and at the same time will be able to develop currently unpublished figures in a manner consistent with these established guidelines Rounding Guidelines Statistics In order that estimates for publication or other release derived from Canada the NLSCY microdata file correspond to those produced by Guidelines Statistics Canada users are urged to adhere to the following guidelines regarding the rounding of such estimat
29. of the child The second group of households included 2 087 new households with children aged 0 11 months 7 932 new households with children 1 year of age and 6 952 new households with children aged 5 years For these households the initial contact procedures were the same except for the fact that no tracing was done for people who had moved Households were updated and the interviewer gathered the demographic data and relationships After this stage if there were no eligible children in the household the interview ended otherwise it continued in the same way as for the households in the first group with questions asked about dwelling conditions and the selection of the PMK 2001 2002 Household Collection Household There were three collection periods for the household collection Collection Period 1 November and December 1998 2 February and March 1999 3 April May 1999 The overall sample was split evenly among the three collection periods and each period lasted approximately six weeks The For the household collection data were collected from a variety of Household respondents using different data collection instruments Except for the Collection questionnaires asked of 10 to 15 year olds all of the information for the household collection was collected in a face to face or telephone interview using computer assisted interviewing CAI Instruments After completing the contact and demographic data questionnaire the
30. or on the due date Categorical Questions R Before After On due date Q Compared to other babies in general would you say the the child s health at birth was R Excellent Very good Good Fair Poor Quantitative Quantitative estimates are estimates of totals or of means Estimates medians and other measures of central tendency of quantities based upon some or all of the members of the surveyed population They also specifically involve estimates of the form where X Y is an estimate of the surveyed population total quantity and Y is an estimate of the number of people in the surveyed population contributing to that total quantity Example of a An example of a quantitative estimate is the average number of Quantitative days of care received by babies who required special medical Estimate care following birth The numerator is an estimate of the total number of days for which babies required special care The denominator is the number of babies who required special care at birth NLSCY Data Users Guide 153 2001 2002 Example of a Quantitative Question Q For how many days in total was this care received R Days Q What was the child s weight at birth in pounds and ounces R Pounds Ounces Tabulation of Categorical Estimates Estimates of the Number of Children Proportions and Ratios NLSCY Data Users Guide Estimates of the number of children with a certain characteristic can be obtained from the mi
31. order to improve the translation Other items which had a negative biserial correlation with the ability being measured were also changed to improve their fit into the logistic model The Bilog MG software was used to calculate the scaled scores for both the reading and mathematics test NLSCY Data Users Guide 143 2001 2002 Response Rates for Teachers and principals questionnaire Response Rates The response rates for the Teacher s Questionnaire and the Principal s Questionnaire were computed by province type of school and the child s age In Cycle 3 the parents of 16 558 children reported that their children were attending school The children ranged in age from 4 to 15 and in level from Kindergarten to Grade 10 The parents of 94 of the 16 558 children gave us written permission to collect data from their children s teacher and principal The school boards of 97 of the children whose parents had given permission that is 91 of all children attending school agreed to take part in the survey After obtaining parental and school board consent we sent a questionnaire to each child s teacher The response rate for the latter questionnaire was 67 Taking into account the questionnaires that were never mailed to teachers because we were unable to obtain parental or school board permission we collected questionnaires from the teachers of 61 of all children attending school Response Rates by province Teacher s Questionnai
32. sample This augmented sample will allow for ongoing cross sectional analyses to supplement the primary longitudinal research As such at each cycle it will be possible to get a snapshot of Canadian children of all ages At the present time it is not planned to follow this augmented component of the sample longitudinally or if so it will be done on a limited scope NLSCY Data Users Guide 54 2001 2002 Children Who Immigrate to Canada It should be noted that the children who immigrate to Canada at any point in time after the Cycle 1 sample was selected and who are in the age cohorts covered in the Cycle 1 sample will not be included in either cross sectional or longitudinal estimates at this time The number of children excluded by this criterion is small Estimates of the number of children immigrating to Canada will be monitored and a decision may be made in the future to introduce a new sample into the NLSCY to cover these children NLSCY Units of Analyses Unit of Analysis Child Defining Longitudinal Households The unit of analysis for the NLSCY is intended to be the child and eventually the young adult For each cycle of the NLSCY extensive information will be gathered on the child s family parent s and neighborhood It is true that families or households are relatively straightforward units of analysis with cross sectional data but the situation becomes problematic with longitudinal data Households change
33. standard errors These different degrees of confidence are referred to as the confidence levels Confidence intervals for an estimate are generally expressed as two numbers one below the estimate and one above the estimate as where k is determined depending upon the level of confidence desired and the sampling error of the estimate Confidence intervals for an estimate can be calculated directly from the Approximate Sampling Variability Tables by first determining from the appropriate table the coefficient of variation of the estimate and then using the following formula to convert to a confidence interval Cl IC X Kay X Kay ye ae ae where x is the determined coefficient of variation x and NLSCY Data Users Guide 170 2001 2002 t 1 if a 68 confidence interval is desired t 1 6 if a 90 confidence interval is desired t 2 if a 95 confidence interval is desired t 3 if a 99 confidence interval is desired Note Regarding Release Guidelines Release guidelines applying to estimates also apply to confidence intervals For example if the estimate is marginal then the confidence interval is marginal and should be accompanied by a warning note to caution subsequent users about high levels of error Example of Using the C V Tables to Obtain Confidence Limits Example A 95 confidence interval for the estimated proportion of babies who were of low birth weight would be calculated as follows Estim
34. sub samples of 7 073 and 7 153 were created for analysis The item non response rate was approximately 2 1 for most of the 47 items involved in the analysis Individuals were excluded from the analysis on the basis of the following criteria individuals with eight or more items coded missing individuals with one or more refusals individuals with two or more missing items under prosocial behaviour conduct disorder hyperactivity anxiety and emotional disorder and individuals with one or more missing items for the other factors After the criteria were applied 6 620 and 6 683 individuals remained in the sub samples to be analysed Data were imputed for 26 items The number of imputations ranged between 1 and 159 for those 26 items A total of 363 values were imputed NLSCY Data Users Guide 89 2001 2002 Six Factors Six factors were identified for this age group hyperactivity for 4 to 11 inattention ABECS06 prosocial behaviour ABECS07 Year Olds emotional disorder anxiety ABECS08 physical aggression conduct disorder ABECSO9 indirect aggression ABECS10 and a new factor property offence ABECS11 The items making up each factor are listed in the table below BEHAVIOUR SCALE FOR 4 TO 11 YEAR OLDS FACTOR SCORE ITEMS Hyperactivity inattention ABECSO ABECQ6B 6l 6N 6P 6S 6W 6 6HH 6QQ Prosocial behaviour aia ABECQ6A 6D 6H 6M 6U 6BB 6GG 600 6SS 6UU Emotional disorder anxiety oe ABECQ6F
35. table This rate does not provide an indicator of the quality of cross sectional estimates as such an indicator would account for the non response rate in previous cycles Instead the rates shown below reflect the efficiency of the data collection process in Cycle 3 NLSCY Data Users Guide 129 2001 2002 NLSCY Cross Sectional Response Rate by Province Province Households Respondent Response Rate deter Satin ca rad Sone ___Contacted Households _ _ __ _ Newfoundland d WZB deusas TOT a ONO sce Prince Edward Island _______ 1 030 ONTE 948 a ae eee Nova Scotia 2 235 2 018 90 e a a Si SERES S E E E E ESE E E New Brunswick aaan 2 181 Ts acs 1 954 190 Quebec 6 963 6 294 90 ee bes See anen Wai asi SS EE eh ees Ss Se ae ss a Oak Ontario 10 501 8 651 82 4 _ 4 _ 4 Manitoba 2 528 2 250 89 oo Se eee eS at Seek Se ee a a See ee eee a eee ee ee ee Saskatchewan 2 619 2 306 88 TR E Mig eo tas oe SS i nd m a Sle ea le et Al ete nt RNa ie De Alberta 3 583 3 117 87 Fa ee ene eT 4 British Columbia E SE ES Poe 2 813 Tera B a TOTAL 36 736 31 963 87 The cross sectional sample included longitudinal households sampled in Cycles 1 and 2 as well as households contacted for the first time in Cycle 3 newborn children selected from th
36. the respondent decided that she he wished to terminate the interview If there was enough information collected to establish this household as a responding household then all remaining items on the questionnaire and on questionnaires that had not yet been started were set to not stated The one exception was that if it was known that a certain section or a certain questionnaire was not applicable then these questions were set to not applicable 3 The third situation in which not stated codes were used was as a result of consistency edits When the relationship between groups of variables was checked for consistency if there was an error often one or more of the variables was set to not stated For derived variables if one or more of the input variables to the derived variable had a refusal don t know or not stated code then the derived variable was set to not stated NLSCY Data Users Guide 39 2001 2002 Coding of Open ended Questions Open ended A few data items on the NLSCY questionnaire were recorded by Format interviewers in an open ended format For example in the Labour Force Section a PMK who had worked in the previous 12 months was asked a series of open ended questions about the current or most recent job lt What kind of business service or industry is was this lt What kind of work are were you doing lt At this work what are were your most important duties or activities How they are recorded The
37. to cooperate b our inability to contact the respondent This second cause may be the result of a move or of a temporary absence when attempts at contact were made The model must therefore include two distinct phenomena mobility and cooperation Regional Separate models have been developed for each region in the Models country in order to take into consideration the characteristics of each one Note that the decision to cooperate or not in a survey is made by an adult As a result the explanatory variables for these models are in fact characteristics of adults Without entering into the details of each regional model here are some of the conclusions that were drawn People with a lower income show lower response rates than people with a higher income People with a lower level of education show lower response rates than people with a higher level of education People living in a large city show lower response rates than people living in smaller cities The presence of a spouse in the household is associated with better response rates NLSCY Data Users Guide 133 2001 2002 Weighting Process Other Sources of Bias First Problem Second Problem In order to minimize the risk associated with this potential bias the models were used for the weighting process see Section 7 This technique helps correct sampling weights in order to account for the potential bias resulting from non response However it does not g
38. was a single or two parent family NLSCY Data Users Guide 69 2001 2002 In order to give a flavour for the types of families associated with various SES scores the following examples are given for illustration purposes It should be noted that the SES scores given in these examples are approximate and do not correspond to actual records on the NLSCY file Many more examples are possible for each score involving both one and two parent families Types of Families SES SCORE EXAMPLE sectional A Family in Which both the PMK and spouse have a university degree BA BSC they are both employed professionals the household income is 80 000 the PMK has a university degree BA BSC and the spouse has grade 13 the PMK is employed as a semi professional and the spouse is employed in a semi skilled clerical position household income is approximately 65 000 the PMK has grade 13 and the spouse grade 12 the spouse is employed in a semi skilled manual position and the PMK has a semi skilled clerical position is not in the labour force household income is approximately 55 000 the PMK and spouse have both completed grade 12 the PMK is employed in a semi skilled manual position and the spouse in an unskilled manual position household income is approximately 30 000 neither the PMK nor the spouse have completed high school the PMK is employed in an unskilled manual position and the spouse is employed i
39. weight 1500 to 2499 grams or very low birth weight lt 1500 grams NLSCY Data Users Guide 127 2001 2002 NLSCY Data Users Guide 128 2001 2002 Types of Errors Non Sampling Errors Defining the Term Respondent Chapter 9 Data Quality The estimates derived from this survey are based on a sample of children Somewhat different values might have been obtained if a complete census had been taken using the same questionnaires interviewers supervisors processing methods etc The difference between the estimates obtained from the sample and the results from a complete count taken under similar conditions is called the sampling error of the estimates Interviewers might misunderstand the instructions respondents might make errors while answering the questions the answers might be incorrectly entered on the questionnaire and errors might be introduced while processing and tabulating the data These are all examples of non sampling errors In certain circumstances it is not possible to gather all the data about a child The definition of the term respondent used in Cycle 1 was again used for Cycle 2 According to this definition a child is a respondent if there is enough information about at least one child in his household Cross sectional and longitudinal response rates Cross Sectional Response Rate The cross sectional response rate or collection rate at the household level is shown in the following
40. weight female babies to the number of low birth weight male Ratios babies The user is interested in comparing these estimates in the form of a ratio How does the user determine the coefficient of variation of this estimate 1 First of all this estimate is a ratio estimate where the Ay l numerator of the estimate is the number of low birth weight female babies and denominator Ae of the estimate is the number of low birth weight male babies 2 Refer to the table for the 0 to 3 age group The questions on birth weight were applicable only to children in the 0 to 3 age group 3 The numerator of this ratio estimate is 45 690 The figure closest to it is 45 000 The coefficient of variation for this estimate is found by referring to the first non asterisk entry on that row namely 10 3 4 The denominator of this ratio estimate is 38 395 The figure closest to it is 40 000 The coefficient of variation for this estimate is found by referring to the first non asterisk entry on that row namely 10 9 5 The approximate coefficient of variation of the ratio estimate is 2 i x respectively 4 amp tt A therefore given by Rule 4 whichis gt where 1 and a are the coefficients of variation of x 13nd That is ap 0 103 0 1097 0 150 The ratio of low birth weight female babies versus low birth weight male babies is 45 690 38 395 or 1 19 1 Since the c v for the estim
41. will be 8 for a 2 digit variable 98 for a three digit variable 998 etc NLSCY Data Users Guide 37 2001 2002 Don t Know The respondent may not know the answer to a particular item Again the CAI system has a specific function key to describe this situation On the NLSCY data file the code used to indicate that the respondent did not know the answer to an item is 7 For a variable that is one digit long the code will be 7 for a two digit variable 97 for a three digit variable 997 etc Not In some cases a question was not applicable to the survey respondent Applicable A code 6 96 996 has been used on the data file to indicate that a question or derived variable is not applicable lt Insome cases a single question or series of questions was not applicable For example the question on number of hours per week the child is cared for in a daycare centre CCRCQ1G1 is only applicable for children for whom this type of care is used CCRCQ1G 1 Otherwise there will be a code 996 for this question In other cases an entire section of the questionnaire was not applicable or even an entire questionnaire For example the Motor and Social Development Section was applicable only to children 0 to 3 years old For all children outside of this age group i e 4 years and older the motor and social development variables have been set to not applicable 6 96 996 etc For cases where the PMK did not have a
42. 11 Questionnaire then this component still exists for the child with all variables set to not stated Likewise if a parent completed a Child Questionnaire for one child in the household but refused to do so for a second child then there is a record for this second child with not stated values for all variables Parent Questionnaire Response Rates NLSCY Data Users Guide 135 2001 2002 The PMK and his or her spouse partner answered this questionnaire Again we determined the valid response rate obtained in order to assess the completeness of the data Out of the 24 692 PMKs and their spouse partners Response Rates there were answers to all relevant questions in 74 of the cases a valid answer was obtained for more than 90 of questions submitted to 95 of the adults less than 50 valid answers were gathered for 1 5 of the adults Child Questionnaire Response Rates Response Rates In order to assess completeness of the child data we determined the rate of answered questions among those that were relevant to the child In the sample of respondents consisting of 20 102 children there were answers to all relevant questions in 63 of the cases avalid answer was obtained for more than 90 of questions submitted to 98 of the children less than 50 valid answers were gathered for less than 1 of the children NLSCY School component NLSCY Data Users Guide 136 2001 2002 School component The Sc
43. 16 6 to 33 3 Estimates should be flagged with the letter M or some similar identifier They should be accompanied by a warning to caution subsequent users about the high levels of error associated with the estimates 3 Unacceptable Estimates have a sample size of less than 30 or very high coefficients of variation in excess of 33 3 Statistics Canada recommends not to release estimates of unacceptable quality However if the user chooses to do so then estimates should be flagged with the letter U or some similar identifier and the following warning should accompany the estimates The user is advised that specify the data do not meet Statistics Canada s quality standards for this statistical program Conclusions based on these data will be unreliable and most likely invalid These data and any consequent findings should not be published If the user chooses to publish these data or findings then this disclaimer must be published with the data NLSCY Data Users Guide 158 2001 2002 NLSCY Data Users Guide 159 2001 2002 Chapter 11 Approximate Sampling Variability Tables Introduction In order to supply coefficients of variation which would be applicable to a wide variety of categorical estimates produced from this microdata file and which could be readily accessed by the user a set of Approximate Sampling Variability Tables has been produced These look up tables which can be found in Appendix 3
44. 33 000 8 500 to 33 000 8 500 38 000 9 500 to 38 000 9 500 9 500 10 000 0 3 years 7 000 4 11 years 11 500 4 7 years 12 000 8 11 years 12 500 37 000 9 500 to 37 000 9 500 NLSCY Data Users Guide 178 2001 2002
45. 6H Q6M Q6GG and Q600 from the OCHS and ABECQ6D Q6U Q6BB Q6SS and Q6UU from the Montreal Longitudinal Survey the last four items are from a scale devised by K Weir and G Duveen NLSCY Data Users Guide 87 2001 2002 Results Two and three year olds There were 3 909 two and three year olds in the sample The group was split into two sub samples of 1 932 and 1 977 individuals and the analysis for this age group was performed separately for each sub sample The non response rate for most items was about 2 2 Some individuals were excluded from the analysis that produced the factors The exclusion criteria were as follows individuals with eight or more items coded missing individuals with one or more refusals individuals with two or more missing items under hyperactivity and emotional disorder and individuals with one or more missing items for the other theoretical factors After the criteria were applied there were 1 742 and 1 773 individuals left in the sub samples to be analysed Data were imputed for only 12 items The number of imputations ranged between 1 and 8 for those 12 items A total of 34 values were imputed Factor Analysis The factor analysis derived five factors for this age group hyperactivity inattention ABECS01 prosocial behaviour ABECS02 emotional disorder anxiety ABECS03 physical aggression opposition ABECS04 and separation anxiety ABECS05 The items making up each factor are listed in th
46. C MM and RR from the Ontario Child Health Study Anxiety Items include AD1CQ01Y and AD1CQ1II from the Montreal Longitudinal Survey and several of the OCHS emotional disorder items AD1CQO1F Q V and CC Indirect aggression Items include AD1CQ01J R Z LL and TT from Lagerspetz Bjorngvist and Peltonen of Finland Physical aggression Items include AD1CQ01 X from the Montreal Longitudinal Survey and AD1CQ01G AA and NN from the Ontario Child Health Study Inattention Items include AD1CQ01P from the Ontario Child Health Study and AD1CQ1EE KK QQ from the Montreal Longitudinal Survey Prosocial behaviour Items include AD1CQ01A H M GG and OO from the Ontario Child Health Study and AD1CQ01D U BB SS and UU from the Montreal Longitudinal Survey Constructing In Cycle 1 to construct the Behaviour Scale for the NLSCY a factor the analysis was conducted to test the theoretical construct In order to Behaviour be consistent with the behaviour scale created from the parent Scale questionnaire the factor structure which emerged from the 4 11 behaviour scale was imposed on the 10 11 behaviour scale NLSCY Data Users Guide 111 2001 2002 Normalized Weight Score Calculation Producing Final Scores Results In the factor analysis the items were multiplied by the child s normalized weight An individual s statistical weight is normalized by dividing his her weight AWTCW01 by the average weight of all i
47. Cronbach s Items that lowered Cronbach s cronbach s alpha the alpha if the item most if excluded is excluded Factor PPA A EE Conduct disorder and physical aggression AD1CS03 Hyperactivity inattention AD1CS04 Prosocial behaviour AD1CS05 I I I I I I I I I I I I I 4 I I I I I 4 I I I I I I I I 4 I I I I I 4 I I I I I The scores for these factors could not be computed in 566 16 5 597 17 4 585 17 621 18 1 and 587 17 1 cases respectively because of unreported values My Parents and Me self complete 10 15 This section was part of the self complete questionnaire given to children in the 10 to 15 age group The objective was to complement the Parenting Section on the Child s Questionnaire completed by the PMK by gathering information directly from the child regarding his her perception of his her relationship with parents For the self completed questionnaire it was also considered important to obtain a measure of parental supervision i e monitoring as this has been shown to be linked to child outcomes there is a correlation between a lack of supervision and negative outcomes such as juvenile delinquency and other risk taking behaviours Objective My Parents and Me NLSCY Data Users Guide 114 2001 2002 Scale Used for My Parents and Me My Parents and Me Scale CPMCcS1 CPMCbS2B CPMCcS3 Questionnaire Items Analysis of the NLSC
48. For Number of Rotation Groups 24 Correction For Household Non Response The weighting strategy applied to these children is similar to that used in Cycles 1 and 2 The LFS sample is made up of six rotation groups each of which is a representative sub sample of the LFS target population In the NLSCY we used 13 rotation groups Consequently the first adjustment is 6 13 Further to this adjustment the adjusted weight is obtained by multiplying the LFS weight by 6 13 In surveys such as the NLSCY some households do not provide responses for a variety of reasons refusal special circumstances language problems temporary absence This non response is usually compensated for by proportionally correcting the sub weights of the responding households The correction is made by multiplying the sub weight of the responding households by the following factor Sum of adjusted weights of households sampled within a stratum of the NLSCY Sum of adjusted weights of responding households within a stratum of the NLSCY In this equation the adjusted weight is the weight obtained after Correction 1 A different correction was made in each of the strata specially defined for non response by the LSF The strata are defined using the following information province economic region census metropolitan area type of sector urban rural apartment frame whether special region or not Each of the strata has at least 30 children and a resp
49. Guide 29 2001 2002 NLSCY Data Users Guide 30 2001 2002 Consistency Editing The Goal of Consistency Editing Consistency Between Cycles Consistency Edits for PAPI Data File for 10 to 15 year olds Data File for Teacher s File After the pre edit consistency editing was carried out to verify the relationship between two or more variables Example In the Socio Demographic Section for children who were not born in Canada Question CSDCQ 2B asks on what year they first immigrated to Canada There was a consistency edit which compared this question to the year of birth of the child If the year of immigration was before year of birth then year of immigration was set to not stated in the edit Editing was also performed to ensure consistency between cycles Example The responding child s height in Cycle 3 should not be less than the height reported in Cycle 2 Flags were set for inconsistencies between cycles These variables appear on the Secondary data file Appendix 5 and contain Z in the variable name For PMK and Spouse variables the data was linked using a unique person identifier allowing the comparison to be made if the PMK was the same in both cycles or if the PMK was the spouse in the previous cycle and vice versa For the questionnaires that were collected using a paper version essentially the same steps of editing were carried out In the pre edit however there was an additional requireme
