Home

Collect SMART User Manual for Project Managers

1. at GOUNTRYRIDGEDR EET PERV TO FONLAR CT Ma he Fan ai H In order to collect data in the field as efficiently as possible when possible have interview teams consider parking and walking between households This works well in densely populated areas In rural areas this can be done as well by parking near clusters of homes Another strategy is to start at the furthest point from the base station and work back towards the base station While it might be tempting for an interview team to split up and knock on doors individually this is not recommended for safety reasons Implementing the survey Page 52 Special Sampling Situations If using Census data the cluster boundaries are typically formed by roads or natural features of the landscape like rivers or streams It is very important to only interview households within the selected clusters This is an important assumption of cluster sampling that must be adhered fo in order for the results to be accurately and validly generalized to the entire sample frame This is particularly important when clusters share borders Shown in the map to the right cluster 19 is north of the road and cluster 20 is just south of the road Interview teams must maintain awareness of where they are in the cluster and conduct interviews in the correct clusters Proper training and providing an overview map can help teams figure with this In a household survey it is important to
2. ccccsceesscseeeessceeenesseees 56 Ble OO SCO JORDENE a EEE 57 Day ol se 16 1100 JE SE SNE 57 ase E E E E 60 EO Next SURVEY Ten 60 CHAPTER 4 ANALYZING AND REPORTING RESULTS cssccsseccsseccssccessceeseces 62 MOC Oe EE E EE EE EA ENE 62 Analysis of Tracking FOrMS essessessesrresesrcesesrresssrteressterssstersssteressteressteresstereseeeresereres 62 Bole SONIC EE EE ESEE ET 63 Weighted FSIS serrreisssrereisresrrrisrerireit tens Ekti i EEEE EERTE i TEE TEE 65 Femme 68 APFENDIK os sisuna amaan ERROR BOOKMARK NOT DEFINED FO 1219 6 A EN 70 0 Oo ose st oh eeeh saeecceausiecanccovind esaseabevsie a EEEE 70 Questionnaire Design Resources rernvrnnnnrnnnnnrnrnnnnnrnrnnnnnnnrnnnnnnnnnnnnnnnrnnnnnnernennnseeeenne 70 MR USE Jr ysdaieodsdeanbuausaisensdtanen 70 el SOG SSE EE 71 OHH SOUS ama aa alal aa 71 Eee TS UN os ll uae 71 Appendix B Team ASSIGAMEME LIST arianismi ama 7 2 Appendix C Staff and Equipment Check Sheet esessssreressreresereressrerensreresererens Ja Appendix D Survey Press Release Template Community Health Assessment 74 Appendix E Community Flyer for Data Collection eessssessesseseesereereesesrereesesrereessss 75 Appendix F Base Station LAYOUT avsa 76 Appendix G Tracking Form Template wniseiacsetdosacericnsasavidctachstorsiqnsdeiadebueniasevedslascnieis 77 Appendix H Collect SMART Pocket Guide Google Android Tablet 78 Appendix l Team NEMNE 80 Appendix J Collect S
3. Do respondents interpret the questions the same way Do closed ended questions have a response option that applies to each respondent Are skip patterns followed correctly How long does the survey take Creating forms in CDC s Epi Info 7 If you are using Epi Info M you can create a new questionnaire from the main Epi Info menu Select Create Forms or select Tools gt Create Forms Epi Info 7 Menu File View Tools StatCalc Help Data AA i e ull enus 7 120 cars NUM INS Implementing the survey Page 20 1 The form designer window will open You can select File gt New Project to create a new project name 2 Tab to or select the Form Name field only use letters and numbers no spaces don t start a form name with a number 3 To create fields right click on the workspace to open the Field Definition dialog box or drag the field type from the list of options on the left 4 The canvas for the form you are creating will be displayed in the in the Form Designer window 5 With the form loaded right click on the canvas a pop up MENU will appear 6 Move the mouse over the New Field option 7 The list of field types that you can add to your form will appear Select a Field or Variable Type from the list i e Text 8 Then you type in the Question or Prompt for the field Click OK The field is created and displayed on the canvas 9 To add fields simply select the type of field you want
4. After you have downloaded the demographic data for your sampling frame the first step is to sort or rank your clusters from lowest to highest largest to smallest according to the number of housing units or total population Next create a new column titled Cumulative housing units and sum the housing units For example if your first cluster has 10 households the cumulative housing unit would be 10 Then add that 10 to the total from the next cluster If Cluster 2 had 13 households then the Cumulative housing unit for Cluster 2 would be 23 and so on Implementing the survey Page 28 Next determine the total number of housing units in your sampling frame In the example presented below it is assumed that you are conducting a 30 by 7 sample The next step is to generate 30 random numbers between 1 and the total sum of housing units You can use a random number generator within your spreadsheet software or one online such as http www random org integers Finally match up the random numbers to the Clusters using the Cumulative of housing units Below is an example to clarify the process Stage 1 Example Cluster Population HU CUM HU CUM Random Range Numbers 1 0 10 2 25 10 11 20 3 21 10 21 30 36 10 41 40 5 7 11 41 51 6 29 11 52 02 i 3 74 Oe a n om s 87 98 imi 55408 a In this example the column HU CUM signifies our cumulative housing unit total and HU CUM Range shows the range Th