50. LD FACTOR SCORE ITEMS ose ho FS N a iy ee General Self______ J AC1CS02__ AC1CQ01A AC1CQ01B AC1CQ01C ACICQ01D Physical AC1CS01 AC1CQ01E AC1CQ01F AC1CQ01G AC1CQ01H Appearance i Cronbach s Cronbach s alpha coefficients raw values were calculated with Alpha for SAS using the normalized weighted data Please note that in About Me general Cronbach s alphas calculated with SAS are lower than those Scales produced by the SPSS software package For the general self score the Cronbach alpha was 0 728 The item that affects the factor the most is AC1CQO1C If it were removed from the analysis the Cronbach s alpha would drop to 0 629 For the physical appearance score the Cronbach alpha was 0 874 The item that affects the factor the most is AC1CQO1E If it were removed from the analysis the Cronbach s alpha would drop to 0 811 Once the factors were determined the next step was to calculate the scores for each of the two factors For the general self factor scores could not be calculated for 555 individuals 16 2 due to missing values for the items comprising this factor For the physical appearance factor scores could not be calculated for 589 individuals 17 2 due to missing values for the items comprising this factor NLSCY Data Users Guide 119 2001 2002 Depression Scale self complete 12 15 In order to be consistent with the depression scale created from the Depression parent questionnaire the factor s
51. Y Data The scale that was used was also used in the Western Australia Child Health Survey It was developed by Lempers et al 1989 based on work of Schaefer 1965 and Roberts et al 1984 and measures parental nurturance rejection and monitoring This information will complement the constructs measured in the parent completed Child s Questionnaire positive child parent interaction hostile ineffective child parent interaction and consistent child parent interaction aversive and non aversive parent management techniques The objective of the My Parents and Me scale is to measure the child s perception of his her relationship with his her parents and parental supervision Below is a description of the items that were included on the 10 15 year old questionnaires to measure family relations the analysis used to construct the scale and the results of these analyses Questions CPMCcQ1A to CPMCcQ1Q were taken from the Western Australia Child Health Survey In addition to these questions questions CPMCcQ1R to CPMCcQI1T were also used The scale was developed by Lempers et al 1989 based on work of Schaefer 1965 and Roberts et al 1984 and measures parental nurturance rejection and monitoring To construct the My Parents and Me Scale for the NLSCY a factor analysis was conducted to test the theoretical construct In the factor analysis the items were multiplied by the child s normalized weight An individual s statistical w
52. Y Data Users Guide 6 2001 2002 Response Rates at Child Level Computer Generated Response Codes Chapter 3 Response Rates Since the child is the unit of analysis response rates are presented at the child level rather than the household level In many cases there may be more than one child per household consequently there may be data obtained for one child but not for another in the same household On the computer each household represents a case and status codes are automatically given to the case each time an interviewer enters it Within each case are components for example there each child has its own component Consequently it s possible to have one household with different response codes for each component Complete information may be available for one child but not another in this case a partial or non response code would appear for one child component while a fully complete response code appears for the other At the household level this case would have a partial response code As the panel ages a larger proportion of the sample will come from the early years cohort in which only one child per household is selected Thus the response rates at the household and at the person level should gradually become very similar Table 1 Overall Child Level Response Rates NLSCY Cycle 3 Sample Not eligible Eligible Full Partial Refusal Unable to trace Other non response inclu
53. Y children living in a household where the PMK had a spouse partner and for 14 8 of children the PMK did not have a spouse partner CDMPDO6A The two ways of describing the child s family are very similar The only reason for the small differences is a result of the few cases where the child lived with a parent but the parent was not selected to be the PMK These estimates for family derived variables are based on unweighted data NLSCY Data Users Guide 59 2001 2002 Step Blended Children living with two parents are classified as being members of and Intact intact step and or blended families based on the relationship of Families these children to the parents Intact Family An intact family consists of a married or common law couple in which all children are the natural and or adopted offspring of both members of the couple For the NLSCY children 75 9 were a member of an intact family CDMCD16 For the NLSCY children 4 3 were step children themselves CDMCD03 and 8 5 lived in a step family CDMCD15 Step Family A step family consists of a married or common law couple residing in the same household with at least one step child living with them who is the biological or adopted child of one parent but not the other It should be noted that a child who is the biological child of both parents is said to belong to a step family if at least one of these parents has a step child residing in the household For the NLSCY child
54. able below shows the relation between the raw gross scores and the scaled scores by grade for the NLSC mathematics test The scaled scores for this test range from 200 to 999 for Cycle 3 NLSCY Data Users Guide 141 2001 2002 Relation between raw gross scores and scaled scores by grade for the Cycle 3 mathematics test Gross Scaled scores scores P Grate Grades Grades Grades Grades Grade Grades Grades a0 e e7 s8 s2 ae s so se e 24 aes a7 a5 ae sae s e7 o s ss a ase a90 se se 652 With the expansion of the tests in Cycles 2 and 3 the ceiling effect measured in Cycle 1 has been greatly reduced Although the skill indicator was still used to elevate children to the next level test the process is problematic and prone to human error During the implementation of the tests in Cycle 3 children who received a perfect score on the skills indicator were divided into two equal size groups one group to receive their regular level tests the other to receive next level tests Unfortunately half of the children targeted through this process namely those who were to receive their regular level tests were not sent tests Although these missing tests would not affect the classically derived scaled scores they had to be adjusted for during the IRT calculation of scaled scores NLSCY Data Users Guide 142 2001 2002 Reading comprehension test The reading compreh
55. allow the user to obtain an approximate coefficient of variation based on the size of the estimate calculated from the survey data Coefficients of The coefficients of variation c v are derived using the variance Variation formula for simple random sampling and incorporate a factor which reflects the multi stage clustered nature of the sample design This factor known as the design effect was determined by first calculating design effects for a wide range of characteristics and then choosing from among these a conservative value to be used in the look up tables which would then apply to the entire set of characteristics Sample For the NLSCY the sample was constructed taking account the Requirements following requirements A sufficient sample was required in each of the 10 provinces to allow for the production of reliable estimates for all longitudinal children who were 0 to 11 years of age in Cycle 1 It was also necessary to have a large enough sample to produce estimates for Cycle 1 at the Canada level by seven key age groupings or cohorts 0 to 11 months 1 year 2 to 3 years 4 to 5 years 6 to 7 years 8 to 9 years and 10 to 11 years _ In each province a sufficient sample size was required for Cycle 2 to produce reliable estimates for all children who were 0 to 11 years of age in Cycle 1 NLSCY Data Users Guide 160 2001 2002 The tables that follow show the design effects sample sizes and population counts by prov
56. antly high or low a pop up message would appear asking the interviewer to confirm the answer with the respondent Flow Pattern All flow patterns were automatically built into the CAI system Edits Example In the Child Care Section the PMK is asked he she used daycare or babysitting in order that he she or a partner spouse could work or study Based on the response given the flow of the questions could be different If Child Care was used the CAI system continued with a series of questions about the specific care method s used for the child If not the CAI system automatically skipped this series of questions General Some consistency edits were included as part of the CAI system and Consistency interviewers were able to slide back to previous questions to correct Edits for inconsistencies Instructions were displayed to interviewers for handling or correcting problems such as incomplete or incorrect data Example In the collection of the Labour Force Section the number of weeks working not working and looking for work should not total more than 52 weeks If this was the case the system generated a pop up window which stated the error and instructed the interviewer to slide back to the appropriate question to confirm the data and make corrections as required NLSCY Data Users Guide 25 2001 2002 Consistency Edits Between Cycles Data Capture Paper and Pencil Questionnaire s Data Capture for PAPI Questi
57. ate falls in the 0 0 to 16 5 range 15 0 this estimate is acceptable and the finding that 70 8 of low birth weight babies were born prematurely requires no warning note NLSCY Data Users Guide 169 2001 2002 How to Use the C V Tables to Obtain Confidence Limits Although coefficients of variation are widely used a more intuitively meaningful measure of sampling error is the confidence interval of an estimate A confidence interval constitutes a statement on the level of confidence that the true value for the population lies within a specified range of values For example a 95 confidence interval can be described as follows Introduction If sampling of the population is repeated indefinitely each sample leading to a new confidence interval for an estimate then in 95 of the samples the interval will cover the true population value Using the standard error of an estimate confidence intervals for estimates may be obtained under the assumption that under repeated sampling of the population the various estimates obtained for a population characteristic are normally distributed about the true population value Under this assumption the chances are about 68 out of 100 that the difference between a sample estimate and the true population value would be less than one standard error about 95 out of 100 that the difference would be less than two standard errors and about 99 out 100 that the differences would be less than three
58. ate of X 5 5 t 2 alpha estimate of X 7 3 073 expressed as a proportion is the coefficient of variation of this estimate as determined by the tables Clx 0 055 2 0 055 0 073 0 055 2 0 055 0 073 Cix 0 055 0 008 0 055 0 008 Clx 0 047 0 063 With 95 confidence it can be said that between 4 7 and 6 3 of babies who were 0 to 3 years old at the time of the survey were of low birth weight NLSCY Data Users Guide 171 2001 2002 How to Use the C V Tables to Do a T test Standard errors may also be used to perform hypothesis testing a procedure for distinguishing between population parameters using sample estimates The sample estimates can be numbers averages percentages ratios etc Tests may be performed at various levels of significance where a level of significance is the probability of concluding that the characteristics are different when in fact they are identical Hypothesis Testing Let x and Ra be sample estimates for two characteristics RA of interest Let the standard error on the difference be Oy If 3s between 2 and 2 then no conclusion about the difference between the characteristics is justified at the 5 level of significance If however this ratio is smaller than 2 or larger than 2 the observed difference is significant at the 0 05 level That is to say that the characteristics are significantly different NLSCY Data Users Guide 172 2001 2002
59. be attached for subsequent users about the high level of error associated with the estimate NLSCY Data Users Guide 174 2001 2002 GEOGRAPHICAL RELEASE CUT OFFS CROSS SECTIONAL SAMPLE Province Acceptable Marginal Unacceptable Estimates at or Estimates Estimates at or above between below Newfoundland 7 500 2 000 to 7 500 2 000 Prince Edward 1 000 to 3 500 Island 2 000 12 000 3 000 to 12 000 3 000 British 51 500 13 500 to 51 500 13 500 Columbia RAM Ml provinces Prairie 26 000 6 500 to 26 000 6 500 provinces 41 000 10 000 to 41 000 10 000 NLSCY Data Users Guide 175 2001 2002 RELEASE CUT OFFS BY AGE GROUP CROSS SECTIONAL SAMPLE anil Acceptable Marginal Unacceptable Estimates at Estimates Estimates at or above between or below 11 000 to 42 000 37 500 9 500 to 37 500 9 500 orm 41 000 10 0000 47 000 10 000 NLSCY Data Users Guide 176 2001 2002 GEOGRAPHICAL RELEASE CUT OFFS LONGITUDINAL SAMPLE Province Acceptable Marginal Unacceptable Estimates at Estimates Estimates at or or above between below Newfoundland Prince Edward Island Nova Scotia New Brunswick 2 500 to 9 500 Qu bec Ontario Manitoba Aberia Bish Columbia NLSCY Data Users Guide 177 2001 2002 RELEASE CUT OFFS BY AGE GROUP LONGITUDINAL SAMPLE Age Group Acceptable Marginal Unacceptable Estimates at Estimates or above between Estimates at or below 0 23 months 19 500 5 000 to 19 500 5 000
60. bout the child as well as the child s parent s characteristics of the family and the neighbourhood This section provides an outline of the content for each section of the questionnaire included in the NLSCY data As part of the NLSCY processing system there were some basic quality checks performed for each section of the questionnaire Any items for which there was a high level of non response or which were frequently involved in edit failures were looked at in detail Where appropriate comparisons were made to external data sources and analyses were carried out to investigate possible reasons for differences from these other sources Any concerns about potential data quality problems for any items in a particular section of the questionnaire are discussed in this section of the documentation NLSCY Data Users Guide 71 2001 2002 General Validation Procedures Before the section by section discussion of content and validation results the general validation procedures used for the scale data are presented PLEASE NOTE THAT MOST SCALES WERE DEVELOPED AND VALIDATED IN CYCLE 1 IN SUBSEQUENT CYCLES THE SAME FACTOR STRUCTURE WHICH EMERGED FROM THE CYCLE 1 ANALYSIS WAS IMPOSED THIS ENSURES THAT THE SCALES ARE CONSISTENT ACROSS TIME TO ALLOW FOR LONGITUDINAL ANALYSIS AND CROSS SECTIONAL COMPARISONS IN THE SECTIONS DESCRIBING THE VALIDATION OF THE SCALES WHERE THE ANALYSIS WAS DONE USING CYCLE 1 DATA THE VARIABLES CITES WILL BEGIN WITH
61. cing practices for the hostility ineffective parenting factor Results Cycle 1 Children There were 4 696 children in the sample for the age group 0 to 23 aged 0 to 23 months The group was split into two sub samples of 2 311 and months 2 385 individuals and the analysis for this age group was performed separately for each sub sample The non response rate for the seven items ranged from 1 9 to 2 5 Some individuals were excluded from the analysis that produced the factors The exclusion criterion was as follows individuals with one or more missing items After the criterion was applied there were 2 245 and 2 307 individuals left in the sub samples to be analysed No imputation was done The factor analysis derived two factors for this age group positive interaction APRCS01 and hostile ineffective APRCS02 The items making up each factor are listed in the table below PARENTING SCALE FOR CHILDREN AGED 0 TO 23 MONTHS FACTOR Hostile ineffective APRCS02 APRC4 5 NLSCY Data Users Guide 95 2001 2002 Cronbach s Alpha for Children aged 0 to 23 Months Missing Values Children aged 2 to 11 Cronbach s alpha raw value was computed with SAS using normalized weighted data In general Cronbach s alphas computed by SAS are lower than those produced by SPSS For the positive interaction factor APRCS01 Cronbach s alpha was 0 727 The item that had the greatest effect on this factor was APRCQ7 Removing it low
62. composition frequently due to divorce of parents or children leaving the parental nest Attempts have been made in other studies to define longitudinal households but the implementation of this concept has never been straightforward No single definition has been found to be appropriate for most analytic tasks and many definitions exclude the portion of the population that has undergone the change Unfortunately this is often a significant as well as interesting population to study It has been suggested that a superior alternative is to use the individual as the unit of analysis and present family and household variables as a characteristic of the individual Thus the files which have been constructed for all NLSCY data consist of child records In order to understand the family situation estimates such as of the number of children in single parent families or the number of children living in low income households can be produced For a more detailed examination of units of analysis in longitudinal studies see G D Duncan and M S Hill Conceptions of Longitudinal Households Fertile or Futile Journal of Economic and Social Measurement 1985 13 361 375 NLSCY Data Users Guide 55 2001 2002 PMK and Spouse Person Most Knowledgeabl e In each NLSCY household for Cycle 3 for each selected child a question was asked about who in the household was the person most knowledgeable about this child This person was lab
63. crodata file by summing the final weights of all records possessing the characteristic s of interest These estimates may be cross sectional or longitudinal A X Proportions and ratios of the form are obtained by Y a summing the final weights of records having the characteristic A of interest for the numerator X b summing the final weights of records having the characteristic of interest for the denominator A Y then c dividing the numerator estimate by the denominator estimate 154 2001 2002 Tabulation of Quantitative Estimates Estimates of Quantities Example of a Quantitative Estimate Estimates of quantities can be obtained from the microdata file by multiplying the value of the variable of interest by the final weight for each record then summing this quantity over all records of interest For example to obtain an estimate of the total number of days of special care received by infants who were born prematurely multiply the number of days for which special care was received by the final weight then sum this value over all records for which the child was born prematurely To obtain a weighted average of the form A A X the numerator X is calculated as for a Y quantitative estimate and the denominator Y is calculated as for a categorical estimate For example to estimate the average number of days spent in special care by premature babies a estimate
64. cy APRCS05 The items making up each factor are listed in the table below PARENTING SCALE FOR CHILDREN AGED 2 TO 11 FACTOR SCORE i ITEMS eee esse oe ea awe cte E EE esse ste cee action __ APRCSO3 __ APRCQ1 2 3 6 7 ies _APRCS04 APRC Q4 8 9 13 14 15 18 NLSCY Data Users Guide 96 2001 2002 Item inverted when computing the score Cronbach s alphas for these factors are given in the table below Normalized weighted data were used for the computations CRONBACH S ALPHA FOR THE PARENTING SCALE FOR 2 AND 3 YEAR OLDS Factor I Cronbach s Item that lowers Cronbach s alpha raw Cronbach s alpha the alpha ifthe item bic anes Puchi ieee horn eee st mostifitis excluded _ _isexcluded _____ Positive interaction 0 808 APRCQ2 0 749 _ APRCS03 a aed at aca ese Ineffective APRCS04 0 706 APRCQ13 0 654 Consistency 0 660 APRCQ12 1 0 569 APRCS07 The scores for these factors could not be computed in 408 482 and 534 cases respectively because of unreported values Parenting There were 18 135 children in the sample for the age group 2 to 11 scale for The group was split into two sub samples of 9 090 and 9 045 children individuals and the analysis for this age group was performed aged 2 to 11 separately for each sub sample The non response rate for the seven items analysed was about 2 5 The exclusion criterion was as follows individuals with one or more items coded missing were e
65. d in English only and so in Quebec the sample represents only the English schools which is referred to as the normative sample The children from the normative sample received the complete test The scaled scores are units of a single scale with equidistant intervals that covers all of the grade levels The scale was developed using a Thurstone procedure derived from the classical testing theory The fact that a short test was used for children in the NLSCY sample meant that it was not possible to directly associate the CTC scaled scores with the gross scores obtained in the survey For this reason the CTC normative sample was used to calculate the percentile rank for each gross score but using only the 15 of the 20 questions of the short NLSCY test Only 15 of the 20 items were normalized for the appropriate grade level while the remaining 5 items were taken from the test of the next grade level The normative score was then interpolated by inserting the percentile rank obtained with the 15 questions of the short test between the percentiles of the complete test For example using level 6 we find in the short test a percentile rank of 2 2 for a raw gross score of 1 On the complete test the percentile ranks of 2 0 and 3 7 correspond to raw gross scores of 5 and 6 and to scaled scores of 332 and 348 respectively After linear interpolation we obtain a scaled score of 334 for the gross score of 1 on the short version of the test The t
66. des longitudinal respondents who did not respond in Cycle 2 3g035 16718 aar Coe a Longitudinal Other Cohort BE ey T NLSCY Data Users Guide 7 2001 2002 Response Rates Cycle 3 Non Response Responding Sample by Age and Province As a percentage of all eligible children a response rate of 88 was achieved with 87 of cases being fully completed and 1 of cases being partially completed In 3 of cases non responses occurred because the respondent no longer resided at the address or phone number on file and attempts to trace their current location were unsuccessful In 6 of cases households refused to participate and in 3 of cases other non responses occurred Examples of other non responses include unable to interview due to unusual circumstances i e death in the family illness the household was absent during the collection period and unable to interview due to language problems In total 31 194 children were retained on the final data file The following two tables present the responding sample by province and age group Table 2 Province and Sample Size Province Responding Sample Size Newfoundland NLSCY Data Users Guide 8 2001 2002 Table 3 Age of Child and Responding Sample Size NLSCY Data Users Guide 9 2001 2002 NLSCY Data Users Guide 10 2001 2002 Chapter 4 Data Collection Computer Assisted Interviewing Computer Assisted Interviewing The CAI System Case Ma
67. dits were what is known as soft edits and some were hard The first example was a hard edit and the second a soft edit For all edit failures the records for the entire household were reviewed manually for obvious mistakes A correction had to be made for the hard edit failures For the soft edit failures a correction was made if it was deemed appropriate to do so NLSCY Data Users Guide 125 2001 2002 Sources of Error The major source of error for relationship data had to do with step children There were several cases where a female parent was living with a biological child and a spouse or common law partner The relationship of the male partner to the child was coded as unrelated For questionnaires completed in French this relationship was often coded as in law In the edit the relationship code was changed to step child for these cases As a result of the relationship edits the number of children in step families increased by close to 40 Medical Biological Medical Biological Children Under 2 Birth Weight The Medical Biological Section was completed for children in the 0 to 3 age group The major objective was to collect information on factors such as gestational age and birth weight These factors have been shown to have a direct impact on a child s growth and development For example in the long term underweight babies face higher risks of poor health as well as longer lasting developmental difficulties
68. e table below BEHAVIOUR SCALE FOR 2 AND 3 YEAR OLDS FACTOR SCORE ITEMS Hyperactivity inattention ABECSO ABECQ6B 6l 6N 6P 6S 6HH 1 6QQ Prosocial behaviour ABECSO ABECQ6D 6U 6BB 6SS 6UU 2 Emotional disorder anxiety ABECSO ABEQCE6F 6K 6Q 6V 6MM 3 6RR Physical aggression opposition ABECSO ABECQ6G 6W 6X 6E1 6R1 4 6T1 6Z1 6NN Separation anxiety ABECSO ABECQ6CC 6DD1 6PP1 6LL1 5 6TT1 NLSCY Data Users Guide 88 2001 2002 Cronbach s alpha raw value was computed with SAS using normalized weighted data in general Cronbach s alphas computed by SAS are lower than those produced by SPSS For hyperactivity inattention ABECS01 Cronbach s alpha was 0 798 The item that had the greatest effect on this factor was ABECQ6P as removing it lowers Cronbach s alpha to 0 762 The table below shows the Cronbach s alpha for each factor first including all items then excluding the item having the greatest effect Cronbach s Alpha for 2 and 3 Year Olds CRONBACH S ALPHA FOR THE BEHAVIOUR SCALE FOR 2 AND 3 YEAR OLDS FACTOR CRONBACH ITEM THAT LOWERS CRONBACH S ALPHA CRONBACH S S ALPHA IF RAW ALPHA THE MOST IF THE ITEM IS IT IS EXCLUDED EXCLUDED Hyperactivity inattention 0 798 ABECQ6P 0 761 ABECS01 Prosocial behaviour ABECQ6SS ABECS02 ABECSO03 opposition ABECS04 ABECS05 Children aged 4 to 11 There were 14 226 children in the 4 to 11 age group Two
69. e LFS and the birth register Since a good number of households were contacted for the first time in Cycle 3 the overall response rate for Cycle 3 is lower than that for Cycle 2 Cross Sectional Response Rate The table below gives the response rate for households contacted by Sample for the first time in Cycle 3 as well as for respondent households Source contacted in at least one previous cycle NLSCY Cross Sectional Response Rate by Sample Source Households Respondent Contacted Households Response Rate Longitudinal Households Selected in Cycle 1 Longitudinal Households Selected in 5 Year Old Children Selected from Birth Register i t i NLSCY Data Users Guide 130 2001 2002 As well the reason for household non response will be different depending on whether the household is longitudinal In fact longitudinal households are usually more apt to take part in the survey having already done so in the past However some households may have moved between the second and third collection cycles As a result it is sometimes necessary to track down the longitudinal children before proceeding with collection This operation is not always successful Longitudinal children who move may thus lead to some erosion of our longitudinal sample The following tables show the distribution of non responding longitudinal and new households by reason for not responding New Households Added to Cycle 2 Non Respondents by R
70. e and therefore unofficial The use of actual variance estimates would likely result in estimates with lower variances for example estimates listed as unacceptable in the Approximate Sampling Variability Tables could move up to the marginal category Approximate Sampling Variability Tables Remember If the number of observations on which an estimate is based is less than 30 the weighted estimate should be classified as unacceptable regardless of the value of the coefficient of variation for this estimate This is because the formulas used for estimating the variance do not hold true for small sample sizes NLSCY Data Users Guide 163 2001 2002 How to Use the C V Tables For Categorical Estimates Introduction Rule 1 Estimates of Numbers Possessing a Characteristic Aggregates Rule 2 Estimates of Proportions or Percentages Possessing a Characteristic NLSCY Data Users Guide The following rules should enable the user to determine the approximate coefficients of variation from the Sampling Variability Tables for estimates of the number proportion or percentage of the surveyed population possessing a certain characteristic and for ratios and differences between such estimates The coefficient of variation depends only on the size of the estimate itself On the Sampling Variability Table for the appropriate geographic area or age group locate the estimated number in the left most column of t
71. e assigned to each occupation is the centre of its piece in the logistic distribution This transformation was employed to scale the 16 occupations For each occupation group x the following values were computed p the percentage of individuals with an occupation less than occupation x based on the Pineo Porter McRoberts category pp the percentage of individuals with an occupation less than or equal to occupation x based on the Pineo Porter McRoberts category phi p p In p 1 p In 1 p phi pp pp In pp 1 pp In 1 pp The recoded logit value for occupation x was assigned to be PINEOLOG phi pp phi p pp p PINEOLOG for both the PMK and spouse partner was then used in the derivation of both SES NLSCY Data Users Guide 65 2001 2002 Household Income Final Derivation of Cross sectional and Longitudinal SES Final Derivation of Cross sectional SES The last variable used in the derivation of SES was household income More detail regarding the collection of household income and data quality issues can be found in Section 9 17 To derive SES income was coded in 1 000s of dollars and a few outliers with incomes greater than 150 000 were recoded to 150 000 Thus the five variables that were used to derive both SES were CEDPD04 years of schooling for the PMK CEDSD04 years of schooling for the spouse partner PINEOLOG PMK the pineo occupation code for the PMK transfo
72. e coefficient of variation of the proportion is not releasable then the coefficient of variation of the corresponding quantitative estimate will also not be releasable 173 2001 2002 Coefficients of variation of such estimates can be derived as required for a specific estimate using a technique known as pseudo replication This involves dividing the records on the microdata files into subgroups or replicates and determining the variation in the estimate from replicate to replicate Users wishing to derive coefficients of variation for quantitative estimates may contact Statistics Canada for advice on the allocation of records to appropriate replicates and the formulae to be used in these calculations Pseudo Replication Release Cut offs for the NLSCY In the tables that follow cut off numbers are given for NLSCY estimates in order for them to be of acceptable marginal or unacceptable quality Users are encouraged to use these cut offs when publishing data from the NLSCY First a table is given to show the cut offs at the provincial regional and Canada level Then a table is given to show the cut offs for the various age cohorts An interpretation of what is meant by the various cut off levels can be found in Section 11 4 Cut off Numbers For example an estimate for Nova Scotia of 5 000 would fall into the marginal range This would mean that the estimate should be flagged and a note of caution would
73. e described below This particular definition of SES was Proposed by Dr Douglas Willms Atlantic Centre for Policy Research in Education University of New Brunswick NLSCY Data Users Guide 62 2001 2002 Education The education variable used in the construction of SES was years of Years of schooling Two such variables were derived independently one for School the PMK and one for the spouse partner CEDPD04 for the PMK and CEDSD04 for the Spouse partner For the PMK the years of schooling variable was derived based on items CEDPQ01 years of elementary and high school and CEDPQ04 highest level of education attained beyond high school To create a somewhat continuous interval level education variable these two items were recoded to form years of schooling in the following manner CEDPD04 Condition 00 CEDPQ01 1 03 CEDPQ01 2 no schooling 1 to 5 years 06 CEDPQ01 3 6 years 07 CEDPQ01 4 7 years 08 CEDPQ01 5 8 years 09 CEDPQ01 6 9 years 10 CEDPQ01 7 10 years 11 CEDPQ01 8 11 years 12 CEDPQ01 9 12 years 13 CEDPQ01 10 13 years 16 CEDPQ04 6 BA BSC 18 CEDPQ04 7 Masters 20 CEDPQ04 8 or 9 MD PHD An extra year was then added to CEDPD04 if the PMK had a diploma from a trade school or community college i e if CESPDQ04 4 or 5 then CEDPD04 CEDPQ04 1 The same procedure was used to set up a years of schooling variable for the spouse partner CEDSD04 1 In cases where the PMK had not
74. e of measurement errors There are many ways to measure reliability One of the most commonly used reliability coefficients is Cronbach s alpha Cronbach 1951 Alpha is a measure of the internal consistency of the items within the factor It is based on the average covariance of items within the factor It is assumed that items within a factor are positively correlated with each other because they are attempting to measure to a certain extent a common entity or construct Cronbach s a has several interpretations It can be viewed as the correlation between this scale or factor and all other possible scales containing the same number of items which could be constructed from a hypothetical universe of items that measure the characteristic of interest In the hostile ineffective parenting factor for example the seven questions actually used for inclusion on the scale can be viewed as a sample from the universe of many possible items Parents could also have been asked How often do you raise your voice when you discipline your child or How often do you threaten punishment more often than you use it Cronbach s a tells how much correlation can be expected between the scale which was used and all other possible seven item scales measuring the same thing 80 2001 2002 Another Interpretation of Cronbach s Alpha Another interpretation of Cronbach s a is the squared correlation between the score an individual obtains on a particula
75. ead Office processing staff for further processing As part of the scanning system some quality checks were built in to flag unusual entries to warn the operators of potentially incorrect entries The operator visually reviewed the questionnaire responses and manually entered the correct values In cases where more than one response was checked off by the respondent the operators were instructed to accept the first response Errors remaining within the questionnaires were then edited at a later stage NLSCY Data Users Guide 26 2001 2002 Minimum Completion Requirements Defining One of the first steps in the NLSCY processing was to define the Requirements requirements for a responding household No In some cases no NLSCY information was collected for a sampled Information household This happened for example when an interviewer was Collected unable to make contact with a selected household for the entire collection period in other cases the household refused to participate in the survey special circumstances such as an illness or death ina family or extreme weather conditions sometimes prevented an interview from taking place For cases where no information was collected for a household the household was dropped from the NLSCY file and the sampling weights for responding households were inflated to account for these dropped households Partial In other cases it was possible to carry out some of the interview but a Informati
76. ease Guidelines Release Guidelines NLSCY Data Users Guide Before releasing and or publishing any estimate from the NLSCY users should first determine the quality level of the estimate The quality levels are acceptable marginal and unacceptable As discussed in Chapter 10 sampling and non sampling errors both influence data quality For the purposes of this document however estimate quality is based solely on the sampling error illustrated by the coefficient of variation as shown in the table below First the number of children who contribute to the calculation of the estimate should be determined If this number is less than 30 the weighted estimate should be considered to be of unacceptable quality For weighted estimates based on sample sizes of 30 or more users should determine the coefficient of variation of the estimate and follow the guidelines below These quality level guidelines should be applied to weighted rounded estimates All estimates can be considered releasable However those of marginal or unacceptable quality level must be accompanied by a warning to caution subsequent users 157 2001 2002 QUALITY LEVEL GUIDELINES Quality Level of Guidelines Estimate 1 Acceptable Estimates have a sample size of 30 or more and low coefficients of variation in the range 0 0 to 16 5 No warning is required Estimates have a sample size of 30 or more and high coefficients of variation in the range
77. eason for Not Responding NLSCY New Households Added to Cycle 2 Non Respondents by Reason for Not Responding Non Responding Households SEES pee mes eae ep Cae ee OS Sas eee nee eee ee Cee eee ee dea ey NLSCY Longitudinal Households Not Responding to Cycle 2 by Reason for Not Responding Non Responding Households Refusal 1393665 Not tracked down sn oe ata ee i Net ee Se de ects aes 4 12 _ No neat home 2 si anaes ine rae eerie ste dios OD azote 42 __ Language barrier mri et 28 a ore ates A hee ibaa Dias oh 01 Special circumstances sickness weather conditions 100 1 5 Gte a tah ct ea a eM rasan renner ON Sen arse eae Partial response rejected for lack of information techie OMe fel 1 1 Other or reason unknown 84 4 Toal te ee ee ee egg T 100 NLSCY Data Users Guide 131 2001 2002 Given the survey method applied to the first two collection cycles it was unfortunately impossible to obtain an exact longitudinal response rate taking into consideration all the components of erosion Ideally this rate would be the simple ratio of the number of longitudinal children responding to the second cycle to the number of children contacted for the first cycle However the number of children present in non responding households during the first cycle is unknown The number of children present in households not responding to the LFS is also unknown It is therefore impossib
78. economic status and income ratio Derived Variables In the second and third cycle of the survey two distinct measures of socio economic status were calculated one longitudinal and one cross sectional The derivations of cross sectional SES and of longitudinal SES differ only with respect to the standardization of the components The derivation of the non standardized components of SES i e parents education level parents occupational prestige and household income was the same for both SES measures Sociologists often use the term socio economic status SES to refer to the relative position of a family or individual in an hierarchical social structure based on their access to or control over wealth prestige and power In studies of children s academic and social emotional development SES is often operationally defined through measures describing the occupational prestige educational levels and economic positions of children s parents Socio Economic Status The measure of SES is calculated for each household assigned to each selected child in that household It was derived from five sources the level of education of the PMK the level of education of the spouse partner the prestige of the PMK s occupation the prestige of the occupation of the spouse partner and household income The method of constructing each component of SES and the construction of the overall cross sectional and longitudinal SES measure ar
79. ed their questionnaire re 8733 9 6 The province of residence is missing for 4 children Response The response rate for the Principal s Questionnaire and hence the Rates by percentage of children whose principals returned their province questionnaire varied widely from province to province and appeared to be lower in Quebec Response Rates by school type Principal s Questionnaire Type of School Principal of the Schools Questionnaires Returned That Have Received a Questionnaire po Frequenc Other Missing 188 60 0 6647 4498 67 7 Type of School Children Consent of Parent Consent of School Board Children whose Principal have returned their questionnaire Co EES TE Private o s1 Other Missing 736 666 NLSCY Data Users Guide 147 2001 2002 Response As in the case of the Teacher s Questionnaire the response rate for Rates by the Principal s Questionnaire appeared to be about equal for public school type school students and Catholic school students but it appeared to be lower for children attending private schools Response Rates by age Principal s Questionnaire Children Consent of Parent Consent of School Board Children whose Principal have Going to returned their questionnaire School Cum Cum ee ea er 914 60 1 6 1219 175 94 6 840 68 9 765 65 9 93 9 8 147 104 E 93 3 66 5 9 1 3 1024 9 94 5 69 0 981
80. efers to the NLSCY Cycle Names A indicates the first cycle B the second cycle C the third etc SE refers to the section of the questionnaire where the question was asked or the section from which the variable was derived C refers to the collection unit or the unit to which the variable refers There are four possibilities 7 C is the variable refers to the child P the PMK S the spouse partner H the household b the lower case letter refers to the NLSCY Cycle in which the variable first appeared on the file It should be noted that while variables do exist for various units of analyses i e the PMK the spouse partner and the household it will only be possible to produce child estimates from the NLSCY file The characteristics of the PMK spouse partner and household can be used to describe attributes of the child For example it will be possible to estimate the number of children living in a household with low income or the number of children for whom the PMK has scored high on the depression scale etc However it will not be possible to produce estimates of the number of low income households or depressed PMKs NLSCY Data Users Guide 32 2001 2002 Example b indicates the variable was new in Cycle 2 In subsequent cycles new variables will also be identified using the lowercase letter representing the cycle New variables in Cycle 3 will contain a c in Cycle 4a d etc Some r
81. ehension skills In this questionnaire school contact information was also gathered principal s name school address telephone number The Informed Consent Questionnaire Cognitive Two tests were administered to respondents in order to assess Measure cognitive measures They are lt Math and Reading Skills Indicator and lt The Peabody Picture Vocabulary Test Revised PPVT R Math and School children in grade 2 or higher were given a brief mathematics Reading and vocabulary reading test of about 12 questions This placement Skills test was designed to make it possible to determine the level of the Indicator math computation and reading comprehension tests that would subsequently be administered in the schools For grade 2 children the interviewer read the questions and recorded the answers on an answer sheet For children in grade 3 or above the child read the questions and gave the interviewer the answer NLSCY Data Users Guide 16 2001 2002 The Peabody The Peabody Picture Vocabulary Test Revised PPVT R was Picture administered by the interviewer to each selected child between 4 and Vocabulary 5 years old as well as to children aged 6 years and older who were Test Revised not yet in grade 2 The oral consent of the PMK was obtained before PPVT R the test was administered The purpose of the test was to assess the child s level of receptive vocabulary After having completed the full NLSCY interview and lea
82. eight is normalized by dividing his her weight CWTCWO01C by the average weight of all individuals Consequently the sum of the normalized weights is equal to the sample size NLSCY Data Users Guide 115 2001 2002 Score Calculation Missing Values Results Cycle 3 Three Factors Identified for 10 15 Year Olds Once the factor structures were analysed and the items included in each factor were determined the scores were calculated Imputation was done for missing values The imputed values were imputed using the SAS PRINQUAL procedure that determines which of the possible values for an item is the most plausible for an individual in view of his her response profile the response profiles of others in the sample and the number of factors included in the analysis If too many values were missing the final score was set to missing To produce the final scores 1 was subtracted from each item so that the lowest score would be 0 The final score was derived by totaling the values of all items with non missing values A score of 0 indicates the following for the three factors that were found to exist in the My Parents and Me scale a low degree of parental nurturance for the parental nurturance score a low degree of parental rejection for the parental rejection score a low degree of parental monitoring for the parental monitoring score In the sample of 10 15 year olds there were 5 539 children The sample was divided into
83. eke 104 My Parents and Me Scale BPRCbS07 and BPRCbS08 Parent 105 Child Scales from Self completed Questionnaire 00000 108 Feelings and Behaviour self complete 10 15 0 2 eee aeee 110 My Parents and Me self complete 10 15 0 0 cece cece eee 114 About me self complete 10 15 00 2 0 cece eee eee 117 Depression Scale self complete 12 15 0 cc eee eee 120 Ed cationi Parent iscissi ted ai ted ane aa wineries Sauce Sree ede 120 Socio demographic Characteristics 0 00 c cece cece 121 Labour Force Parent sicscwre enw aewre Sheieoure iene ct oee wee eee 122 Work Duration Derived Variables sacs s ov aaes cane vad wea aa soee es 124 Demographic Variables 1 2 4ocehected nee on a ete E ea Pehl a 125 Medic al Bi l gical mereen eese a a a eE EE Ea AE Ea EA o 126 Chapter 9 Dat amp Qualty int drost lace demia daada Opi e ook dimly Doig 129 Cross sectional and longitudinal response rates nunun nunnan 129 Component Non Response ssssrrseurrrsrrrrrrrrrrrrrrrrrrrr 135 Parent Questionnaire Response Rates nunna anann naaran 135 Child Questionnaire Response Rates 0 06 c cee eee eee eee 136 NLSCY School component sss coces Santee feel hek wee ieee Be bates 136 Response Rates for Math and Reading tests 0c cece eee eee 137 Response Rates for the Teacher s and Principals Questionnaires 144 Chap
84. eled as the PMK The intention was that the PMK would provide the information for all selected children in the household and then give socio demographic information about himself herself and him her spouse partner In some rare cases it might have been appropriate to label two different people in a household as PMKs For example in the case of a step family it may have been appropriate to label the mother as the PMK for one child and the father for another However in order to simplify the interview procedures only one PMK was selected per household Relationship The following is the breakdown of the relationship of the PMK to the of PMK to NLSCY children for Cycle 3 NLSCY Children For 93 0 of responding children the PMK was the mother 92 1 the biological mother and 0 9 the step adoptive or foster mother For 6 4 of the children the PMK was the father for 0 6 of children the PMK was not a parent Cases Where the PMK was not the Parent For the majority of cases of the PMK not being a parent the child had a parent living in the household but the parent was not selected as the PMK For the most part this situation occurred when a child had a very young mother living with her own parents i e the child s grandparents and the grandmother was selected as the PMK Spouse Partne ras PMK If the PMK had a partner residing in the household at the time of the interview then this person was labeled as the spouse Spouses
85. eneral APRCQ18 How often do you have to discipline your child repeatedly for the same thing The answer categories for these items were of two types 1 never 2 about once a week or less 3 a few times a week 4 one or two times a day 5 many times each day 1 never 2 less than half the time 3 about half the time 4 more than half the time 5 all the time NLSCY Data Users Guide 77 2001 2002 In the calculation of the score for this hostile ineffective parenting factor the categories were rescaled to 0 to 4 i e the category never was scored as 0 the category about once a week or less less than half the time was scored as 1 and the category many times each day all the time was scored as 4 In order to compute the score these values were summed across the seven items involved in the factor resulting in a hostile ineffective parenting score in the range 0 to 28 A score of 0 represents the absence of a problem and a score of 28 is the highest possible score with respect to problems For most of the scores calculated for the NLSCY a score of 0 represents the absence of a problem However there are exceptions to this which are noted in the documentation for each particular scale Example of Factor Score Computation Continued Note that the second item that loaded into the hostile ineffective parenting factor APRCQ08 Of all the times you talk to your child about his her behaviour what propor
86. ension test like the mathematics test was also developed in part from the CAT 2 However since the CAT 2 contain only English passages French passages had to be chosen from another source by educators at Universit de Sherbrooke The test is designed to measure basic reading skills The test s objectives cover information recall analysis of passages identification of the main idea interpretation of various types of writing and critical evaluation For each grade level the test developed for the NLSCY consists of four reading passages totalling 20 questions Each test consists of two original English passages and two original French passages in order to make the test as fair as possible In addition between two consecutive grades there are always two common passages with more or less 10 questions Similarly to the mathematics test each child who took the reading test was also given a raw gross score and a scaled score Since the CTC did not have standards for this test the approach of the item response theory IRT described for the mathematics tests was the only option for this test Unlike the approach of the mathematics test the three parameter logistical model was chosen for the reading test This model takes into consideration the difficulty and the discrimination of the item and a pseudo guessing parameter that seems more prevalent for this test A number of corrections were made to the English and French passages in Cycle 3 in
87. erally speaking each child in the NLSCY sample represents about 300 children in the population The The weighting phase is a step which calculates how many people each Weighting respondent represents As the target population is not the same for the Phase cross sectional sample and the longitudinal sample the number of persons each child represents is not the same Consequently two series of weights must be calculated lt one for the cross sectional sample lt one for the longitudinal sample These weights appear on the NLSCY data files CWTCW01C for cross sectional weight CWTCWO01L for longitudinal weight and must be used to derive meaningful estimates of the characteristics measured by the survey For example to estimate the number of children living in single parent families in 1996 we would select the records in the cross sectional sample of Cycle 2 with that characteristic and sum the weights found on those records NLSCY Data Users Guide 43 2001 2002 The Longitudinal Sample or Cross sectional Choice of Sample Dependant on Analysis Longitudinal Weight Cross sectional Weight Flows The choice of which sample to use depends on the type of analysis to be done The longitudinal sample pertains to the child population at the time this sample was selected i e 1994 95 The sum of the longitudinal weights is equal to the available demographic estimates for January 1995 Only the longitud
88. ers Cronbach s alpha to 0 656 For the hostile ineffective factor APRCS02 Cronbach s alpha was 0 394 It should be noted that there were only two items for this factor and the alpha can only be derived if one of the 2 items is removed After identifying the two factors the next step was to calculate scores for each Scores could be calculated for only 132 individuals for the positive interaction factor and for only 124 individuals for the hostile ineffective factor because of missing values for the items for these factors There were 18 135 children in the sample for the age group 2 to 11 The group was split into two sub samples of 9 090 and 9 045 individuals and the analysis for this age group was performed separately for each sub sample The non response rate for each of the eighteen items ranged from 2 1 to 2 7 Some individuals were excluded from the analysis that produced the factors The exclusion criteria were as follows individuals with two or more items coded missing under positive interaction and hostility and individuals with a single missing item under consistency After the criteria were applied there were 8 815 and 8 772 individuals left in the sub samples to be analysed Data were imputed for 12 items The number of imputations ranged between 1 and 16 A total of 91 values were imputed The factor analysis derived three factors for this age group positive interaction APRCSO3 and hostility APRCS04 and consisten
89. es a Estimates in the main body of a statistical table are to be rounded to the nearest hundred units using the normal rounding technique In normal rounding if the first or only digit to be dropped is 0 to 4 the last digit to be retained is not changed If the first or only digit to be dropped is 5 to 9 the last digit to be retained is raised by one For example in normal rounding to the nearest 100 if the last two digits are between 00 and 49 they are changed to 00 and the preceding digit the hundreds digit is left unchanged If the last digits are between 50 and 99 they are changed to 00 and the preceding digit is incremented by 1 b Marginal sub totals and totals in statistical tables are to be derived from their corresponding unrounded components and then are to be rounded themselves to the nearest 100 units using normal rounding c Averages proportions rates and percentages are to be computed from unrounded components i e numerators and or denominators and then are to be rounded themselves to one decimal using normal rounding NLSCY Data Users Guide 150 2001 2002 d Sums and differences of aggregates or ratios are to be derived from their corresponding unrounded components and then are to be rounded themselves to the nearest 100 units or the nearest one decimal using normal rounding e In instances where due to technical or other limitations a rounding technique other than normal rounding is used resulting