5. http wwwn cdc gov epiinfo script eihat aspx Epi Into M User Guide http wwwn cdc gov epiinfo user guide index him Questionnaire Design Resources NC Division of Public Health s Community Health Assessment Guidebook includes Community Health Opinion Survey templates in English and Spanish as well as an Epi Info template http publichealth nc gov Ihd cha Behavioral Risk Factor Surveillance Survey optional modules on specific topics http www cdc gov brtss questionnaires National Health Interview Survey has questions that have been tested and validated http www cdc gov nchs nhis quest data related 1997 forward htm PubMed search to see if there are questions that have already been tested http www ncbi nim nih gov pubmed Sampling Resources A Guide to Sampling for Community Health Assessments and Other Projects http cphp sph unc edu PHRSTS IntroSampling pdf Two Stage Cluster Sampling General Guidance for Use in Public Health Assessments http cehp sph unc edu PHRSTS TwoStageSampling pdf Sample size calculator openEpi com IRS requirements for health assessments conducted by hospitals http www irs gov Charities amp Non Profits Charitable Organizations New Requirements for 501 28c 29 283 29 Hospitals Under the Affordable Care Act Pre joined demographic data for Census Blocks via TIGER Line Shapefiles http www census gov geo maps data data tiger data html Census Block Groups tabular data from
6. need to calculate response rates and compare your sample demographics to the geographic area as a whole The data to calculate your response rates should be on the tracking forms that your field teams have so it is vitally important that your volunteers understand how those forms are to be used and then they have to use them More detail regarding the specifics of preparation of the data for use will be provided in the Analysis section of this manual including comparing your sample demographics to the sampling frame demographics Often in door to door surveys the elderly and women are oversampled As the project manager you can make small adjustments to your methods fo try to avoid this bias For example if you have already surveyed 100 residents and 70 of them are women let your survey teams know that if a male answers the door they should try even harder to convince him to participate citing the vital importance of both female and male opinions Again coercion is discouraged but taking extra time to explain the importance of diverse community participation should be encouraged and emphasized Similarly adjustments should be made if you notice on the tracking forms that certain teams are getting a lot of refusals As discussed earlier talk fo the team members about their challenges and split them up if needed or send a team that has had good success to their possibly stubborn neighborhood Completion Rate The completion rate describes how clo
7. answer You might also find that resoondents will get distracted and want fo tell you a story they think is somewhat related to a question Be polite and hear them out However if this happens repeatedly you will need to gently guide them back to the survey Sir don t want to take up anymore of your time than necessary today If it s ok with you I would like to ask you the next question on Or if they still offer up stories you can say l am sorry but I have to ask you this next question so I can get back out there and talk to your neighbors about their concerns too e Avoid grilling participants f resoondents falter or are slow to respond offer to move on Do not cross examine or grill the participant Remember this is an opinion survey and there are no right answers This isn t a test of the respondent s knowledge If they keep saying oh don t know remind them that there are no right answers and we just want their opinion However don t press them for an answer if they don t want to give you one Implementing the survey Page 39 Probing A common practice in interviewing Is fo probe a respondent for additional information It is important to recognize that probing should also be done in a standardized way so not to influence the response A probe is generally used when a respondent s answer is unclear or irrelevant Examples of Responses Requiring a Probe Interviewer In the past 12 months did yo
8. bilingual Now that your team is place you are ready to develop your questionnaire and select your sample Let s start with the questionnaire first Questionnaire Design Well developed questions on a thoughtfully designed questionnaire are critical to the success of collecting data and will allow programs to ask and answer questions about the health priorities of a community Time spent during the questionnaire development phase will lead to higher data quality and easier data analysis There are five components of designing a questionnaire 1 Work with stakeholders to create questions addressing the topic of interest 2 Continue to prioritize and revise questions as necessary with input from stakeholders 3 Keep the questionnaire to 50 questions or less so that it can be completed in under 20 minutes 4 If necessary have the finalized questionnaire translated into other languages 5 Pilot test the questionnaire and revise or edit as necessary Although there is often interest in many types of questions focus on those that cannot be answered from other data sources For example for a community health needs assessment you might consider including questions about community concerns services needing improvement health care access and health behavior A pre disaster questionnaire may focus exclusively on emergency preparedness and special medical needs In addition to the questions the questionnaire should include opening and closing
9. coding errors Another problem is unintelligible text entries This is not uncommon and can happen when interviewers use abbreviations or when they enter typos into the text box field In these cases you will have to decide how to recode the data and document any subsequent changes Missing Data There are very few datasets that are 100 complete Usually there are at least a few missing values and in the worst case there might be so much missing data that a proper analysis is not possible With smaller datasets the easiest way to identify missing data is by looking at the actual dataset You can also examine frequencies and many statistical software programs have features that will help you identify missing data which is necessary when working with larger datasets Analyzing and reporting results Page 63 Sometimes missing data occurs randomly and sometimes it can occur in patterns e g you may find that a number of people are all missing data on the same variable When there are patterns to missing data this may provide clues to why it is missing e g one interviewer may not be appropriately entering data and Is repeating this error for all of the interviews that they conduct When there is a large amount of missing data for a single subject this may indicate that the interview or questionnaire was terminated early However sometimes there may be pieces of data that are intentionally missing such as when a questionnaire includes
10. disaster on the population is unknown when the health status and basic needs of the affected population are unknown or when the response and recovery efforts need to be evaluated l Malilay J Flanders WD Brogan D A modified cluster sampling method for post disaster rapid assessment of needs Bull World Health Organ 1996 74 4 399 405 Frerichs RR Shaheen MA Small community based surveys Annual Rev Public Health 22 231 47 2001 Table of Contents Page 9 CASPER can also be used before a disaster occurs to assess community preparedness including personal preparedness hazard mitigation knowledge and intent to evacuate CASPER Is also well suited to routine health data collection Health departments and hospitals offen must conduct periodic community health needs assessments The same CASPER methods that apply after a disaster can be used for these assessments with the added benefit of having more time fo plan Besides community health needs assessments CASPER can be used for other surveys that gauge community opinion and knowledge on a variety of topics In 2010 CASPER was used to determine levels of potassium iodide KI distribution and resident education in a 10 mile radius around a nuclear power plant in North Carolina In another community in North Carolina CASPER was used to gauge public opinion on how to reduce teen pregnancy and to assess how much of a problem teen pregnancy was in the community ee aA e e e Lyd Onlin
11. from 600 to 3 000 people Blocks are the smallest geographical unit used by the U S Census Bureau On average there are about 39 blocks per block group Below is an example of 30 clusters determined by Census Blocks within a North Carolina County js Stage 1 Hy eee Community Health Opinion Moniga County NC If you decide to use U S Census Data it is recommend that your PSU be either the block or block group As you consider the choices think about the following factors Implementing the survey Page 27 ae Small easy to navigate Small not many households Field maps easier to make and read Blocks Large lots of eligible Large more difficult to participants navigate American Community Often not possible in rural Block Groups Survey updated yearly areas with less than 30 block groups in a county Population estimates only available every 10 years The CDC recommends the use of blocks but this is often problematic in areas where there are lots of blocks with fewer than 10 households The main advantage of using blocks Is that they are small and easy to navigate You can often park your car in one area and walk the entire block If your sample frame contains over 60 block groups you might consider using these as your sampling unit particularly if you find that the majority of the blocks have fewer than 10 households Block Groups typically contain between 600 and 3 000 people so