90. es followed for each of these steps are described in detail in the following pages NLSCY Data Users Guide 73 2001 2002 Factor Analyses Factor Analysis The factor structure of each scale was determined based on data for Scales from the first cycle The factor structure imposed on the scales already used in the first cycle and repeatedly utilized in the second cycle of the survey was the result of analyses of data from the first cycle The following is a summary of the procedures used in the factor analysis for each scale 1 The sample of respondents for each scale and age group if the scale used different questions for different groups was randomly divided into two half samples This was done to find out whether different samples would yield the same results 2 Principal component analysis was carried out separately on each half sample to find out how many factors should be extracted in the factor analysis performed subsequently In principle the same number of factors as was found in the literature was expected In practice however some scales showed a different number of factors because in some cases factors combined while in others new factors emerged 3 Factor analysis was done on each half sample and the factor structure and loading of each factor were compared across the half samples 4 In the factor analysis the items for each child in the appropriate age group were used multiplied by the child s normalized wei
91. es in this area could make a difference The depression rating scale includes twelve questions each of which contains four response categories In order for the lowest score value to be 0 the value for each question was reduced by 1 in calculating the score As well the answer categories were reversed for questions having a negative loading BDPPQ12F Q12H and Q12u The total score BDPPS01 may therefore vary between 0 and 36 a high score indicating the presence of depression symptoms NLSCY Data Users Guide 100 2001 2002 Results The factor structure of each scale was determined based on data from the first cycle The factor structure imposed on the scales already used in the first cycle and repeatedly used in the second cycle of the survey was the result of analyses done based on data from the first cycle Unweighted Data In analysing this scale unweighted data were used The sample size was 13 439 PMKs However once the observations containing mostly missing values were eliminated the analysis dealt with only 13 140 PMKs The non response rate for the various questions in the rating scale was roughly 2 0 whereas for the total score a non response rate of 2 2 was obtained There was no imputation for the variables in this rating scale In spite of the possibility of extracting more than one factor from the depression rating scale a single factor analysis was used since the interest was in developing a global depressio
92. et by the child s teacher in class using a multiple choice questionnaire NLSCY Data Users Guide 139 2001 2002 This test was a shortened version of the CAT 2 mathematical operations test The CAT 2 mathematical operations test measures the student s ability to do addition subtraction multiplication and division operations on whole numbers decimals fractions negatives and exponents Problem solving involving percentages and the order of operations are also measured The short version of the test developed for the purposes of the NLSCY now consists of 20 questions at each level except for level 9 10 which consists of 15 questions The tests were expanded in Cycle 3 to include overlapping items between each level An extra five items were added to each level selected from the test of the next level Mathematics test Each child who took the mathematics test was given a raw gross score a scaled score referred to as the classical scaled score and an IRT scaled score The raw gross score is obtained simply by adding the number of correct answers The Classically derived scale score and the IRT scaled score are described as follows The IRT derived scaled score The approach of the item response theory IRT was used successfully in Cycle 2 to derive scores for the reading comprehension tests Unlike the approach of the classical theory the IRT makes it possible to scale the scores without preset population standards Using comm
93. evisions were made to the content of the questionnaire between cycles If the revision resulted in a change to the meaning or the values of a question the variable was treated as new and contains a c Q refers to the variable type There are six possibilities Format for refers to the variable for a question that was asked directly on one of the Variable NLSCY questionnaires Names Cont S refers to a score calculated for one of the scales used on the questionnaire D means the variable was derived from other questions that were asked on the questionnaire I means the variable is a flag created to indicate that an item has been imputed X means the variable is a flag created to indicate an inconsistency in reported data between the current and previous cycles nnx refers to the question or variable identification Generally nn is a sequential number assigned to the variable and x is a sequential alphabetic indicator for a series of variables of a similar type NLSCY Data Users Guide 33 2001 2002 Acronym Names for Questionnaire Sections The following table gives the acronym names that were used for each section of the various NLSCY questionnaires This acronym is embedded in the variable name for all variables on the NLSCY data file The acronym is the second and third characters of the variable name Table 10 varia l Collected or Derived from the Geographic sample information Househ
94. f variation 164 2001 2002 Rule 3 Estimates of Differences Between Aggregates or Percentages Rule 4 Estimates of Ratios NLSCY Data Users Guide The standard error of a difference between two estimates is approximately equal to the square root of the sum of squares of each standard error considered separately That is the standard error of a difference a K 1 A is Te eS eT o ay do o is estimate 1 x is estimate 2 and alpha 1 and alpha 2 f Ky A respectively The A where are the coefficients of variation o coefficient of variation of d is given by Oy d This formula is accurate for the difference between separate and uncorrelated characteristics but is only approximate otherwise Where the numerator is not a subset of the denominator for example the ratio of the number of low birth weight female babies to that of low birth weight male babies the standard deviation of the ratio of the estimates is approximately equal to the square root of the sum of squares of each coefficient of variation considered separately multiplied by the ratio itself RIE The standard error of ratio is therefore r a Ta 5 Raay T ot where 1 and gt are the coefficients of variation of Xi the number of low birth weight female babies and Ay the number of low birth weight male babies respectively R Ons The coefficient of variation of is given by The formula will tend to
95. for Cycle 2 but did have a spouse for Cycle 3 A 6 had no spouse for Cycle 2 but did have a spouse for Cycle 3 3 2 0 7 was the same person for both survey cycles but had a different soouse ae NLSCY Data Users Guide 57 2001 2002 NLSCY Data Users Guide 58 2001 2002 Family Derived Variables Relationship Grid Single Parent Families Using NLSCY data a child s family may be described in several different ways Many of the family variables used to describe the NLSCY children were derived from what is known as the relationship grid As part of the household roster some basic demographic information was collected for all members of the child s household As part of this questionnaire the relationship of everyone in the household to everyone else was asked Using this information it was possible to create an extensive set of variables to describe the child s family situation The following are some of the family derived variables for the children that exist on this second micro data file for the NLSCY The names of the derived variable are given in brackets There are two ways to describe the parental situation of children using NLSCY data Using the relationship grid a child s single parent status was derived There were 84 4 of children living with two parents 15 4 with one parent and 0 2 without a parent CDMCD04 A child s parental status can also be defined in terms of the PMK There were 85 2 of the NLSC
96. fusals don t know and and Child file not stated Mark All That Apply questions were destrung and values converted to Yes 1 or No 2 responses Databases files were created for each section of the Adult and Child questionnaires Small data base files were created for each section of each Separate DBF questionnaire A record was kept for the section only if the files from Step section was applicable For example the section on 1 temperament was only applicable for children 3 months to 3 years old Therefore a temperament record was only created for children in this age group Within several sections different wording was used for different age groups For example in the Activities section Question 3 asks In the past 12 months outside of school hours how often has the child taken part in any clubs groups or community programs with leadership The wording for 4 to 5 year olds CAACQ3D1 was such as Beavers Sparks or church groups The wording for 6 to 9 year olds CAACQ3D2 was such as Brownies Clubs or church groups Initially these questions were stored as separate variables As part of the pre edit the two variables were collapsed into one output variable CAACQ3D The various wordings are given for these types of questions in the data dictionary in Appendix 4 The flow patterns for each section were processed and valid skips were assigned not applicable codes 6 96 996 NLSCY Data Users
97. g this factor NLSCY Data Users Guide 109 2001 2002 Feelings and Behaviour self complete 10 15 This section was part of the self complete questionnaire given to children in the 10 to 15 age group The objective of this section was to determine the child s perception of his her general behaviour and the child s engagement in risk taking behaviours Feelings and Behaviour This section replicates the behaviour checklist included on the Child s Questionnaire completed by the PMK for those aged 10 11 and the one on the Teacher s Questionnaire It is intended to provide indicators of the following behaviours conduct disorder hyperactivity inattention physical aggression indirect aggression emotional disorder anxiety and prosocial behaviours In Cycle 2 the factor scores were derived based on the factorial structure identified in Cycle 1 Behaviour Checklist Analysis of The following indicates the constructs or factors that the behaviour the NLSCY scale was intending to measure the items that were included in the Data factor and the sources for the items NLSCY Data Users Guide 110 2001 2002 Conduct disorder Items include AD1CQ01C E G L O T AA DD FF JJ and PP from the Ontario Child Health Study OCHS Hyperactivity Items include AD1CQ01B N P S and W from the Ontario Child Health Study and AD1CQ1HH from the Montreal Longitudinal Survey Emotional disorder Items include AD1CQ01F K Q V C
98. ght An individual s statistical weight is normalized by dividing his her weight AWTCW01 by the average weight for all individuals Thus the sum of the normalized weights is equal to the sample size NLSCY Data Users Guide 74 2001 2002 Distance Between Answer Categories Data Transformation Using Optimal Scaling 5 Once the factor structures were analysed and the items included in each factor were determined scores were calculated To produce the scores 1 was subtracted from each item so that the lowest possible score would be 0 A score of 0 indicates that the child has no problems for all factors in the behaviour scale except for the Prosocial factor where a score of 0 indicates the absence of prosocial behaviour Some items were imputed The imputed values were computed by a procedure the SAS PRINQUAL procedure that determines which of the possible values for an item is the most plausible for an individual in view of his her response profile the response profiles of others in the sample and the number of factors included in the analysis 6 The score for each factor on the scale was derived by totaling the values of the items that made up that factor including imputed values The score was set to missing if too many of the values of an items included in the factor were unreported A value may be missing if the parent refused to answer or did not know the answer to the item Factor analysis requires that the data ha
99. graduated from high school but had completed a ha aR degree or certificate then the post secondary degree or certificate took precedence For example if the PMK had completed only grade 10 but had masters then AEDPD04 was set to 18 It was decided that years of schooling was an interesting derived variable itself and therefore this variable has been included on the NLSCY master file for the PMK and spouse partner CEDPD04 and CEDSD04 NLSCY Data Users Guide 63 2001 2002 Occupational status is an important indicator of SES The occupation variable used in the derivation of SES was a modified version of a scale developed by Pineo Porter and McRoberts 1977 The classification system groups occupations described in Statistics Canada s 1980 Standard Occupational Classification into 16 somewhat homogeneous categories ordered from 1 to 16 where code 1 represents the highest level of occupation and code 16 the lowest The 16 category scale provides a ranking of occupations according to their social standing or prestige For the NLSCY for both the PMK and the spouse partner a detailed description was taken of the job considered to be his or her main job during the previous 12 months The information was used to code occupations into the 1980 classification and in turn into the 16 prestige categories For the purposes of deriving both SES the order of the Pineo Porter McRoberts scale was reversed The final scale used in the derivation of bo
100. he NLSCY Population They are the gt longitudinal sample and gt cross sectional sample Longitudinal The longitudinal sample consists of different cohorts Sample The first cohort consists of the children who were sampled in Cycle 1 at age 0 11 these children will be followed until they are 25 years of age The second cohort consists of children who were sampled in Cycle 2 at age 0 1 these children will be followed until they are 5 years of age The third cohort consists of children who were sampled in Cycle 3 at age 0 1 these children will be followed until they are 7 years of age possibly 9 years of age The longitudinal sample is also used for cross sectional purposes to cover specific age groups Cross From the Cycle 3 file we can produce cross sectional estimates for sectional ages 0 15 years A large sample of 5 year olds was included in Cycle Sample 3 to allow for reliable provincial estimates of this age group Non With each cycle there are respondents for which we are unable to response and collect information Based on interviewer notes from previous cycles missing we determine hard core non respondents exclude them from the information sample and do not attempt to trace them In most longitudinal surveys only respondents to the first cohort are followed and interviewed However a number of surveys including the NLSCY try to re contact people from the initial cohort even if they missed one or more waves of interv
101. he table headed Numerator of Percentage and follow the asterisks if any across to the first figure encountered This figure is the approximate coefficient of variation The coefficient of variation of an estimated proportion or percentage depends on both the size of the proportion or percentage and the size of the total upon which the proportion or percentage is based Estimated proportions or percentages are relatively more reliable than the corresponding estimates of the numerator of the proportion or percentage when the proportion or percentage is based upon a sub group of the population For example the proportion of female babies who were of low birth weight is more reliable than the estimated number of female babies who were of low birth weight Note that in the tables the c v s decline in value reading from left to right When the proportion or percentage is based upon the total population of the geographic area or age group covered by the table the c v of the proportion or percentage is the same as the c v of the numerator of the proportion or percentage In this case Rule 1 can be used When the proportion or percentage is based upon a subset of the total population reference should be made to the proportion or percentage across the top of the table and to the numerator of the proportion or percentage down the left side of the table The intersection of the appropriate row and column gives the coefficient o
102. herefore necessary to derive a correction factor that takes this overlap into account In addition one final factor is needed to ensure that these weights produce estimates consistent with the demographic estimates produced from other sources We are dealing with three types of households those selected in Cycle 1 those selected in Cycle 2 and those selected in Cycle 3 However the target populations for these three samples overlap in the cases of children selected in 1994 who are now 5 years of age and children aged 5 selected from the birth register in 1998 This overlap must be taken into account in order to ensure that our sample does not systematically overestimate the characteristics of the population NLSCY Data Users Guide 51 2001 2002 To take the relative contribution of each sample into account we Multiplier calculated a multiplier factor for each province If an optimal Factor combination of these samples is to be obtained this factor must take into account the accuracy of each sample s estimates For example an estimate from a highly accurate sample is considered more important than that from a sample of low accuracy Accordingly the former would have a high adjustment factor and the latter a low adjustment factor An example will illustrate this approach Let us suppose that 30 longitudinal children aged 5 years were sampled in New Brunswick in 1994 and 10 children from the same age group were selected from the birth
103. hild records were determined to be Longitudinal complete enough to be kept codes 000 and 001 These children Child came from 12 100 longitudinal households which is the number of Records households maintained in the Cycle 3 NLSCY files There were 18 612 child records for the responding longitudinal households Out of these there were 994 longitudinal child records that were not acceptable but were kept because there was at least one acceptable child record for the household Missing Variables on missing components for the household were imputed or Components set to not stated The longitudinal file also contains 194 records that were created for some longitudinal children for whom no data was collected in this cycle These are children who are now deceased or who have moved out of the country but who will be kept on the longitudinal file for weighting purposes For these records all variables except for the longitudinal weight CWTCWO01L have been set to not stated NLSCY Data Users Guide 28 2001 2002 Head Office Editing Stages of For CAI questionnaires for the NLSCY two stages of editing were Editing conducted lt Pre edit lt Consistency Editing The purpose of the Pre edit was to carry out some basic formatting and preliminary editing Table 9 outlines some of the procedures used Table 9 lt Non response values from the CAI system were recoded to complete Adult standard non response codes for re
104. hild s grade level type of school and language of instruction whether the child looks forward to school behaviour problems at school absenteeism parental hopes for the child s educational outcomes number of school changes and residential moves For children in grade 1 or higher additional questions were asked concerning other aspects such as skipping and repeating grades achievement special education parents perception of school climate and importance of good grades to parents NLSCY Data Users Guide 83 2001 2002 The Teacher s and Principal s Questionnaire s Child s Grade Education Section Behaviour Scale NLSCY Data Users Guide The Teacher s and Principal s Questionnaires provide additional information about the child and his her school achievement and behaviour At the data collection stage six different questions were asked to determine the child s grade This was done due to differences in grade classification among provinces At the processing stage these six questions were collapsed into one variable On the record layout an indication is given as to what the code means for each province For example if the grade code CEDCD01 is 10 this refers to secondary 1 for Qu bec and grade 7 for all other provinces A similar procedure was carried out for grade skipped CEDCDO2 and grade repeated CEDCDO03 The child s grade was also collected on the Teacher s Questionnaire There was not always c
105. hool component of the NLSCY collects information every two years on a nationally representative group of children within their school environment This information is used jointly with the information collected earlier about the same children in the Household component of the survey By collecting both the child s household information and school performance the NLSCY can obtain a more complete picture of the child s development Every child surveyed in the NLSCY from grades two to ten are given mathematics and reading tests In order for a test to be administered the consent from parents and the school board are required These tests were constructed by selecting a subset of questions from the Canadian Test Centre s Canadian Achievement Tests second edition CAT 2 The mathematics test is a shorten version of the CAT 2 mathematical operations test This test measures the student s ability to do addition subtraction multiplication and division operations on whole numbers decimals fractions negatives and exponents Problem solving involving percentages and the order of operations are also measured School component The reading comprehension test is developed in part from the CAT 2 Since the CAT 2 contains only English passages French passages were developed in co operation with educators at Universit de Sherbrooke The test is designed to measure basic reading skills The test s objectives cover information recall anal
106. icular factor is usually based on a series of items since one single item usually cannot measure the factor or construct with adequate precision During the development of the NLSCY when consideration was being made of what specific scales should be used to measure a particular concept scales were as much as possible selected that had been used in other studies where the psychometric properties of the measures produced by the scale were available with complete references Evaluation of In many instances the wording of certain questions was modified Scale Data and in some cases new questions were added Sometimes the scale that was used had not previously been used for children in Canada or had only been used for very small samples Given these concerns and further concerns regarding interviewing conditions it was felt that the factor structures of the scales used in the NLSCY could be different from the ones given in the literature Therefore the project team felt the need to carry out an extensive evaluation of the scale data to ensure that the psychometric properties found in other studies also held true for the NLSCY experience There were three major steps in the analyses of the scale data First a new factor analysis was performed on all scales to determine the constructs or factors inherent in each scale Then scale scores were calculated based on this factor structure Finally reliability measures were produced The general procedur
107. id you have that job one year ago without a break in employment since then There is moreover a derived variable CL FPD33 for indicating the number of weeks worked by the PMK in a job or company the previous year In the first cycle of the survey an employment vector of 53 weeks was established based on information about each job held to a maximum of six jobs To reduce the respondent s response burden this collection method was abandoned in favour of a more general section A good many variables derived from Cycle 1 were reproduced but it should be noted that while considerable effort was made to keep the same definitions the collection tool was changed substantially With the current collection tool it is still possible to gather labour force data for the previous year but in a more general way A series of questions was used to determine the number of weeks worked in the 12 previous months the number of weeks the individual was absent from work the number of weeks the individual was without work but seeking employment and so on Moreover the tool focuses on the current main job or if applicable the most recent job A detailed description of this job was obtained employer type of company nature of the work main duties status hours worked salary 124 2001 2002 Other Derived Variables This release includes other derived variables which describe the employment picture over the reference year such as number
108. iew In Cycle 3 and Cycle 4 an attempt was made to re interview people who responded to Cycle 1 but not Cycle 2 or 3 NLSCY Data Users Guide 3 2001 2002 Cycle 1 Cycle 2 Why use these children for Cycle 2 Longitudinal In Cycle 1 22 831 children were interviewed After sub sampling 16 903 respondents to Cycle 1 form the longitudinal cohort that will be followed until these respondents reach the age of 25 Close to 5 000 children used in the sample for Cycle 1 were selected from the National Population Health Survey NPHS That sample 5000 children was given back to the NPHS These 5000 children will remain part of NPHS and will not be followed and interviewed by NLSCY Due to costs and response burden the sample for Cycle 2 was reduced To try and decrease response burden for families that had more than 2 selected children a sub sample from Cycle 1 was taken to keep only two children per household for the Cycle 2 interviews The siblings selected for Cycle 2 were part of a responding longitudinal household this was an inexpensive way of adding children to the sample It also allowed us to continue the comparisons of children within a family versus between families These children are not considered longitudinal respondents even though they live with a longitudinal family because they were not eligible in Cycle 1 The new sample of children from the Labour Force Survey was included to ensure an unbiased sample