12. household Are you at least 18 years old START INTERVIEW Latitude Longitude Date Time You will return to the Record List screen If you want to view the Map screen select the Back button Create a Sync File Mid point and at the end of the day create a sync file which will upload your locally saved files to the server creates a back up copy and allows the project manager to see the survey files From the Record List tap on the Menu Icon a at the upper right hand corner From the Menu Icon select Create Sync File Record ID 1 Geate Sync File ClusterSurvey DateTimd Generate Map Preview 20 3 Analyze Data Record ID 2 ClusterSurvey DateTime Latitude Preview 66 7 35 9102 Record ID 3 ClusterSurvey DateTime Latitude Preview 45 66 9 11 2013 1 26 57 PM 35 9103 Record ID 4 ClusterSurvey DateTime l Latitude Preview 23 2 9 11 2013 12 02 08 PM 35 91 Record ID 5 A ClusterSurvey DateTimel Latitude Preview 9 11 2013 12 13 31 PM 35 9102 Record ID 6 E ClusterSurvey DateTime Latitude Preview 33 55 9 11 2013 12 23 17 PM 35 9103 Click OK once the survey files have been transferred to the server HELP From the Map View tap on the Menu Icon at the upper right hand corner to access the Epilnfo Help Screen Map View New Hanover Survey A GT jelp Skippers A info Settings My Location Refresh Points Still have questions Call Matt Simon at 919 928 4477 Appe
13. if necessary e Revisit common problems or issues teams were having e g make sure to charge mobile device tablet switch to paper surveys if there are any technical issues tracking forms reminders etc e If time allows ask for feedback or for people to share their experiences from the previous survey day Implementing the survey Page 59 Check In l Collect all team equipment including tablet tablet chargers team packets GPS units mounts chargers if using Track check in of equipment in Staff and Equipment Check Sheet Verify with staff or volunteers who is committed to returning the next day When you receive a team s mobile device tablet open up Collect SMART o Sync the files in case the team forgot to sync after the last survey Ask teams to walk you through tracking forms as they explain be sure to check o Tracking forms have cluster number written on top o All addresses have been entered and there are no missing status codes teams often forget to enter in the last address of the day o The number of interviews they have completed matches what is on the tracking forms On the Collect SMART website open the Manage Project page and choose View Survey Files o Open the latest xml file synced by the survey team o Make sure that the synced file is on the server o Verify the number of surveys complete in the synced file list Before Next Survey Day l 2 Fully charge mobile devices tablets Check s
14. no coverage at times Texting could possibly be available in areas with low coverage but not guaranteed At a minimum interviewer teams should contact base station after they complete the first interview and on their way back to the base station after the final survey for that day is complete Remind team members to check their cell phone or radio often Base station staff may try to contact teams and leave messages as necessary Ensure that the interviewers equipment especially the GPS unit tablet is kept secure and in their possession at all times Interviewers should avoid leaving equipment in the car or on the dashboard when taking lunch or breaks Since personal data Is being collected it is very important that the GPS unit tablet is locked when not in use or if paper surveys are being used that the survey copies are in the interviewers possession at all times If interviewers recognize houses or persons being interviewed remind them to maintain the confidentiality of resoonses and not share any information Personal Safety The following are personal safety tips that can be used while training interviewers e In remote areas be aware that public restrooms may not be nearby Never use the restroom in a residence but instead use restrooms in town areas or a public place such as a gas station or fast food establishment e When approaching a home wait and observe the area before getting out of the car Look at what s going on