109. in estimates to be published or otherwise released which differ from corresponding estimates published by Statistics Canada users are urged to note the reason for such differences in the publication or release documeni s f Under no circumstances are unrounded estimates to be published or otherwise released by users Unrounded estimates imply greater precision than actually exists Sample Weighting Guidelines for Tabulation Sample Design The sample design used for the NLSCY was not self weighting When producing simple estimates including the production of ordinary statistical tables users must apply the proper demographic load If proper weights are not used the estimates derived from the microdata file cannot be considered to be representative of the survey population and will not correspond to those produced by Statistics Canada In effect the weight assigned to each child reflects the number of children represented by a particular respondent For any analysis dealing with correlation analysis or any other statistics where a significance measure is required it is recommended that a sample weight be used This weight is obtained by multiplying the demographic load by the sample size and dividing this total by the total estimated population This produces a mean weight of 1 and a sum of weights equal to the sample size NLSCY Data Users Guide 151 2001 2002 Benefit of Using an Adjusted Weight Software Difference
110. inal children i e those selected in 1994 are givena longitudinal weight other than 0 For each cycle the longitudinal weight of the panel is recalculated to take into account the further erosion non response that occurs between the two cycles of the survey i e about two years It is this one that is usually better suited to longitudinal analysis based on a comparison of the data for more than one year as it allows for the life courses of the children to be quantified over time The cross sectional sample makes it possible to do estimates based on data from a single cycle A separate cross sectional weight is calculated for each cycle For Cycle 1 the longitudinal sample and the cross sectional sample have the same target population As the target populations are identical only one series of weights was needed for this cycle Flows may be calculated using cross sectional estimates produced for two cycles However the flows thus measured are net flows They are calculated based on a snapshot taken for each reference period As a result they mask all transitions that cancel each other out Here is an example to illustrate this phenomenon A researcher wishes to know whether the number of young people who smoke increased between 1994 and 1996 He can therefore calculate the number of smokers in 1994 using the Cycle 1 sample and a second estimate for 1996 using the cross sectional sample for Cycle 2 By comparing these two estimates he
111. inal scale to determine whether alcohol consumption had an effect on global family dynamics However it was not used in the analysis of the scale This scale is aimed at providing a global assessment of family functioning and an indication of the quality of the relationships between parents or partners For this reason and because of the small number of questions no attempt was made to measure the various aspects of family functioning Other surveys have shown that the relationship between family members has a considerable effect on children The results of the Ontario Child Health Study have shown for example that there is an important link between family dysfunction and certain mental conditions in children The family functioning scale was administered to either the PMK or the spouse partner as part of the Parent Questionnaire The unit of analysis for the scale is the family The scale includes twelve questions each of which contains four response categories In order for the lowest score value to be 0 the value of the categories was reduced by 1 in calculating the score The order of the categories was reversed for questions having a negative loading BFNHQ01A Q01C Q01E Q01G Q011 and Q01K The total score BFNHS01 may therefore vary between 0 and 36 a high score indicating family dysfunction 2001 2002 Results Non response Rate Cronbach s Alpha for Family Functioning Scale Distribution of Values for the Fa
112. ince and age groupings used to produce the Approximate Sampling Variability Tables First the tables for the cross sectional samples Design Effect Sample Size Population CROSS SECTIONAL SAMPLE Province Design Sample Size Population Effect Newfoundland 1 001 100 089 1 Prince Edward Island 2 2 545 Nova Scotia 1 293 167 311 New Brunswick 1 664 Qu bec Ontario Manitoba Saskatchewan Alberta British Columbia NLSCY Data Users Guide 161 2001 2002 CROSS SECTIONAL SAMPLE Age Group Design Sample Size Population Effect 740151 740 151 3 866 0 to 23 years Total 0 to 13 4 1 20 025 5 482 536 ears Design effects for the longitudinal sample are as follows CYCLE 1 LONGITUDINAL SAMPLE Province Design Sample Size Population Effect Newfoundland 43 ae Nova Scotia 29 1068 14722 New Brunswik 23 958 m5913 49 ae Er as Prince Edward Island 2 0 2 0 2 9 2 3 Qu bec 4 9 Ontario 4 2 Manitoba 3 4 2 8 3 2 3 6 Saskatchewan Alberta 32 1465 489604 British 1 333 574 160 Columbia provinces Prairies NLSCY Data Users Guide 162 2001 2002 CYCLE 1 LONGITUDINAL SAMPLE Age Group Design Sample Size Population Effect 2 to 3 years 3 654 752 598 4 to 5 years 10 to 13 years 4 1 4 519 1 565 342 Total 2 to 13 15 468 4 673 390 ears All coefficients of variation in the Approximate Sampling Variability Tables are approximat
113. incipal In cases where the child was in grade 2 or higher the PMK was asked to give permission to allow the teacher to administer a skills test in math computation and reading comprehension to the child The school collection involved three questionnaires These questionnaires were mailed out to teachers and principals who were asked to complete the questionnaires and mail them back to Statistics Canada in the envelopes provided Collection Questionnaire packages were mailed to principals with instructions on Strategy for how the various instruments should be completed The principals were the School then asked to distribute the questionnaires and tests to the teachers Collection Approximately one week after the initial mailing a postcard was sent out to thank all respondents and to remind those who had not yet responded to do so Roughly two weeks later a second questionnaire package was sent out to teachers and principals who still had not responded Finally three weeks later non responding teachers and principals were contacted by telephone and encouraged to participate Teacher s The goal of the teacher s questionnaire was to collect information Questionnaire about the child s academic achievement and behaviour at school as well as information on characteristics of the class and the teacher s instructional practices There were three teacher questionnaires which were completed depending on the circumstances of the child e a
114. ion This correction affected only two households For Cycle 3 a maximum of 2 children were to be interviewed in the new households If the economic family has more than 2 eligible Weighting of Children Sampled in 1994 and 1996 Weighting of Two correction factors are required for these children The first inflates Children the survey weight in order to account for the non response observed Selected during data collection in Cycle 3 For this adjustment the from the homogeneous response group method is used once more However Birth as there is little information on non responding households from this Register in sample HRGs in this particular case correspond to the strata used to 1998 select the birth register sample The second adjustment calculated for these children accounts for the fact that we interviewed twins The basic weight was modified for couples of twins as households with twins have a higher probability of selection than those with just one eligible child Weighting of Itis not necessary to apply all the corrections described in the previous Children section to these children as this was done in Cycles 1 and 2 The Sampled in basic weight we use is therefore the weight obtained in previous 1994 and cycles after the adjustment for non response and before post 1996 stratification Only two corrections were necessary for these children 15t Correction The first correction inflates the basic weight in order to accou
115. ionship or adoption Foster children are considered to be part of the economic family For example if a woman lives in a household with her spouse and two children as well as her sister and her sister s child then all of these individuals would be part of one economic family If a boarder also resided in the household with her child then this would constitute a second economic family Siblings For the NLSCY data siblings include full half step adopted and foster siblings Only siblings residing in the household have been included in the calculation of the sibling derived variables included on the micro data file In the case of common law relationships if both members have brought their own children into the relationship then these children are considered siblings It should be noted that the classification of siblings was age independent If an NLSCY child had an adult sibling for example 21 years of age living in the household then this sibling was included in the calculation of the sibling derived variables The sibling derived variables include total siblings as well as number of older siblings younger siblings and siblings of exactly the same date of birth i e twins CDMCD08 09 10 and 11 Socio Economic Derived Variables NLSCY Data Users Guide 61 2001 2002 There were two derived variables produced from Cycle 1 data to assist analysts in understanding and explaining the socio economic situation of the child s family socio
116. is would be the case notably if all children were living in households that experienced on average an improvement in socio economic status It is therefore essential to be familiar with the rules used to derive the two SES values in order to use the variables properly in the analyses The differences observed from one cycle to the other for the standards of both SES are not yet very pronounced Therefore the use of one measure rather than another in the short term should not produce significant differences in research results But over the long term the proper use of both measures should become more important Normally it is recommended that cross sectional SES be used to accurately measure the relative position of a child in relation to other children in a given cycle whereas the use of longitudinal SES provides a better indication of the progression of an individual s situation from one cycle to the other Two SES Values NLSCY Data Users Guide 68 2001 2002 Missing values i e not stated values were ignored in the standardization In the standardization of the spouse partner variables CEDSD04 and PINEOLOG SP if the PMK did not have a spouse partner these records were ignored The SES composite was then calculated by taking the unweighted average of the five standardized variables If one of the five variables had missing data due to non responses refusal don t know etc then the average was taken over the remaining n
117. is section which were part of a scale Items BFFCQ01 BFFCQ02 BFFCQ03 and BFFCQ04 are intended to measure how well the child gets along with peers It is part of the Peer Relations Sub scale from the Marsh Self Description Questionnaire developed by H W Marsh The object of the friends scale is to measure how well the child feels he she gets along with his her peers In order to understand how the factorial structure was determined in Cycle 1 below is a description of the items that were included on the questionnaire in Cycle 1 to measure peer relations the analysis used to construct the scale and the results of these analyses In Cycle 1 questions AA1CQ01 to AA1CQ04 were used to construct the scale This set of questions on getting along with peers is the Peer relations Subcale from the Marsh Self Description Questionnaire To construct the Friends Scale for the NLSCY a factor analysis was conducted to test the theoretical construct In the factor analysis the items were multiplied by the child s normalized weight 2001 2002 Once the factor structures were analysed and the items included in the factor was determined the score was calculated No imputation was done on the values If any values were missing the final score was set to missing A value may be missing if the child refused to answer or did not know the answer to the question Score Calculation To produce the score 1 was subtracted from each item so that the lo
118. ized weights is equal to the sample size Once the factor structures were analysed and the items included in each the factor were determined the scores were calculated No imputation was done for missing values If any values were missing the final score was set to missing To produce the final scores 1 was subtracted from each item so that the lowest score would be 0 The final score was derived by totaling the values of all items with non missing values A score of 0 indicates the following for the two factors that were found to exist for in the About me scales a lack of general self esteem for the general self scale and a negative perception of physical appearance for the physical appearance score In the sample there were 3 434 children aged 10 or 11 years They were divided into two sub samples of sizes 1 705 and 1 729 and analysis was done on each sample The non response rates for the 8 items ranged from 14 to 15 8 Individuals with missing values were excluded from the analysis conducted for the purpose of constructing the factor After these exclusions the sub samples contained 1 371 and 1 413 individuals respectively for analysis purposes As a result of factor analysis two factors were identified the general self factor and the physical appearance factor The items that comprised each factor are described in the following table NLSCY Data Users Guide 118 2001 2002 GENERAL SELF SCALE FOR CHILDREN AGED 10 AND 11 YEARS O
119. le to compute an exact rate since the exact denominator of this rate is unknown Longitudinal Response Rate In keeping with the custom for longitudinal surveys we decided to publish the response rate among respondents for Cycle 1 In the table below which gives these rates by province the percentage reported is the ratio between the number of respondents for the cycle in question and the number of respondents in Cycle 1 NLSCY Longitudinal Response Rate by Province Children Selected in Cycle 1 Province No of Respondents No of Respondents No of Respondents in Cycle 1 in Cycle 2 in Cycle 3 467 NLSCY Data Users Guide 132 2001 2002 Non Response Non response is a type of error that can result in bias in survey Bias estimates Biased estimates can occur when the characteristics of non respondents differ significantly from those of survey respondents Bias resulting from non response during the first contact was dealt with in the manual for the first cycle As few households were added for the second cycle and since similar results would be obtained this study is not taken up for the second cycle A considerable amount of information is available to evaluate this potential bias As a result we attempted to model the non response to Cycle 2 event using variables obtained during the first collection cycle In this context the non response event may have two causes a the decision made by the respondent not
120. mily Functioning Scale The factor structure of each scale was determined based on data from the first cycle The factor structure imposed on the scales already used in the first cycle and repeatedly used in the second cycle of the survey was the result of analyses done based on data from the first cycle In analysing this scale unweighted data were used The sample size for the scale was 13 439 families However once the observations containing missing values were eliminated the analysis dealt only with 13 190 families The non response rate for the different variables was between 1 3 and 1 4 whereas for the total score a non response rate of 1 9 was obtained There was no imputation for the variables in this scale Following single factor analysis all 12 variables of the scale were kept since the loading values were well above the established threshold The Cronbach alpha coefficient calculated using SAS software was 0 88 The variable AFNHQ0O1L showed the highest correlation 0 66 with the total score once the variable was removed whereas the variable showing the lowest correlation was AFNHQ01A with a correlation of 0 51 The Cronbach alpha coefficient calculated by omitting one variable was stable at about 0 87 for the 12 variables When the values for the factor score for the family functioning scale are examined for the NLSCY children the distribution that is observed is not a continuous one In fact the most common sc
121. n an unskilled manual position household income is approximately 25 000 neither the PMK nor the spouse have completed high school neither the PMK nor the spouse are in the labour force household income is approximately 15 000 there is no spouse the PMK has not completed high school the PMK is not in the labour force the household income is less than 10 000 NLSCY Data Users Guide 70 2001 2002 Introduction Factors Affecting Child Growth NLSCY Processing System Chapter 8 Content and Validation of NLSCY The NLSCY was designed to follow an ecological or holistic approach to measuring child development The survey captures the diversity and dynamics of the factors affecting children To ensure that all relevant topic areas affecting child development were adequately addressed by the survey a multidisciplinary consultation was carried out at the inception of the survey The selection of specific subject areas priorities and survey questions was very much a group effort with input and advice from the NLSCY expert advisory group which consists of researchers in the area of child development and the social sciences federal departments representatives from the provinces and territories responsible for child development programs It was recommended that the NLSCY cover a broad range of characteristics and factors affecting child growth and development Extensive information was gathered a
122. n index Following the analysis the 12 variables of the scale were all kept as components of this factor since all 12 loading values met the established threshold The Cronbach alpha coefficient calculated using SAS software was 0 82 The variable ADPPQ12D showed the highest correlation 0 68 with the total score once the variable was removed whereas the variable showing the lowest correlation was ADPPQ12L with a correlation of 0 33 The Cronbach alpha coefficient calculated by omitting one variable was between 0 79 and 0 82 for the 12 variables Single Factor Analysis Weighted data could not be used since the weights developed for the NLSCY are for children only and not for parents NLSCY Data Users Guide 101 2001 2002 Family Functioning Scale Parent Introduction Global Assessment of Family Functioning Effect of Family Relations on Children Administerin g the Family Functioning Scale NLSCY Data Users Guide 102 Questions related to family functioning i e BFNHQO01A to BFNHQ0O1L were developed by researchers at the Chedoke McMaster Hospital of McMaster University and have been used widely both in Canada and abroad This scale is used to measure various aspects of family functioning e g problem solving communications roles affective involvement affective responsiveness and behaviour control Question BFNHQO1M drawn from the Follow up to the Ontario Child Health Study was added to the orig
123. nagement Transmission of Cases Data collection for the household component of the NLSCY relied heavily on CAI technology The use of computer assisted personal interviewing CAPI technology allows for high quality collection of complex population specific content sections For example the system facilitates the collection of the relationships of all household members to each other i e the relationship grid This wealth of information will enable a detailed analysis of family structures an important concept for analysis of the child information This type of collection would be very difficult to implement in a paper and pencil environment The CAI system has two main parts 1 Case Management and 2 the survey specific components The Case Management system controls the case assignment and data transmission for the survey For the NLSCY a case refers to a household selected for the NLSCY sample The Case Management system also automatically records management information for each contact or attempted contact with respondents and provides reports for the management of the collection process The Case Management system routes the questionnaire applications and sample file from headquarters to the regional offices and from the regional offices to the interviewer laptops The returning data take the reverse route To assure confidentiality all data is encrypted for transmission The data are unencrypted only once they are on a
124. nce hours worked and work arrangements e g shifts during the previous year Information was collected on the main job and on all jobs for a one year period Respondents were asked to identify what they considered to be their main job over the previous year if they had more than one job A complete description was recorded for this main job and industry and occupation coding was carried out using 1980 Standard Industrial Classification codes and 1980 Standard Occupational Classification codes Data on wages and salaries for this main job were collected Wage rate data provides an additional source of information on income This data will be useful in analysing choices which parents particularly mothers face in deciding to stay at home or to return to the labour force NLSCY Data Users Guide 123 2001 2002 Work Duration Derived Variables Work Duration Derived Variables Jobs Held During the Previous Year Response Burden Current Collection Tool NLSCY Data Users Guide With the data collected in the Labour Force Section it was possible to create a series of derived variables to describe the stability of work for the PMK and spouse partner over the previous year As mentioned above a series of questions were asked about all jobs the PMK and spouse partner held during the previous year As well in order to address absences within a job the following question was asked as the initial lead in question to a job D
125. ndividuals Consequently the sum of the normalized weights is equal to the sample size Once the factor structures were analysed and the items included in each the factor were determined the scores was calculated Some items were imputed The imputed values were imputed using the SAS PRINQUAL procedure that determines which of the possible values for an item is the most plausible for an individual in view of his her response profile the response profiles of others in the sample and the number of factors included in the analysis To produce the final scores 1 was subtracted from each item so that the lowest score would be 0 The score for each factor on the scale was computed at by totaling the values of the items that made up the factor including imputed values The score was set to missing if too many of the values of any items included in the factor were unreported A value may be missing if the child refused to answer the item A score of 0 indicates that the child has no problems for any of the factors in the behaviour scale with the exception of the prosocial factor where a score of 0 indicates the absence of prosocial behaviour In the sample there were 3 434 children aged 10 or 11 years They were divided into two sub samples of size 1 705 and 1 729 and analysis was done on each sample The non response rates for the 8 items ranged from 13 6 to 16 7 Individuals with missing values were excluded from the analysis conducted f
126. nt In some cases a value was captured that was not allowable for a particular item This was possible due to the fact that the scanning operator was given the ability to overwrite the edits These invalid entries were set to not stated values in the pre edit Editing for flow patterns was carried out at the consistency editing stage for the paper questionnaires One data file was produced for the 10 11 12 13 and 14 15 questionnaires For questions that did not apply to an age group the variables were set to not applicable codes 6 96 996 In this cycle there were 3 Teachers questionnaires with many of the same questions These are to be released in July 2001 Questions that were not asked from a teacher were set to not applicable codes 6 96 996 NLSCY Data Users Guide 31 2001 2002 Naming Convention and Coding Structure for NLSCY Variables The NLSCY microdata file documentation system has employed certain Introduction standards to label variable names and values The intent is to make data interpretation more straight forward for the user A naming convention has been used for each variable on the NLSCY Naming data file in order to give users specific information about the variable Convention All variable names are at most eight characters long so that these for Variables names can easily be used with analytical software packages such as SAS or SPSS C SE C Q nnxorB SE C b Q nnx Format for Variable C r
127. nt for non response The adjustment used in this step is identical to that calculated in determining longitudinal weight NLSCY Data Users Guide 50 2001 2002 ond Correction Weight Integration Correction For Overlap Of Target Populations The second adjustment attempts to minimize the impact of rare interprovincial migrations Some children selected in 1994 or 1996 had moved or changed province since the first interview This can sometimes distort weights for the new province of residence For example the weight of a child selected in Ontario is far greater than that for a child selected in Prince Edward Island When a child selected in Ontario moves to Prince Edward Island this will have an enormous impact on the estimates for Prince Edward Island if he she retains his her original weight This type of migration is very rare among the target population In this context it is not reasonable to assume that the sampled child who has moved from Ontario to Prince Edward Island represents a large number of children in the target population who have followed the same life course Rather such a case should be considered uncharacteristic The weight of these children has therefore been corrected downward Using the three weight calculation methods presented in the previous sections it is possible to produce estimates for their respective target population In some cases however these target populations are not mutually exclusive It is t
128. of weeks worked part time number of weeks worked full time etc Demographic Variables Demographic Variables Edits on the Relationship Grid Data Examples of Editing The demographic variables discussed in this section refer to variables collected on the household roster As part of the household roster some basic demographic information e g age gender marital status was collected for all members of the child s household The relationship grid was also completed as part of this questionnaire i e the relationship of everyone in the household to everyone else Using this information it was possible to create an extensive set of variables to describe the child s family situation Most of these derived variables are critical to the analyses of NLSCY data and are described in Section 8 NLSCY Concepts and Definitions If was necessary to perform an extensive series of edits on the data that were collected as part of the relationship grid There were some edits that were carried out as part of the CAI system during collection However in the data that were received at Head Office there were still inconsistencies The following are some examples of the types of editing that was carried out in all relationships reported a person could not have more than two parents and the difference in age between a husband and wife had to be less than 29 years In total there were over 30 relationship edits performed Some of the e