15. or address points using the built in GPS 4 Digitally collect survey responses 5 Upload responses to a central server 6 Track team progress and locations 7 Conduct real time analysis of results The Collect SMART tools will immediately provide the following innovations 1 Electronic surveys reduce data entry time and errors common with paper forms 2 Cloud based database for survey responses for added data security 3 Real time survey frequency analysis using a web based project management dashboard 4 Integration with mobile Epi Info to create and share questionnaires and analyze results Materials and resources associated with Collect SMART include MATERIALS RESOURCES Collect SMART software suite CASPER Online introductory module http www collectsmartdata com http tinyurl com C ASPERTrain Community Health Opinion Just in Epi Info software download Time Training http wwwn cdc gov epiinfo html downloads htm http tinyur com CHOSJITTrain Epi Info Community Health Assessment Tutorial http www cdc gov epiinfo script eihat aspx Table of Contents Page 8 Introduction to CASPER The Community Assessment for Public Health Emergency Response CASPER is an epidemiologic sampling method that enables public health and emergency management professionals to rapidly collect data in order to determine the health status and basic needs of affected communities during all ohases of a disaster response The CASPER
16. put tablet in Airplane mode to save battery The surveys need to have the Cluster and Survey number entered on the form and need to be saved The surveys will be saved on the device and will need to be pulled off later Teams can use paper surveys as a back up I saved my record and it took me out of the interview form Saving the record will automatically take you to the Record List You can scroll down to the most recent interview and tap on the interview to get back In The Cluster view shows that have completed all the clusters in my survey but have not completed them all accidentally opened one How to I begin a new interview in that cluster You can still begin an interview you can restart a point that is saying it s already complete or from the list at bottom hard press a point until it opens up an interview form We are at starting point 22 3 but the tracking form says 22 1 The starting points are just suggested starting sites for you to find homes to do the interview Remember the cluster itself is a randomly selected cluster so any home within that cluster is fair game The tracking form is just tracking homes you are going to and the cluster _ interview _ that you are working on The most important thing Is to track the addresses of where you are going NO IR IC how many interviews you are completing in the cluster and use the random starting points as a suggested place fo start Appendix Page 81
17. record after completing the interview Once the record is completed it will show up in the record list and then interviewers will need to send a sync file over to the Collect SMART cloud based server This sync file is an encrypted version of the data ep7 that for import in the desktop version of Epi Info Note that an xml version of the data is also sent to the server There is a help screen with some information about the app and the sampling method that can accessed from the Map View screen Also a pocket guide guiding users through the Collect SMART mobile application is included in Appendix H Monitoring Teams and Analyzing Results in Real time Once teams are in the field the project manager can monitor team progress using the Project View dashboard on the Collect SMART website This dashboard displays the point status in three different ways amp Green dots incomplete interviews X Gray x s completed interviews amp Yellow dots interviews in progress A Red triangles Survey teams Both points and surveyors can quickly be located using the drop down menu in the upper left part of the screen There are several management buttons at the top of the screen e Survey or Team dropdown To view specific teams or survey points select from the dropdown menu e Hide Clusters Cluster boundaries and labels can be turned on and off and point status and surveyor location can be refreshed manually using the buttons at the top
18. sample such as at a health fair or other central location like a shopping center would necessitate the collection of at least twice as many surveys and it still would not guarantee that the sample could be generalizable to the entire county More participants do not always lead to better survey results In North Carolina the two stage cluster method is an accepted sampling strategy for conducting surveys for Community Health Assessments For additional information see A Guide to Sampling for Community Health Assessments and Other Projects and Two Stage Cluster Sampling General Guidance for Use in Public Health Assessments In Stage 1 you randomly select 30 census blocks with a probability proportional to size within the sample frame Size is typically the number of housing units or the number of occupied housing units within the cluster Then for Stage 2 you randomly select 7 households within each cluster to conduct an interview at The gold standard would be to return to that randomly selected household as many times as it takes until the interview is completed or refused However this approach can be very expensive and time consuming and it is known that some replacement of selected households will have to occur in order to complete the data collection in a timely manner If no one is home at the randomly selected household or the respondent refusing to be interviewed then teams should be instructed to find the next closet house until