129. of the survey Of the selected children in these Cycle 2 households a maximum of two were chosen for whom data was to be collected The first contact was established with these households using the address and telephone number provided during Cycle 2 Next the interviewer confirmed that at least one member of the household list provided in 1996 97 still lives at the address If none of the individuals on the list were in the household contacted the file for the household was transferred to the trace folder and the interview with the household was ended If one of the individuals on the list was a member of the household contacted the interview continued beginning with the confirmation or updating of the contact information mailing address and residence telephone number and the updating of the list of household members During this final phase if one of the children selected was no longer part of the household information as to why parents separation departure for a foster home etc the date of the child s departure and the child s new address or other relevant information for tracing them was obtained Then the new members of the household were added to the list If at least one of the selected children was no longer a member of the household a new household file was created and transferred to tracing The Tracing file includes all household members from the first cycle who were no longer part of the contacted household The inte
130. old dwelling characteristics MM Variables collected as Basic demographic variables for each household part of the household member These variables are included on the roster NLSCY data file for the child the PMK and the spouse partner Demographic derived information of the household roster and to explain the living relationship grid arrangements of the child ball ie lore and spouse partner on the Adult Questionnaire Health PMK and Spouse on the Adult questionnaire and for the Child on the Child questionnaire Z oono PMK and Spouse in the Health section of the Conditions Adult questionnaire Restriction of Activities PMK and Spouse in the Health section of the Adult questionnaire Depression scale Parent Questionnaire this scale was administered to the PMK ED Education children 4 to 13 years old on the Child s Questionnaire and about the PMK and spouse partner on the Adult Questionnaire Labour force PMK and spouse partner on the Adult Questionnaire pa er PMK collected on the Adult Questionnaire Family functioning scale Adult Questionnaire this scale was administered to the PMK or teas es to measure how family members relate to each other MD Medical biological Child s Questionnaire 0 to 3 years of age NLSCY Data Users Guide 34 2001 2002 T develooment PR_ Parenting style CR Child care Child s Questionnaire 0 to 13 years PPVT test 4 to 6 years old if child in
131. ollection Cycle Collection Cycles NLSCY Data Users Guide 1 This manual will assist National Longitudinal Survey of Children and Youth data users It s purpose is to lt document data quality and other analytical issues regarding the NLSCY and lt facilitate the manipulation of the micro data files The National Longitudinal Survey of Children and Youth NLSCY is a long term study conducted in partnership by Human Resources Development Canada HRDC and Statistics Canada The primary objective of the NLSCY is to monitor the development and well being of Canada s children as they grow from infancy to adulthood The NLSCY is designed to follow a representative sample of Canadian children aged newborn to 11 years into adulthood with data collection occurring at two year intervals Each collection cycle used by NLSCY consists of a number of months sometimes over the period of two calendar years during which interviews with respondents are completed Each cycle marks the beginning of the collection phase when the longitudinal survey respondents are followed up Collection Start Collection End 1 December 1994 April 1995 2 December 1996 April 1997 October 1998 June 1999 2001 2002 Objectives of the NLSCY Data Release Strategy Contact Person at Statistics Canada Contact Person at Human Resources Development Canada NLSCY Data Users Guide 2 The objectives of the NLSCY are gt
132. on complete interview was not obtained for a variety of reasons Some respondents were willing to give only a certain amount of time to the completion of the survey In some cases an interviewer completed a portion of the survey with the respondent and made an appointment to continue at another time but was unable to re contact the respondent Criteria for It was necessary to come up with a criteria for deciding what to do Partial with these partial interviews If the majority of the survey had been Response completed obviously the preference was to keep this case and label itas a responding household However if only very minimal information was collected the decision was made to drop the household and treat it as a non responding household In order to make this assessment the data collected for each selected child in the household were examined This was done by looking at certain key questions across the Child Questionnaire An assessment was made as to whether or not there was an adequate amount of information collected for at least one child in each household If there was the household was maintained in the responding sample NLSCY Data Users Guide 27 2001 2002 Missing All missing variables for households were set to not stated or Variables imputed If there was not adequate information for at least one child then the household was dropped from the responding sample and treated as a non response In total 17 618 longitudinal c
133. on missing items If there was no spouse partner in the household i e the PMK had no spouse partner then the average was taken over the three applicable variables CEDPD04 PINEOLOG PMK and HHINC For two parent families i e for cases where there was a PMK anda spouse partner if two or more out of the five input variables were missing then SES was set to not stated For single parent families i e there was no spouse partner if one or more out of the three input variables were missing then SES was set to not stated Consideration of Missing Data The values for SES range from 2 000 to 1 750 The distribution of SES scores is as follows for children on the file Examples of SES SES SCORE RANGE CHILDREN WITH SCORE IN RANGE L Cross Sectional Longitudinal 5 or over 2 2 3 T Llar olele a gt nw alol ajlSl g L S S 5 92 o plololoa D V o o es n n a gt o aR SISIEN 2 o ola D N 7 En gt D 3 L o Note These numbers are based on unweighted data With this procedure the SES score for single parent families will tend to be lower because household income on average will be lower However the SES score will properly reflect the level of education and the occupational prestige of the single parent Nevertheless for most regression analyses where SES is used as a control variable it would be useful to include a dummy variable denoting whether the family
134. on test items linking grades standards are estimated from the entire population of children taking the test for this cycle Scores are derived ranking each child within a level then the scores are scaled to reflect the progression of scores throughout all the levels In order to ensure comparability from year to year each sample from each cycle must represent equivalent populations Among the single dimension models the two parameter and the three parameter logistical model were chosen for math and the reading tests respectively The two parameter model takes into consideration both the difficulty and the discrimination of the item while the three parameter model also considers the pseudo guessing component In this way the IRT takes into consideration the pattern of responses Two children with the same raw gross score will not have the same scaled score unless they answered exactly the same way For example a child who only answered the 5 easiest questions correctly would have a lower scaled score than the one who only answered the 5 hardest questions correctly The scaled scores of the Cycle 3 mathematics test ranged from 100 to 600 NLSCY Data Users Guide 140 2001 2002 The classically derived scaled score The scaled score is derived from standards norms established by the Canadian Test Centre CTC The CTC developed these standards from a sample of Canadian children from all 10 provinces however the test has been develope
135. on the homogeneous response group HRG method which involves an attempt to consolidate those individuals with the same propensity to respond These groups are formed using the characteristics for each child reported in Cycle 1 A correction factor is then derived for each HRG as follows Homogeneous Response Group HRG Sum of adjusted weights in the HRG Sum of adjusted weights of respondents in the HRG Two HRG Sets Two distinct HRG sets were constructed one for the sample of children selected in Cycle 1 and another for the sample of children selected in Cycle 2 Both sets are required as these samples do not necessarily react to the same non response mechanism As there is every reason to believe that this mechanism changes in accordance with the number of times an individual is surveyed the non response adjustment model must take this fact into account Lastly the constraints represented by adjustment factor range and minimum HRG size are imposed during HRG formation in order to obtain reasonable reliable correction factors Post stratification The purpose of the second adjustment factor is to ensure consistency between survey estimates and demographic estimates produced by Statistics Canada This method is known as post stratification For the sample of children selected in Cycle 1 the target population is the set of children aged 0 to 11 in early 1995 As a result the post stratification adjustment for this sample ensure
136. onnaire s Questionnaire Grooming For this second cycle of the NLSCY edits were also performed to ensure consistency between cycles for data that was not expected to change Data from the previous cycle feedback variables were included in the CAI system for the current cycle When inconsistencies were identified the interviewer was asked by the system to confirm the Cycle 2 data with the respondent through a series of questions Example For the Chronic Conditions questions if a chronic condition such as asthma was reported in the previous cycle but not indicated as being present in the current cycle the system prompted the interviewer to ask questions to determine if the current data was in fact correct or if the condition had changed since the previous cycle Some questionnaires for the NLSCY were completed on paper and pencil questionnaires PAPI The 10 11 12 13 and 14 15 year old Self Completed Questionnaires the Teachers Questionnaires and the Principals Questionnaire were all completed by PAPI All of these documents were completed directly by a survey respondent Data capture for these questionnaires were accomplished through scanning at a centralized area at Statistics Canada s Head Office Prior to scanning the documents were groomed and verified for completeness During this process any document containing at least one respondent completed item was scanned and a file containing each record was provided to H
137. onse rate of at least 70 Strata that are too small or have a response rate of less than 30 are grouped together until these restrictions are met Following the survey it is possible that information is gathered only for one child in a household although two children are in the sample According to the NLSCY release strategy both these children are considered respondents as we have considerable information about their parent s For this reason it is not necessary to apply a correction factor for the non response of the children gt These restrictions are designed to ensure that the adjustment factor is relatively stable and not too large NLSCY Data Users Guide 48 2001 2002 3 Correction For Households With More Than One Economic Family 4 Correction For Households children 2 children are chosen at random This correction takes this With More selection process into account in economic families and affects only Than Two 5 households as very few households have more than two children Eligible under 1 year of age Children oe This is the last correction to be calculated for these children before weight integration NLSCY Data Users Guide 49 2001 2002 Sometimes a household includes more than one economic family When this occurs the child selection procedure requires the selection of one of these families at random This correction is the inverse of the selection probability of the family in the household in quest
138. onsistency across the data collection units on what the correct grade was In the edit priority was placed on what the teacher said in the case of discrepancies On the micro data file the variables on language of instruction BEDCQ12A and type of school BEDCQ08 were set to not stated because of confidentiality concerns In the Education Section there was one question BEDCQ13 which asked the number of days the child had missed since the beginning of the school year The answer to this question obviously depends on the collection date which has not been included on the micro data file because of confidentiality concerns Therefore this variable has been suppressed and a derived variable was created BEDCD04 to indicate the percent of days missed since the beginning of the school year 84 2001 2002 Objective of the Behaviour Scale Similar Behaviours Factor Analysis for the Behaviour Scale Theoretical Constructs NLSCY Data Users Guide The objective of the behaviour scale is to assess aspects of the behaviour of children two years of age and older Initially an attempt was made to measure the following behaviours for children aged 2 and 3 hyperactivity emotional disorder anxiety physical aggression inattention prosocial behaviour separation anxiety and opposition For children between 4 and 11 years of age an attempt was made to measure similar behaviours separation anxiety
139. or the purpose of constructing the factor After these exclusions the sub samples contained 1 352 and 1 398 individuals respectively for analysis purposes As a result of imposed factor analysis five factors were identified hyperactivity inattention prosocial behaviour emotional disorder anxiety physical aggression conduct disorder and indirect aggression The items that comprised each factor are described in the following table NLSCY Data Users Guide 112 2001 2002 BEHAVIOUR SCALE FOR 10 AND 11 YEARS OLD AD1CQ01J AD1CQ01R AD1CQ10Z AD1CQ10LL and AD1CQ01TT AD1CQ1F AD1CQ1K AD1CQ1Q AD1CQ1V AD1CQ1CC AD1CQIIlI AD1CQ1MM and ADiCQ1RR AD1CQ1G AD1CQ1X AD1CQ1AA AD1CQ1FF AD1CQ1JJ and AD1CQiNN AD1CQ1B AD1CQ11 AD1CQ1N AD1CQ1P AD1CQ1S AD1CQ1W AD1CQ1HH and AD1CQiQQ AD1CS05 AD1CQ1A AD1CQ1D AD1CQ1H AD1CQ1M AD1CQ1U AD1CQ1BB AD1CQ1GG AD1CQ100 AD1CQ1SS and AD1CQ1UU Conduct disorder and physical aggression Prosocial behaviour SESE EESE EE EEE EEE a eee eel E Cronbach s Cronbach s alpha coefficients raw values were calculated with Alpha for SAS using the normalized weighted data Please note that in Behaviour general Cronbach s alphas calculated with SAS are lower than Scale those produced by the SPSS software package Cronbach s alphas for these factors are given in the table below NLSCY Data Users Guide 113 2001 2002 CRONBACH S ALPHA VALUES FOR BEHAVIOUR SCALE 10 11 YEAR OLDS
140. ore is 12 This is a result of the fact that there are 12 items in the scale and four possible rescaled values 0 to 3 Many respondents had a rescaled score of 1 for every item in the scale and thus an overall score of 12 This means that the respondent answered agree to all of the items in the scale which were positive and disagree to all of the negative items as opposed to the more extreme answers of strongly agree or strongly disagree Basically this artifact in the scale score is due to the fact than many respondents were consistent in their answering pattern across items Weighted data could not be used since the weights developed for the NLSCY are for children only and not for families NLSCY Data Users Guide 103 2001 2002 Activities Activities Scale 10 13 Years BACCS6 Questionnair e Items Analysis of the NLSCY Data The object of the activities scale is to measure the child s participation in home responsibilities In Cycle 2 the factor scores were derived based on the factorial structure identified in Cycle 1 Below is a description of the items that were included on the questionnaire to measure activities the analysis used to construct the scale and the results of these analyses all from Cycle 1 In Cycle 1 questions ACCSQ6A ACCSQ6F were tested and questions ACCSQ6A ACCSQG6E were used to construct the scale Only Children aged 10 and 11 years answered these questions This set of question
141. ours smoking and drinking and activities 12 13 years friends and family school feelings and behaviours delinquent behaviour smoking drinking drug use health general depression and puberty and about work and sources of money 14 15 years friends and family school feelings and behaviours delinquent behaviour smoking drinking drug use health general depression and puberty and about work and sources of money work during the school year summer work sources of money and how they spent their money Interview For the household collection the interview length for responding NLSCY Length for households was approximately two hours Household Collection The total amount of time that it took to complete the major questionnaires that were part of the NLSCY household collection are presented in the table below The table gives median interview times i e the time at which 50 of the cases took more time and 50 took less It should be noted that all extreme times high and low were removed before these times were derived since they often represent a problem with the time clock procedure rather than a real interview time Table 5 outlines the length of time required to complete the various questionnaires Table 5 Type of Questionnaire Time in Minutes All questionnaires in the household interview All Child Questionnaires for the household 31 All Parent Questionnaires for the household for the PMK and spouse
142. overstate the error if at and Ay are positively correlated and understate the error if Ay and Ap are negatively correlated 165 2001 2002 In this case Rules 3 and 4 are combined The c v s for the two ratios are first determined using Rule 4 and then the c v of their difference is found using Rule 3 Rule 5 Estimates of Differences of Ratios Release guidelines applying to estimates also apply to confidence intervals For example if the estimate is marginal then the confidence interval is marginal and should be accompanied by a warning note to caution subsequent users about high levels of error Warning Note on Confidence Intervals NLSCY Data Users Guide 166 2001 2002 Examples of using C V Tables for Categorical Estimates Introduction The following are examples using actual NLSCY data to illustrate how to apply the foregoing rules Example 1 Using NLSCY data 84 085 babies were estimated to be of low Estimates of birth weight i e less than 2 500 grams How does the user Numbers determine the coefficient of variation of this estimate Possessing a Characteristic Aggregates 1 Refer to the c v table for children in 0 to 3 age group It should be noted that because the question on birth weight applied only to children in this age group this table should be used to determine the c v for this estimate 2 The estimated aggregate 84 085 does not appear in the left hand column the N
143. partner Total for major components Child Parent General amp PPVT amp Informed Consent NLSCY Data Users Guide 18 2001 2002 Remaining Components Tables 6 gives the median interview times for a family with PMK a spouse and 1 3 children The number of selected children 0 to 13 in the household was the factor that had the strongest impact on interview length Table 6 and Minutes 1 child 2 children 3 children Table 7 gives the median interview times for a family with a PMK no spouse and 1 3 children For households in which the PMK had a spouse partner and more than two selected children the interview length was over two hours Table 7 PMK spouse Time in and Minutes 1 child 2 children 3 children School Collection This is the difference between the total time and the time required for the major components This would include time for the interviewer to introduce the survey complete the household roster the relationships set up time for the 10 to 11 Questionnaire the 12 13 Questionnaire and the math and reading skills test time for the computer to generate the various questionnaires etc NLSCY Data Users Guide 19 2001 2002 The School The school collection took place from April to June 1998 For all Collection children in the Cycle 3 sample who were attending school the PMK was asked to give written permission to allow the collection of information from the child s teacher and pr
144. perament Scale NLSCY Data Users Guide 81 2001 2002 Introduction The Infant Characteristic s Questionnaire Questions Measuring Aspects of Temperament Children 3 to 5 Months Aspects of Temperament Children 1 to 3 Years NLSCY Data Users Guide Temperament scales are used to measure the temperament of young children up to and including the age of three based on the parents answers to questions about the degree of difficulty their child presents for them This measure is founded on the assumption that a child s temperament is not solely dependent on biological factors but is also influenced by the parents perception of the difficulty of the child The temperament scale used in the NLSCY for children 3 to 5 months old was developed by Dr John Bates of the University of Indiana This well established scale originally known as the Infant Characteristics Questionnaire ICQ has been used in large scale studies and is considered by specialists to be the best available measure for use in household surveys The ICQ has been adapted for use in other surveys covering different age groups 6 to 11 months 12 to 23 months and two year olds A revised version of the scale devised by Dr Jo Anne Finegan at Toronto s Hospital for Sick Children is used for three year olds For children aged 3 to 5 months the scale is made up of questions ATMCQ01 to ATMCQ12 ATMCQ14 to ATMCQ20 ATMCQ23 and ATMCQ33 is intended to measure the e
145. plain the living arrangements of the child and the socio economic conditions under which the child lives The content areas for each section of the various questionnaires used for the first cycle of the NLSCY are presented in the next section Cross sectional and Longitudinal Estimates NLSCY Design The NLSCY design and sample has been constructed so that it will be possible to produce both cross sectional and longitudinal estimates At present it is possible to obtain cross sectional estimates with Cycle 1 Cycle 2 and more recently with Cycle 3 data It is also possible to obtain longitudinal information from the longitudinal file The allocation of the Cycle 1 2 and 3 sample was such that it is possible to produce estimates at the national level for the specific age cohorts and at the provincial level for aggregated age groups This is true for cross sectional data as well as longitudinal data NLSCY Data Users Guide 53 2001 2002 Longitudinal Cohorts Attrition Augmenting the Sample There are two longitudinal cohorts those who were in the sample beginning with Cycle 1 aged 0 11 years at cycle 1 and those who were in the sample beginning in Cycle 2 0 1 year olds The Cycle 1 longitudinal sample is comprised of all children sampled for Cycle 1 of the survey in responding households excluding those from the integrated sample NPHS and the 3rd and 4th child of each family The plan is to follow these children over
146. pment MSD Scale Standardized Scores Standardized Scores from Previous Cycles The Motor and Social Development MSD Scale was developed by Dr Gail Poe of the U S National Center for Health Statistics The MSD scale consists of a set of 15 questions that measure dimensions of the motor social and cognitive development of young children from birth through 3 years the questions vary by age of the child Each item asks whether or not a child is able to perform a specific task The scale has been used in collections of the National Longitudinal Survey of Youth in the United States and in recent versions of the National Child Development Survey in England A score was calculated for each child by summing the number of yes answers to each item in the scale BMSCS01 Although there were different sets of questions depending on the age in months of the child differences were observed when comparing score within these age bands For example there was a specific set of questions for children 4 to 6 months old It was found that children who were 6 months old had scores that were on average higher than those 4 months olds Therefore a decision was made to produce standardized scores Each child was assigned a standard score such that the mean MSD score was 100 and the standard deviation was 15 for all age groupings of Cycle 2 and Cycle 3 This standardization had been done by 1 month age groups Therefore children who are 0 months old had
147. quent adjustment of salary and income level will also over the long term have a significant impact on the value of the means and standard deviations used to standardize the components of cross sectional SES The variable for cross sectional SES is labeled CINHDO8 The final derivation of longitudinal SES is based on the standards calculated for the first cycle of the survey The same raw values of the components helpful in deriving cross sectional SES are used but the standardization differs in this way Thus unlike cross sectional SES the standardization is not expected to produce a mean of zero anda standard deviation of one for each of the variables By definition the use of longitudinal SES is relevant only for analyses based on longitudinal children The initial standards of the first cycle which were used to derive longitudinal SES were created based on the characteristics specific to households having children aged 0 to 11 These same families in the third cycle of the survey have children aged 4 to 15 The value of longitudinal SES therefore allows us to calculate the net progression of each child in relation to the initial characteristics of his her household 2001 2002 Values of SES A child living in a household where the income has improved appreciably all things being equal will see the value of his her longitudinal SES improve as well However in the same circumstances the value of cross sectional SES may decline Th
148. r Items include BBECQ6D Q6U Q6BB Q6SS and Q6UU from the Montreal Longitudinal Survey the last four items are from a scale developed by K Weir and G Duveen Separation anxiety Items include BBEC6DD1 6LL1 6PP1 and Q6TT1 from Achenbach s Child Behavior Checklist CBCL Opposition Items include BBECQ6E1 Q6J1 Q6R1 and Q6T1 also drawn from Achenbach s CBCL NLSCY Data Users Guide 86 2001 2002 Children aged 4 to 11 Conduct disorder Items include BBECQ6C Q6E Q6G Q6L Q6O this item is coded not applicable for children not in school Q6T Q6AA Q6DD Q6FF Q6JJ and Q6PP from the Ontario Child Health Study OCHS Hyperactivity Items include BBECQ6B Q6l Q6N Q6P Q6S and Q6W from the OCHS and Q6HH from the Montreal Longitudinal Survey Emotional disorder Items include BBECQ6F Q6K Q6Q Q6V Q6CC Q6MM and Q6RR from the OCHS Anxiety Items include BBECQ6Y and Q6Il from the Montreal Longitudinal Survey along with several of the OCHS emotional disorder items BBECQ6F Q6Q Q6V and Q6CC Indirect aggression Items include BBECQ6J Q6R Q6Z Q6LL and Q6TT from Lagerspetz Bjorngvist and Peltonen of Finland Physical aggression Items include BBECQ6X from the Montreal Longitudinal Survey and BBECQ6G Q6AA and Q6NN from the OCHS Inattention Items include BBECQ6P from the OCHS and BBECQ6EE Q6KK and Q6QQ from the Montreal Longitudinal Survey Prosocial behaviour Items include BBECQ6A Q
149. r factor the observed score and the score he she would have obtained if questioned on all possible items in the universe the true score Since a can be interpreted as a correlation coefficient it ranges from 0 to 1 It has been shown that in general a is a lower bound to the reliability of a scale of n items Novick and Lewis 1967 In other words in most situations a provides a conservative estimate of a score s reliability Satisfactory Level of Reliability What is a satisfactory level of reliability It is difficult to specify a single level that should apply in all situations Some researchers believe that reliabilities should not be below 0 8 for widely used scales At that level correlations are affected very little by random measurement error At the same time it is often very costly in terms of time and money to obtain a higher reliability coefficient It should be noted that for some of the factors for which scores were computed for the NLSCY the reliabilities are below this level The Cronbach a is given in the documentation for each score that has been calculated Researchers can determine for themselves whether or not the score has adequate reliability for their specific purposes Finally it should be mentioned that for the NLSCY the Cronbach a for each factor score was computed using SAS Typically the a coefficients calculated using SAS are lower than those calculated using SPSS Parent Reported Scales Tem
150. r regression logistic regression and analysis of variance a method exists which can make the variances calculated by the standard packages more meaningful by incorporating the unequal probabilities of selection The method rescales the weights so that there is an average weight of 1 For example suppose that analysis of all male children is required The steps to rescale the weights are as follows Select all respondents from the file with SEX male variable CMMCQ02 Calculate the AVERAGE weight for these records by summing the original person weights BWTCWO1C from the microdata file for these records and then dividing by the number of records with SEX male For each of these records calculate a RESCALED weight equal to the original person weight divided by the AVERAGE weight Perform the analysis for these respondents using the RESCALED weight However because the stratification and clustering of the sample s design are still not taken into account the variances calculated in this way are likely to be under estimated 156 2001 2002 Calculation of Variance Estimates The calculation of truly meaningful variance estimates requires detailed knowledge of the design of the survey Such detail cannot be given in this microdata file because of confidentiality Variances that take the complete sample design into account can be calculated for many statistics by Statistics Canada ona cost recovery basis C V Rel
151. rade The other education variable included is current school status and whether attendance is full time or part time NLSCY Data Users Guide 120 2001 2002 Socio demographic Characteristics Objective Socio demographic Characteristics Suppression of Variables Questions on Mother Tongue and Language of Conversation Aggregated Variables for Language NLSCY Data Users Guide 121 The objective of the Socio demographic Section was to gather information on immigration ethnic background and the language profile of household members This will allow analysis for various components of the Canadian population and will permit identification of visible minorities As well there were questions on religious affiliation and frequency of attendance at religious services Religion particularly frequency of attendance is acknowledged as having a positive influence on a child s development It was necessary to suppress many of the variables in this section on the micro data file due to confidentiality concerns The questions on country of birth ethnicity and religion have all been suppressed while frequency of attendance at religious services has been included The questions on mother tongue and language of conversation are included on the micro data file but only with aggregated answer categories English only French only English and French only at least one other language indicated The aggregated variable
152. re Province Children Consent of Parent Consent of School Board Questionnaires Returned Going to School Freq Freq Cond Cond 97 5 94 0 97 3 73 2 95 2 70 7 91 5 52 2 94 9 59 5 93 2 64 2 92 6 67 4 95 4 64 7 92 9 7 56 0 94 1 91 3 60 9 The province of residence is missing for 4 children NLSCY Data Users Guide 144 2001 2002 Response The response rate for the Teacher s Questionnaire appeared to be Rates by lower in Quebec British Columbia and Ontario For Quebec and province Ontario this may be due to a lower rate of return by teachers In British Columbia s case while the questionnaire return rate was average the low response rate may have been caused by a lower school board consent rate Response Rates by school type Teacher s Questionnaire Type of School Children Going to Consent of Parent Consent of School Board Questionnaires Returned School 406 644 a i Private 660 59 Other Missing 76 66 Response The response rate for the Teacher s Questionnaire appeared to be Rates by nearly equal for public school students and Catholic school students school type but it appeared to be lower for children attending private schools This could be due to the lower percentage of consent given by private school boards Response Rates by age Teacher s Questionnaire Age Children Going Consent of Parent Consent of School Board to School
153. re determined the final score were calculated To produce the scores 1 was subtracted from each item so that the lowest possible score value would be 0 For the factors a high score indicates the presence Results There were 4 296 children in the sample for the age group 12 to 15 Cycle 3 years The group was split into two sub samples of 2 140 and 2 156 individuals and the analysis for this scale was performed separately for each sub sample The non response rate for the ten items ranged from 1 9 to 2 5 In total 310 cases who had one or more missing values and were excluded from the analysis These cases were given a missing value for the overall score since no imputation was completed The factor analysis revealed one strong factor conflict tactics CCRCS09 Items and J were not included in the factor as they reduced the Alpha Cronbach score The final score included items items A B C D E F G and H Items A and H were reversed in the calculation of the score All values were recoded from 1 5 to 0 4 The final score ranges from 0 28 with a high score indicating a higher degree of parent child disagreements The Alpha Cronbach value for the score is 0 75 Parent Child The parent child cohesion score was created for children aged 12 Cohesion 15 The following items were used in the factor analysis Scale CPRCBb31a CPRCBb31b CPRCBb31c CPRCBb31d CPRCBb31e CPRCBb31f CPRCBb31g and CPRCBb31h NLSCY Data Users Guide 99 2001 2002
154. register in 1998 Moreover suppose the design for the birth register sample is twice as effective as that for the 1994 sample In this case the correction factor for the longitudinal children would be i 2 30 0 6 while the correction factor for the birth register y 10 sample would be 0 4 Note that the sum of the two adjustment factors is 1 Correction Post stratification was carried out on the weights thus far to ensure that For Post the national and provincial estimates agreed with the January 1997 Stratification demographic estimates of the population of children aged 0 to 13 For Cycle 3 post stratification was done by province age group and sex This correction factor was derived for each post stratification as follows Demographic estimate Sum of weights in the post strata This correction ends the weighting process of the cross sectional sample for the second cycle of the NLSCY In this context a sample is more effective if its sampling variance is smaller than that of another sample of equal size selected using a different sampling design _ NLSCY Data Users Guide 52 2001 2002 Chapter 7 NLSCY Concepts and Definitions Introduction There are many variables and concepts which are critical to the analysis of the NLSCY data In this section there is a brief discussion regarding the types of possible analyses with the NLSCY data This is followed by a description of key variables which have been derived to ex
155. ren 4 3 were step children themselves CDMCD03 and 8 5 lived in a step family CDMCD15 Foster children and children living with only one parent are not included in step blended or intact families In the derivation of blended intact and step families if a child was the adoptive child of one parent and the biological child of the other parent then this child was treated like a step child and thus the family labelled as a step family In other Statistics Canada publications children of this type are treated as if they were biological children of both parents NLSCY Data Users Guide 60 2001 2002 Blended Blended families combine children who have different relationships Families with their parents A blended family consists of a married or common law couple living with at least two children one of whom does not share the same natural and or adoptive parents as the other child ren The following are examples of blended families lt a couple with biological children of the female partner as well as biological children of the male partner i e hers and his lt a couple with biological children of the female partner as well as children out of the new union i e hers and theirs The blended family is a sub set of the step family For the NLSCY children 6 2 were members of a blended family CDMCD14 Economic For the NLSCY an economic family is defined as all family members Family related by blood marriage common law relat
156. rging from the weighted analysis were used Factor Analysis Using Weighted Data Calculation of Scores and Item Imputation The results of the factor analyses were used to determine what items loaded into each factor i e were a part of each factor The next step was to calculate a score for each factor This was done by summing the values for each individual item that made up the factor In some cases some rescaling of values was done before the final score was calculated The following example illustrates how factor scores were computed Calculation of Scores for Each Factor NLSCY Data Users Guide 76 2001 2002 Example of Factor Score Computation One of the constructs that emerged in the factor analysis for the Parenting Scale on the Child s Questionnaire was the ineffective parenting factor In the factor analysis on cycle 1 data seven items were found to load into this factor APRCQ04 How often do you get annoyed with your child for saying or doing something he she is not supposed to APRCQ08 Of all the times you talk to your child about his her behaviour what proportion is praise APRCQO9 Of all the times you talk to your child about his her behaviour what proportion is disapproval APRCQ13 How often do you get angry when you punish your child APRCQ14 How often do you think the kind of punishment you give your child depends on your mood APRCQ15 How often do you feel you have problems managing your child in g
157. rmed to the logit distribution PINEOLOG SP the pineo occupation code for the spouse partner transformed to the logit distribution and HHINC household income in thousands of dollars Each of the five variables was standardized to have a mean of zero and a standard deviation of one NLSCY Data Users Guide 66 2001 2002 Consideration of Missing Data for the Derivation of Cross Sectional SES Final Derivation of Longitudinal SES Initial Standards of the First Cycle NLSCY Data Users Guide 67 In the case of cross sectional SES the components were standardized using the means and standard deviations of the variables for all households as observed in Cycle 3 Thus new standards were established based on the data for Cycle 3 families with selected children aged 0 to 15 Given the change in age of the selected children between Cycle 3 and Cycle 2 0 13 years it is expected that our sample allowing for the production of Cycle 3 standards consists of slightly older families This characteristic difference is of some importance as older families are generally expected to present more favourable socio economic characteristics than younger families From one cycle to the next this difference might not be felt but over the long term or over several cycles differences will likely be noticeable The income variable which is utilized to derive SES is expressed in current dollars Thus the cost of living increase and the subse
158. rview with the contacted household was discontinued if all the selected children had left but was continued if at least one of the selected children was still a member of the household NLSCY Data Users Guide 13 2001 2002 Contacting Non responding Households Demographic Information Collected Person Most Knowledgeable PMK New Cross sectional Households NLSCY Data Users Guide 14 In order to achieve the desired response rate an effort was made to recontact non responding households to the first collection in the second collection period and to recontact non responding households to the second collection period in the third collection period For example if in the first collection period a household could not be reached because no one was at home then this case was sent out again with the February sample and further attempts were made at that time to contact the household For households with eligible children basic demographic information was then gathered age date of birth sex marital status and relationships between the members of the household were completed Some questions about dwelling conditions were asked and this questionnaire ended with a question designed to select from among those individuals aged 15 or older the Person Most Knowledgeable PMK about the selected child ren This person became the primary respondent and was labeled as the PMK for this household In most cases the PMK was the mother
159. s The benefit of this adjusted weight is that an over estimation of the significance which is very sensitive to sample size is avoided while maintaining the same distributions as those obtained when using the population weight The disadvantage is that the numerator is not weighted up to the target population and the Coefficient of Variance Tables described in section 12 and presented in Appendix 3 are no longer useful as a measure of data quality Users should also note that some software packages may not allow the generation of estimates that exactly match those available from Statistics Canada because of their treatment of the weight field Definitions of Types of Estimates Categorical vs Quantitative Unit of Analysis Categorical Estimates NLSCY Data Users Guide The NLSCY file has been set up so that the child is the unit of analysis The weight that can be found on each record CWTCWO1C for the cross sectional sample and CWTCWO1L for the longitudinal sample is a child weight Estimates of parents or families cannot be made from the NLSCY microdata file Categorical estimates are estimates of the number or percentage of the surveyed population possessing certain characteristics or falling into some defined category An estimate of the number of persons possessing a certain characteristic may also be referred to as an estimate of an aggregate 152 2001 2002 Examples of Q Was the child born before after
160. s about responsibilities is from the Home Observation for Measurement of the Environment Short Form questionnaire in the National Longitudinal Survey of Youth Ohio State University To construct the Activities Scale for the NLSCY a factor analysis was conducted to test the theoretical construct In the factor analysis the items were multiplied by the child s normalized weight An individual s statistical weight is normalized by dividing his her weight AWTCW01 by the average weight of all individuals Consequently the sum of the normalized weights is equal to the sample size Once the factor structures were analysed and the items included in the factor was determined the score was calculated No imputation was done on the values If any values were missing the final score was set to missing A value may be missing if the child refused to answer or did not know the answer to the question To produce the score 1 was subtracted from each item so that the lowest score would be 0 The final score was derived by totaling the values of all items with non missing values The score ranges from 0 to 15 A score of 0 indicates the respondent does not participate in home responsibilities NLSCY Data Users Guide 104 2001 2002 Results In the sample there were 3 434 children aged 10 or 11 years They were divided into two sub samples of size 1 705 and 1 729 and an analysis was done on each sample The non response rates for the 5 items was 1 3
161. s consistency between the sum of the weights and demographic estimates from January 1995 for each combination of province age and sex As regards the sample of children selected in Cycle 2 the adjustment is made using demographic estimates from January 1997 NLSCY Data Users Guide 46 2001 2002 Weighting of As explained earlier the cross sectional sample is comprised of the Cross children selected in 1994 and children selected in 1996 In the sectional following paragraphs we present the correction factors which when Sample applied to the basic weights make it possible to calculate the weights of the cross sectional sample These correction factors differ according to whether the child was selected in 1994 or in 1996 First of all cross sectional weights were calculated separately for the children selected in 1994 and those selected in 1996 Thereafter each of these two components represents its respective target population However these target populations are not entirely separate It is therefore necessary to apply other correction factors to take this overlap into account The purpose of the last step post stratification is to ensure consistency between survey estimates and demographic estimates produced by Statistics Canada NLSCY Data Users Guide 47 2001 2002 Cross Sectional Weights for Children Surveyed for the First Time in Cycle 3 Children Selected from the Labour Force Survey LFS Sample 1 Correction
162. s for language of conversation are labeled CSDPDO05B CSDSDO5B and CSDCDO05B for the PMK Spouse partner and Child on the micro data file The mother tongue variables are CSDPDO6B CSDSDO6B and CSDCDO6B For the immigrant population a derived variable was created to indicate the number of years since first immigrating to Canada It was possible to put a grouped version of this derived variable on the micro data file CSDPD02B CSDSD02B CSDCD02B 2001 2002 Remote Access Since there are many variables in this section which have been suppressed for the micro data file researchers who are particularly interested in conducting analyses on socio demographic variables are encouraged to consider making use of the remote access service described in Section 13 3 Labour Force Parent Employment Stability Objective Labour Force Respondents and Employment Wages and Salaries NLSCY Data Users Guide 122 2001 2002 Employment stability impacts the home environment both in terms of income and stress levels Research conducted for the Ontario Child Health Study indicates that parental unemployment can adversely impact child mental health The Labour Force Section was completed for both the PMK and the spouse partner The main objective of the section was to determine employment stability as an indicator of the continuity of employment income Questions included periods of absence from work reason for the most recent abse
163. score would be 0 The final score was derived by totaling the values of all items with non missing values A score of 0 indicates the following for the two factors that were found to exist in the My Parents and Me scale a low degree of parental nurturance for the parental nurturance score a low degree of parental rejection for the parental rejection score and In the sample there were 2 258 children aged 12 or 13 years They were divided into two sub samples and analysis was done on each sub sample Individuals with missing values were excluded from the analysis conducted for the purpose of constructing the factor After these exclusions the sub samples contained 1 076 and 1 146 individuals respectively As a result of the factor analyses two factors were identified the parental nurturance factor and the parental rejection factor The items that comprised each factor are described in the following table NLSCY Data Users Guide 106 2001 2002 MY PARENTS AND ME SCALE FOR CHILDREN AGED 12 AND 13 YEARS OLD PARENT REPORT FACTOR t SCORE ITEMS pases asain a et ie a E SS aw NG E TEE aN ei ERE OTE T E Parental 1 BPRCbS07 BPRCQ29A BPRCQ29H BPRCQ291 BPRCQ29L Nurturance BPRCQ29N BPRCQ29R Parental BPRCbS08 BPRCQ29C BPRCQ29G BPRCQ29J BPRCQ29K Rejection BPRCQ29M BPRCQ29P BPRCQ29Q Cronbach s Cronbach s alpha coefficients raw values were calculated with Alpha For 12 SAS using the normalized weighted data
164. se tests are administered in the school to children from grades 2 to 10 During the household interview the parents consent is requested before the tests can be administered to the child at school 3 Since Cycle 2 a skills indicator of mathematics and reading abilities is administered at the home to pre assess their abilities The indicator consists of 10 to 13 questions with multiple choice answers the questions were taken from the second edition of the Canadian Achievement tests CAT 2 The CAT 2 is a series of tests to measure basic skills in a variety of subjects taught in schools 4 In Cycle 3 of the NLSCY a separate version of the mathematics and reading comprehension tests was administered for each academic grade level except for grades 9 and 10 which got the same level tests for a total of eight in all Thus students in Grade 2 completed the level 2 test students in Grade 3 the level 3 test and so on to level 9 for students in Grades 9 and 10 In some instances students were given a higher level test Fifty per cent of the children who scored a perfect score on the home administered skills indicator test were given a higher level school test than their actual level This approach was used to offset the potentially serious problem of ceiling effect encountered during Cycle 1 with the mathematics test especially in Grades 3 and 5 The mathematics and reading comprehension tests were administered from the same bookl
165. separate secure computer with no external access NLSCY Data Users Guide 11 2001 2002 Survey Specific The survey specific component of CAPI includes an introductory Components component with procedures for contact and selection of households Once contact has been made and household composition has been established the CAPI system generates applicable questionnaire components dependent on the household composition and the outcome of the selection procedures Some of the specific components generated included a Parent and General Questionnaire for the Person Most Knowledgeable PMK and spouse partner and Child s Questionnaire for selected children in each household The household roster becomes more difficult when a longitudinal survey interviews more than one longitudinal respondent per household Eligibility rules need to be defined to know when to trace and when to interview An added complexity was due to the fact that households of the Cycle 3 sample were divided into two groups Household Roster 1 longitudinal households that is those that had already participated in Cycle 1 and or 2 of the survey 2 new households with children aged 0 to 23 months 1 years of age or 5 years of age NLSCY Data Users Guide 12 2001 2002 Longitudinal Households 15 Contact The Final Phase of 1 Contact The Tracing File About 50 of all households contacted had already participated in either the first and or second cycle
166. ses Objective Questionnair e Items Questions BPRCQ29A to BPRCQ29R were taken from the Western Australia Child Health Survey The scale was developed by Lempers et al 1989 based on work of Schaefer 1965 and Roberts et al 1984 and measures parental nurturance rejection and monitoring NLSCY Data Users Guide 105 2001 2002 Analysis of the NLSCY Data Imputation for Missing Values Missing Values Results To construct the My Parents and Me Scale for the NLSCY a factor analysis was conducted to test the theoretical construct In the factor analysis the items were multiplied by the child s normalized weight An individual s statistical weight is normalized by dividing his her weight BWTCWO01C by the average weight of all individuals Consequently the sum of the normalized weights is equal to the sample size Once the factor structures were analysed and the items included in each the factor were determined the scores was calculated Imputation was done for missing values The imputed values were imputed using the SAS PRINQUAL procedure that determines which of the possible values for an item is the most plausible for an individual in view of his her response profile the response profiles of others in the sample and the number of factors included in the analysis If too many values were missing the final score was set to missing To produce the final scores 1 was subtracted from each item so that the lowest
167. spouse or common law partner residing in the household all spouse variables e g the Labour Force Section and the Education Section for the spouse have been set to not applicable NLSCY Data Users Guide 38 2001 2002 Not Stated In some cases as part of Head Office processing the answer to an item has been set to not stated The not stated code indicates that the answer to the question is unknown Not stated codes were assigned for three main reasons 1 As part of the CAI interview the interviewer was permitted to enter a refusal or don t know code as described above When this happened the CAI system was often programmed to skip out of this particular section of the questionnaire In the case of refusal it was assumed that the line of questioning was sensitive and it was likely that the respondent would not answer any more questions on this particular topic area In the case of a don t know it was assumed that the respondent was not well enough informed to answer further questions As part of the NLSCY processing system it was decided that all of these subsequent questions should be assigned a not stated code A not stated code means that the question was not asked to the respondent In some cases it is not even known if the question was applicable to the respondent 2 Insome cases a specific questionnaire was not started or it was started but ended prematurely For example there may have been some kind of an interruption or
168. te ets ee RSE na ene 72 ACCOR ALY SOS a Siac fit oleae cna san get A a E E E E a E S a Buca ect 74 Calculation of Scores and Item Imputation 0 cece cece eee 76 Reliability Measures for Scales 0 c cece eee cece tenes 80 Patent Reponed Scales wit ye ereat be cinuiuyss A Raita eres 81 Temperament Scale lt bs tees at enh She eA les te wl Nee he ae Gh ne 81 Ed ation Child serere reat n Sages enh hice e externa a Sine 83 BERAVIOUE Scale ni 45 eh cates can pend theists bie bund Si leds ond Sule NE EE 84 Motor and Social Development sis 2 65 ennics cdieadet ee pene ties ede ete 91 Relationships s airan he raene naaa e a ieee See Le eee Tae e shee oS 93 P renting ocd jce Wane Shank Stale dine Gule ne Siu elteduehG ng ea T 93 QUESTIONS swam E Kaa eernne we wen end tae wes otaG Na eae eee we meee 94 Analysis of NLSCY Data ences accede srra ranie Sees ante ceca meade eek 94 R su lts Cycle T sng ica Ole ints oe oe oe Ald cele eee neha Sy ea EE ieee Ste 95 Parenting Scales 12 15 Year Olds 2 s s 4 0 4 sacens dou ecs ods gece fees 98 Analysis of NLSCY Data cus orstes Sack ees ede Oise ete Meanie pane e and 99 Depression Scale RMI uixscwies new catia elmetne haes ahaa een een 100 PASSUIG ortho cb Atak Nedra eels Grebe Aunts oe ant aired seni TATE at Atk 101 Family Functioning Scale Parent 3 00 s05 c0 4e Setona vas oe uae ad osarms 102 ACUVINGS 4 7 nc ets ox a Dede Oe ree Doohiat sete blee SoU Get ates D