19. scripts skip patterns and appropriate transitions Implementing the survey Page 16 When adding new questions to your survey try to use tested or validated questions when possible First the NC Division of Public Health s Community Health Assessment Guidebook includes Community Health Opinion Survey templates in English and Spanish as well as an Epi Info template There are additional resources out there including the Behavioral Risk Factor Surveillance Survey particularly their optional modules on specific topics and the National Health Interview Survey that have questions that have been tested and validated You can also conduct a PubMed search to see If there are questions that have already been tested For example your program may want to ask questions about health literacy Instead of creating new untested questions identify studies that developed questions and survey instruments through a PubMed search A key point in questionnaire design is to limit the number of questions to avoid survey fatigue and also to make analysis and interpretation of those results easier Working with your community partners It might be tempting to add all of the questions that interest each partner However you need fo first ask before you add Consider if the data are available through another source for that geographic area For door to door surveys the goal is to keep the survey length to about 15 20 minutes The average length of a survey with
20. small sub area which may not be typical of the population as a whole Moreover surveys should avoid sampling only the most accessible members of the affected population For example try not to over sample households along roads near markets or in the center of town The decision about which areas to survey depends on a number of other factors including e Number of survey team staff or volunteers available e Amount of time and the funds available for the survey e Availability of transportation options are survey staff or volunteers able to use personal or work vehicles to reach the survey areas The sampling frame for health departments is typically the county or if you are part of a regional health department multiple counties If you are conducting an assessment in multiple counties you could conduct one sample in each county or you could conduct one assessment for the entire region The CDC recommends that a sampling frame should include at least 800 housing units If there are fewer housing units consider conducting a full census or another sampling methodology such as a simple random sample Below is an example of an assessment for a regional health district in Northeastern North Carolina that included 7 counties The sample size calculator at openEpi com was used to determine that 80 surveys per county would be sufficient to assess the needs of the population Interviews were conducted with 4 residents in each of 20 clusters selecte
21. that interview then go to the next starting point location We ve knocked on the doors in all of our clusters and still need interviews Suggest going to another assigned cluster and going back to the NOs in the first cluster later in the day The map on Collect SMART app is not showing our location we don t have data signal On mobile device verify that the cellular data is not turned off Also check to make sure it is not in Airplane mode Check and see if there is any data signal If you can still get into Collect SMART you will still be able to open cluster points just not see where you are on the map The surveys need to have the Cluster and Survey number entered on the form and need to be saved The surveys will be saved on the device and will need to be pulled off later Teams can use paper surveys as a back up Once you are in a cellular coverage area click on the OPTIONS 3 dots button in the Upper right corner to refresh points Collect SMART won t log us in saying check connection If you are in an area with no connectivity hitting the back button can log you out of Collect SMART and you can t get back in until you are in an area of connectivity On mobile device verify that the cellular data is not turned off Also check to make sure it Is not in Airplane mode Check and see if there is any data signal If Collect SMART does not work you can instead go into the Collect window of Epi Info and collect surveys offline you can