169. tem the response profile of other responding records in the sample as well as the number of factors considered in the analyses Imputation A flag was created for many of the items for which values have Flags been imputed to indicate the records for which imputation has taken place Where these exist the flags have been included on the micro data file The flag on the file which corresponds to an item has the same name as the item itself except that the Q question indicator in the variable name is replaced by I For example some imputation was carried out for APRCQ04 How often do you get annoyed with your child for saying or doing something he she is not supposed to The imputation flag for this item is labeled APRCI04 Raw Items It should be noted that in addition to the scores the raw items for each scale are included on the micro data file This will allow researchers to consider alternate factor structures if desired For the raw items the original values in the 1 to 5 range for the parenting scale have been retained before any rescaling or reversal of values took place NLSCY Data Users Guide 79 2001 2002 Reliability Measures for Scales Introduction Cronbach s Alpha Interpretation s of Cronbach s Alpha NLSCY Data Users Guide Reliability refers to the accuracy dependability consistency or repeatability of score results In more technical terms reliability refers to the degree to which the scores are fre
170. ter 10 Guidelines for Tabulation Analysis and Release 150 Rounding GUIGEINES nieo cue swear me whe emo cee one Sewanee eS 150 Sample Weighting Guidelines for Tabulation 00 cece eee ee 151 Definitions of Types of Estimates Categorical vs Quantitative 152 Tabulation of Categorical Estimates 0 cece eens 154 Tabulation of Quantitative Estimates 0 cece eee 155 Guidelines for Statistical Analysis 0000 ccc eee cence eens 155 C V Release Guidelines 25 4560ae Lawes os aa coir tn hexane axe ceeds 157 Chapter 11 Approximate Sampling Variability Tables 005 160 How to Use the C V Tables For Categorical Estimates 164 Examples of using C V Tables for Categorical Estimates 167 How to Use the C V Tables to Obtain Confidence Limits 170 Example of Using the C V Tables to Obtain Confidence Limits 171 How to Use the C V Tables to Do a T test 00 c cece eee 172 Example of Using C V Tables to do a T TeSt 2 0 cece eee eee eee 173 Coefficients of Variations for Quantitative Estimates 173 Release Cut offs for the NLSCY 0 000 cee eee eee eee eee eens 174 Chapter 1 Introduction General Information Purpose of this manual National Longitudinal Survey of Children and Youth Survey Population C
171. th SES had the following values Occupational Prestige 01 Farm labourer 02 Unskilled manual 03 Unskilled Clerical sales service 04 Semi skilled manual 05 Semi skilled clerical sales 06 Farmer 07 Skilled crafts and trade 08 Skilled clerical sales service 09 Foreman forewoman 10 Supervisor 11 Middle manager 12 Technician 13 Semi professional 14 High level management 15 Employed professional 16 Self employed professional 96 Not applicable this was assigned for the Spouse partner for cases where the PMK did not have a spouse partner 99 Not stated NLSCY Data Users Guide 64 2001 2002 This ordinal scale can be used to rank individuals into the various occupation groups but one cannot assume that the intervals between ranks are equal For example in this scale a middle manager code 11 is ranked higher than a supervisor code 10 which in turn ranked higher than a foreman code 09 However this does not imply that the difference in occupation between the middle manager and a supervisor is equivalent to the difference between a supervisor and a foreman By assuming that the underlying latent construct has a particular distribution one can assign intervals to the various categories Mosteller and Tukey 1977 propose a logit transformation to re express ordinal data on an interval scale To do this the percentage of individuals in each occupation group is considered a piece of the logistic distribution The cod
172. the total number of days as described above b estimate the number of children in this category by summing the final weights of all records for babies which were premature then c divide estimate a by estimate b Guidelines for Statistical Analysis Sample Design NLSCY Data Users Guide The NLSCY is based upon a complex sample design with stratification multiple stages of selection and unequal probabilities of selection of respondents Using data from such complex surveys presents problems to analysts because the survey design and the selection probabilities affect the estimation and variance calculation procedures that should be used In order for survey estimates and analyses to be free from bias the survey weights must be used 155 2001 2002 Variance Estimates Rescaling the Weights Example of Rescaling the Weights NLSCY Data Users Guide While many analysis procedures found in statistical packages allow weights to be used the meaning or definition of the weight in these procedures differ from that which is appropriate in a sample survey framework with the result that while in many cases the estimates produced by the packages are correct the variance estimates that are calculated are not adequate Variances for simple estimates such as totals proportions and ratios for qualitative variables are provided in the accompanying Sampling Variability Tables For other analysis techniques for example linea
173. time and revisit them every two years Analyses of these children will permit researchers the opportunity to perform in depth studies of the long term impact of risk factors such as divorce or the onset of a health condition and protective factors such as positive interactions with parents or academic success at school on these children as they move into adulthood If a child moves out of the household where he or she was sampled at Cycle 1 that child will be traced to wherever he or she resides during future cycles of the survey From a longitudinal perspective the child not the household is the statistical unit of analysis It should be noted that some children who were participants in Cycle 1 of the NLSCY did not participate in the second cycle or may not participate in subsequent cycles due to a variety of reasons This is usually referred to as attrition The number of these children is being carefully monitored and we are making every effort to keep these numbers at a minimum The Cycle 1 sample and its allocation were designed with this in mind and as long as future response rates are not lower than expected the sample will still permit longitudinal research by age cohort at the national level In the second and subsequent cycles it is intended that the NLSCY will add children belonging to age groups no longer covered in the longitudinal sample For example for Cycle 3 a panel of children 0 and 1 year of age was added to the Cycle 3
174. tion is praise is in the opposite direction compared to the other items In fact the item loaded negatively into the factor Therefore when computing the score the values for this item were reversed all the time was scored as 0 more than half the time as 1 and never as 4 In the documentation for each scale any item that was reversed for the scoring algorithm due to a negative loading is indicated Negative Loading NLSCY Data Users Guide 78 2001 2002 Non The score for the hostile ineffective parenting factor is labeled as Response APRCS04 on the record layout for the micro data file An S in the Code 5th position of the variable name indicates a score When the score was being calculated for each factor there was a possibility that one or more of the items making up the score had a non response code don t know refusal or not stated If the number of items with a non response code was above a certain threshold the factor score was set to not stated Generally this threshold value was set at 10 of the items If less than 10 of the items had a missing value then the items with non response codes were imputed before the score was computed The procedure used to impute these missing items is a routine available in SAS in the procedure called PRINQUAL This procedure indicates among valid item values the one that seems the most plausible for a given record It considers the response profile of the record with the missing i
175. tructure which emerged from the Scale parental scale for PMK depression was imposed on the 12 13 BHTCbS1B depression scale Score In order to produce the score 1 was subtracted from each item so Calculation that the lowest score would be 0 The final score was derived by totaling the values of all items with non missing values As well the answer categories were reversed for questions having a negative loading BHTCb11F 11H and 11J The total score BHTCbS1B may therefore vary between 0 and 36 a high score indicating the presence of depression symptoms Education Parent The Education Section was completed for both the PMK and Objective spouse partner The objective was to gather information on the years Education of school completed educational attainment and current attendance Parent at an educational institution Research for example the Ontario Child Health Study and the National Longitudinal Survey of Youth in the United States has indicated a link between maternal educational attainment the home environment and child development The questions on full time and part time school attendance provide an indicator of the main activities of the PMK and the spouse partner The variables CEDPDO2 for the PMK and CEDSD02 for the Values for spouse partner have the following values CEDPD0O2 and CEDSD02 less than secondary l secondary school graduation beyond high school college or university degree including t
176. tudinal weighting process first as it is the simpler of the two Furthermore this weight is used later to determine cross sectional weight Two steps are involved in obtaining the longitudinal weight for children selected in Cycles 1 and 2 These adjustment factors are applied to the basic weight in order to obtain the final longitudinal weight As concerns the sample of children selected in Cycle 1 the basic weight is the final weight before post stratification obtained in Cycle 1 With respect to the sample of children selected in Cycle 2 the basic weight is the final weight before post stratification obtained in Cycle 2 For further information on these weights please consult the documentation from previous cycles In step one an adjustment factor is calculated that accounts for the erosion non response observed since the sample was selected As regards the sample of children selected in Cycle 1 this factor corrects the erosion that affected this cohort in Cycles 2 and 3 With respect to the cohort selected in Cycle 2 the factor inflates the basic weight in order to mitigate the non response for this cohort observed in Cycle 3 These factors are determined by means of models NLSCY Data Users Guide 45 2001 2002 Regardless of whether they responded in Cycle 3 a considerable amount of information on these children was gathered during previous cycles The non response correction strategy makes use of this information It is based
177. two sub samples and an analysis was done on each sample Individuals with missing values were excluded from the analysis conducted for the purpose of constructing the factor After these exclusions the two sub samples contained 2509 and 2584 individuals respectively As a result of the factor analyses three factors were identified for the 10 15 year olds the parental nurturance factor the parental rejection factor and the parental monitoring factor The items that comprised each factor are described in the following table MY PARENTS AND ME SCALE FOR CHILDREN AGED 10 TO 15 YEARS OLD NLSCY Data Users Guide 116 2001 2002 Parental Nurturance Parental Monitoring Cronbach s Alpha for My Parents and Me I CPMCcS3 CPMCcQ1B CPMCcQ1F CPMCcQ1N CPMCcQ1E CPMCcQ1T Cronbach s alpha coefficients raw values were calculated with SAS using the normalized weighted data Please note that in general Cronbach s alphas calculated with SAS are lower than those produced by the SPSS software package Cronbach s alphas for these factors are given in the table below CRONBACH S ALPHA VALUES FOR MY PARENTS AND ME SCALE 10 15 YEAR OLDS s aloha aloha the most if excluded the item is excluded Parental Nurturance 0 88 CPMCQ1M 0 855 CPMCcS1 Parental Rejection 0 73 CPMCQ10 0 504 CPMCbS2B CPMCcQ1R 0 680 Parental Monitoring CPMCcS3 CPMCcQ1T 0 459 About me self complete 10 15 Objective About Me
178. uarantee that there is no bias induced by non response There remains a latent risk and we must remain watchful That is why there is considerable effort to minimize and study non response during both collection and processing All children covered by the NLSCY have been selected among households having already taken part in the Labour Force Survey This method of selection leads to three problems which might produce bias in our estimates The first problem stems from the fact that only respondents to the LFS have been considered for the NLSCY sample It could be that some of the LFS non respondents had children in the appropriate age group These households were not included in the NLSCY sample which could be a source of bias The second problem is due to the fact that only households having children when the LFS was conducted were included in the NLSCY sample It could be that some households were not included in the sample because the dwelling was vacant or their members were out of scope for the NLSCY at the time of the LFS Some of these households may have had children 0 to 13 living in them a few months later when the NLSCY interview took place Since these households were not eligible to be selected some bias may have been introduced NLSCY Data Users Guide 134 2001 2002 Third Problem The third and last problem complements the second In some cases the sampled address where a child was living at the time of selection
179. umerator of Percentage column so it is necessary to use the figure closest to it namely 85 000 3 The coefficient of variation for an estimated aggregate is found by referring to the first non asterisk entry on that row namely 7 3 4 The approximate coefficient of variation of the number of low birth weight babies is estimated to be 7 3 The finding that there were 84 085 babies that were of low birth weight is acceptable and no warning message is required to produce this estimate since the c v for the estimate is in the 0 0 to 16 5 range NLSCY Data Users Guide 167 2001 2002 Example 2 Estimates of Proportions or Percentages Possessing a Characteristic Example 3 Estimates of Differences Between Aggregates or Percentages NLSCY Data Users Guide Using NLSCY data it is estimated that 70 8 59 567 84 085 of low birth weight babies were born prematurely gestational age 258 days or less How does the user determine the coefficient of variation of this estimate 1 Refer to the c v table for children in 0 to 3 age group It should be noted that because the questions on birth weight and delivery time applied only to children in this age group this table should be used to determine the c v for this estimate 2 Because the estimate is a percentage which is based ona subset of the total population i e low birth weight babies who were born prematurely it is necessary to use both the percentage
180. utation and what specific items have been imputed on what records All imputation flags on the NLSCY data file have an l as the fifth character of the variable name For example the name of the imputation flag for household income CINHQO03 is CINHIO3 Derived Variables Combining Items Longitudinal derived variables A number of data items on the data file have been derived by combining items on the questionnaire in order to facilitate data analysis For example in the Labour Force section one of the question is on the Number of Weeks Worked but in the Adult Education section the question is Whether They Are Presently Going To School The combination of these two questions forms a variable that is based on the Actual Situation Of Work And Study Longitudinal derived variables were created to indicate changes between data reported in the current and previous cycles for family structure and PMK and Spouse changes NLSCY Data Users Guide 41 2001 2002 All derived variables on the NLSCY data file have a D as the fifth character of the variable name The name of the variable for the primary care arrangement is CLFPD51 Derived Variable Name NLSCY Data Users Guide 42 2001 2002 Chapter 6 Weighting Estimation Estimation The principle behind estimation in a probability sample such as the NLSCY is that each respondent in the sample represents several other persons in the general population For example gen
181. ve the property of interval or ratio data that is the distance between each answer category of the question should be the same For example in scales where the answer choices are Never Sometimes Often and Always one must assume that the distance between Never and Sometimes is the same as that between Sometimes and Often in the respondent s perception It was felt that this was not necessarily true in the case for the scales used in the NLSCY Therefore before performing the factor analysis for each of the NLSCY scales the data were transformed using optimal scaling The method used was one proposed by Young and several associates Young 1981 which is a variant of Fisher s optimal scaling technique The method is presented as a means of transforming data which are fundamentally nominal or ordinal in nature to interval or ratio level data so that statistical techniques which are appropriately applied only to interval and ratio data may be utilized NLSCY Data Users Guide 75 2001 2002 Initially the factor analysis for each scale to be included in the NLSCY data was carried out using unweighted data At that point in time the final weights had not yet been calculated Once the weights were available work started on repeating the factor analyses using the weighted data See Section 7 for a description of the weighting procedures With the weights the same factor structure was not always observed When there was a discrepancy results eme
182. ving the household the interviewer completed an administrative questionnaire describing the conditions in which the test was administered This is done in order to identify any factors that might have influenced the child s answers and overall reaction to the test Self The objective of the Self completed Questionnaire is to collect Completed information directly from the child on a variety of aspects of his her life Questionnaire These self completed questionnaires are used to supplement and in 10 15 yrs subsequent analyses compare with information obtained from the parent and teacher Starting at age 10 with the PMk s permission the interviewer provides a questionnaire to the child and encourages him her to complete it in a private setting Upon completion the questionnaire is sealed in an envelope to ensure confidentiality The PMK was informed of the confidentiality of the questionnaire before giving permission for the child to complete it The PMK is not allowed to see the completed questionnaire It was hoped that this procedure would increase the likelihood that the child would provide accurate and honest information The following table contains the content of the questionnaires completed by those 10 15 years of age Table 4 The Self completed Contains questions on the topics of Questionnaire for those aged NLSCY Data Users Guide 17 2001 2002 10 11 years friends and family school feelings and behavi
183. was no longer occupied by a family having in scope children at the time of collection In a way this is a frame undercoverage issue linked to the time lag between the LFS interview and the NLSCY interview This situation might occur when the selected occupants have moved before collection takes place As a result it is possible that the NLSCY sample undercovers the population of highly mobile children Component Non Response Component Non Response Responding Household As discussed in Section 5 there were several respondents or components to the NLSCY interview The PMK provided detailed information about each selected child In the Parent Questionnaire and the general questionnaire the PMK provided information about himself or herself and his or her soouse partner The PPVT R test was administered to children in the 4 to 5 age group Children in the 10 to 15 age group completed a questionnaire on their own For school aged children the teacher completed a questionnaire about the child and if the child was in grade 2 or above a Math Test was administered There was a potential for non response for each of these components It should be noted however that when a household was deemed to be a responding household then all required components were created for that household even if there were no data provided for a particular component For example if there was a 10 year old ina responding household who did not complete the 10 to
184. west score would be 0 The final score was derived by totaling the values of all items with non missing values The score ranges from 0 to 16 A score of 0 indicates the respondent does not have a lot of friends and does not have positive relations with other children In the sample in Cycle 1 there were 3 434 children aged 10 or 11 years They were divided into two sub samples of size 1 705 and 1 729 and analysis was done on each sample The non response rates for the 4 items ranged from 10 9 to 11 5 Individuals with missing values were excluded from the analysis conducted for the purpose of constructing the factor After these exclusions the sub samples contained 1 508 and 1 529 individuals respectively for analysis purposes No imputation took place As a result of factor analysis one factor was identified the friends factor AA1CS01 All items AA1CQ01 to AA1CQ04 loaded into the factor Results Cronbach s alpha coefficients raw values were calculated with SAS using the normalized weighted data Please note that in general Cronbach s alphas calculated with SAS are lower than those produced by the SPSS software package The Cronbach alpha for the friends score was 0 779 The item that affects the factor the most is AA1CQ04 If it were removed from the analysis the Cronbach s alpha would drop to 0 689 The final friends score could not be calculated for 397 11 6 individuals due to missing values for the items comprisin
185. xcluded After this criterion was applied there were 8 848 and 8 801 individuals left in the sub samples to be analysed No unreported values were imputed NLSCY Data Users Guide 97 2001 2002 Aversion A factor was derived for this age group aversion APRCS06 The Factor items making up this factor are APRCQ21 22 23 and 24 Items 21 Derived for and 23 were inverted when computing the scores The factor Children weights of variables APRCQ19 20 and 25 were insufficient to be Cronbach s alpha for this factor was 0 569 The item that had the greatest effect on this factor was APRCQ22 Removing it lowers Cronbach s alpha to 0 377 Normalized weighted data were used in the computations The score for this factor could not be computed in 478 cases because of unreported values Parenting Scales 12 15 Year Olds Conflict Tactics Scale The conflict tactics score was created for children aged 12 15 The following items were used in the factor analysis CPRCBb30a CPRCBb30b CPRCBb30c CPRCBb30d CPRCBb30e CPRCBb30f CPRCBb30g CPRCBb30h CPRCBb30i and CPRCBb30j NLSCY Data Users Guide 98 2001 2002 Analysis of NLSCY Data Factor The factor structure of this scale was determined based on data Structure from cycle3 To conduct the analysis on this scale for the NLSCY data a factor analysis was conducted splitting the data into two separate files Once the factor structure was analysed and the items included in the factor we
186. xtent to which the child is fussy unadaptable unpredictable and dull For children 6 to 11 months old the foregoing list was expanded to include ATMCQ13 and ATMCQ24 to ATMCQ27 The expanded list of questions measures the same four aspects of temperament as for children 3 to 5 months old For children between 1 and 3 years old questions ATMCQ1 to ATMCQ15 and ATMCQ17 to ATMCQ33 should theoretically measure the degree to which the child is difficult irregular unadaptable affectively negative and persistent unstoppable 82 2001 2002 Meanings of The respondent in most cases a parent is required to answer Ratings for each question in the scale by assigning a rating between 1 and 7 Specified For all questions except ATMCQ14 a 1 means that the child has a Behaviours favourable response or usually exhibits the specified behaviour while a 7 indicates that the child reacts negatively or seldom displays the behaviour in question If the child is in the middle a 4 is assigned In question ATMCQ14 the meanings of the ratings are reversed Education Child The objective of this section was to obtain some basic information about the child s educational experiences The amount and type of information collected varied depending upon the age of the child with more information being collected for the older children who have had greater school experience Introduction Basic information was collected for all age groups such as the c
187. ysis of passages identifying the main idea interpretation of various types of writing and critical evaluation Each test consists of two original English passages and two original French passages in order to make the test as linguistically equivalent as possible Response Rates for Math and Reading tests NLSCY Data Users Guide 137 2001 2002 Response Rates Among the 9 542 children eligible for the math and reading tests for Cycle 3 lt 86 of parents representing 8 206 children consented to having the school board contacted to administer the math and reading tests lt The school boards of 97 of the 8 206 children consented to administer the tests That meant that from the total number of 9 542 consent was obtained for 7 920 or 83 of all eligible children lt 65 of the tests that were administered were returned lt Due to an operational error approximately 2 of the tests were not sent to the schools To summarize out of the total of 9 542 eligible children we received 5 153 or 54 completed math and reading tests for Cycle 3 This rate is much lower than the 74 total that was obtained from Cycle 2 NLSCY Data Users Guide 138 2001 2002 Testing strategy In cycle 3 the evolution of the NLSCY testing strategy continued 2 The reading comprehension test introduced in Cycle 2 of the survey has seen few modifications while the mathematics test has evolved considerably from its inception in Cycle 1 The

Download Pdf Manuals

Related Search

Related Contents

la lecture en milieu rural : une pratique risquée  IVC24-04 User`s Manual Version 1.0  自 的 各研修で行われる実習及びデモンス トレ…ションで、 オ…トラジオ  TROUBLESHOOTING MANUAL  Department of Interior and Local Government  Dell OpenManage Server Administrator Version 6.2 SNMP Reference Guide  Fujitsu CELSIUS H920    PQA300 Picture Quality Analysis System User Manual  

Copyright © All rights reserved. Failed to retrieve file