22. the US Census Bureau s American Fact Finder system http tacttinder census gov taces nav jst oages searchresults xhtml retresh t Appendix Page 70 Then download the spatial data from TIGER Line Shapefiles http www census gov cgi bin geo shapefiles2010 main Random number generator http www random org integers Staffing Resources Incident Command System Field Guide from Texas Department of State Health Services CASPER Strike Team http www dshs state tx us commprep disasterepi casper aspx Training Resources Interviewer training template http nciph sph unc edu training programs collectsmart Interviewer 20 Training Template pptx Reporting Resources Sample Community Health Needs Assessment from Wake County NC Full Report Appendices D F pertain to the survey htto www wakegov com humanservices data Documents 2013 20Wake Z2ZO0CHNA 20Full 20Report 20FINAL pdf Sample from the 2011 Durham County Community Health Needs Assessment Summary Presentation http www healthydurham org docs CHA 20201 1 20 20key 20findings pdf Sample from the 2011 Durham County Community Health Needs Assessment Appendix pertaining to survey analysis and results http www healthydurham org docs Appendix 20G 20 20Survey pdf Appendix Page 71 Appendix B Team Assignment List Tobe TEAM ASSIGNMENT LIST Day 1 Enter date CLUSTERS SAME er cel DRIVER p Needed T 7 For clusters you may want to note her
23. the respondent refuses to be interviewed surveyors should knock on the door of the next closet house until an interview is conducted Once an interview is completed move onto to the next random start site If resources allow this is the preferred method for implementing the second stage of the sample Implementing the survey Page 34 e Survey Start Sites PS 7 Selected Clusters 1 40 Fe Building Type 20 gt gt Residential x Commercial Other Mobile Home Streets Now that you have your team your questionnaire is designed and you now where you are going to conduct interviews it s time to recruit surveyors Recruiting Volunteers For a typical two stage 30 7 sample 210 interviews plan for 10 teams of two to work over 3 full days you will need at least 20 volunteers or staff each day Typically teams can complete 5 14 interviews per day This can vary depending on your survey instrument the distance teams have fo travel as well as surveyor experience competency and confidence The ideal time to be in the field is Thursday through Saturday from 10am until dinnertime before dark For safety reasons it is recommended that 2 people should be assigned to each interview team so that no one is ever sent out to conduct interviews alone The best candidates would be those with prior experience with door to door activities such as political canvassers fundraisers nurses who make home visits You are looking for individuals w
24. the survey e A brief review of the sampling method so interviewers know how houses were selected e How to use the questionnaire including the intent and meaning of each question and how to record responses and ways to address resoondents questions e Methods for improving the response rate ee ee i ee e How fo track completed surveys Ne Training Template pptx e Confidentiality Interviewer Training e Safety ps i Se ee ee a ee ee a ee er Implementing the survey Page 36 Materials eguipment needed for training e AV laptop and projector If possible it would help to have a software on your laptop e g http droid at screen org that enables you to share your tablet screen so you can demonstrate and project Collect SMART for the trainees e Training packets for each interviewer provide at a minimum Agenda Team Safety Checklist Copy of training slides Questionnaire in color to show probes and instructions Sample Tracking Form e Mobile devices tablets with Collect SMART software installed for allowing interviewers to practice using technology e Sign in sheet to track who attended the training e Training evaluation if applicable The overall goal is to train the interviewer to overcome unwillingness and increase receptiveness among participants in the field What one does or says as an interviewer can make or break a survey Below are the various factors that are involved in motivating s
25. the timeframe rather than assuming the respondent s frame of reference By narrowing down the answers for the respondent the interviewer is leading to a response Feedback Another important area to train interviewers on to maintain rapport with the respondent is providing feedback during the interview Inexperienced interviewers may use feedback inappropriately Interviewers should give feedback only for acceptable performance not good content For example if the respondent is answering questions about smoking and tells the interviewer that he quit Implementing the survey Page 40 a few months ago the interviewer may be tempted to provide a bit of encouragement That is exactly what should be avoided Examples of appropriate feedback include e see e Uh huh e Thank you e That is useful information e That is helpful to know e That is useful for our research e Let me get that down e want to make sure have that right and repeat the answer given and e We have touched on this before but I need to ask every question in the order that it appears in the questionnaire Finishing the Interview Lastly it is important to train interviewers to maintain confidentiality throughout the process Participants should be reminded that interviewers are not collecting identifying information names etc All survey information will be compiled with other responses from throughout the survey area and their identity will remain an
26. their equipment mobile device GPS unit batteries maps clipboards pens pencils questionnaires and leftover incentives You will need to sort through everything to make sure the paperwork Is in order before the teams go back out the next day Implementing the survey Page 43 Materials and Staging As soon as the interview teams are established and the training is prepared the next step is to plan the logistics of survey materials and staging Team Packets Listed below are items to include in survey team packets Three ring binders work well for ease of use l A list of contact numbers including individual cell phone numbers for all the team members as well as staff in the base station This allows other team members to get in touch with each other to help finish up survey clusters A copy of the training materials slides from a presentation offer a nice summary for reference in the field At least one paper copy of the questionnaire for the interviewer to read the survey from as the other team member enters data on the mobile device Itis also helpful to include a handful of paper copies 5 7 of the questionnaire for each team as back up in case there is a problem with the technology Set of tracking forms There should be 2 blank tracking forms per cluster in each of the team packets These will be discussed in greater detail on the following page A tracking form template can be found in Appendix G Field maps
27. to add give it a prompt and the step Is complete The steps above outlined how to create a text field Other field types are also available including number fields which restrict the user to entering just valid numbers date fields checkboxes and drop down lists You can also create code books in Epi Info For more guidance on how to create questionnaires in Epi Info see the Epi Info User Guide from the CDC s website or watch the demonstration videos below e igner tee pi Info pen enebas ab Text tesla ee ANNE bi Unique Identifier pI New Project 5 Open Project 5 Close Project Undo Redo F Check Code E Enter Data FF Durham CHOSCeuntyPage 2 A Label Title 41 Duiham CHOS County 3 2 DuthemCHOSCourty les Text Uppercase labi Multiline i Number ati j 121 Phone Number att Thes sbons are uru ja issue est i eral att Unique kenther lity of life in Durham County Please loo is li Te p to A Date ni Number ELLOW t community eeping in mind yourself and the people m your neighborhood Pt Mart ick the community issu hi he greatest effect on quality of life in Durham County Please 3 uf hoose 3 Remember thi f opumon and your choices will no inked to you m any way Time Ti One u do community blem you consider one of the most important please let me know I can 4 z Ime iso read th t loud as you think about them Read community i if they pref have them EE Date Time Date
28. to fill Interview Team Lead roles so teams might be reporting directly to the Strike Team Leader e Strike Team Leader is responsible for overall survey collection organizing and directing the base station and overseeing the tactical assignments given to the Strike Team e Assistant Strike Team Leader assists the Strike Team leader e GIS Specialist is responsible for random sampling process creating and printing maps assigning teams to clusters and assigning equipment e Epidemiology Specialist is responsible for data collection data analysis of survey results and report writing e Interview Team Leads are responsible for overseeing 5 interview teams of two people each approximately 10 people e Interviewers are responsible for safely conducting household interviews within assigned clusters To learn more about how to organize your data collection team using the ICS structure the Texas Department of State Health Services CASPER Strike Team created this pocket field guide Other team members to consider for your survey team might be Information Technology IT staff to assist with any technical support and if you are operating within a health department and environmental health staff that might have GIS expertise or are trained in interview methods for routine inspections Lastly if you are planning to conduct a survey in communities where residents speak other languages such as Spanish be sure to recruit team members who are
29. to that point area A 11 56 A e Map View GASTON 2015 FINAL Survey Map View GASTON 2015 FINAL Survey Tap the call out box for the survey point you where would like to start an interview Select Begin Interview to open the survey record screen Map View GASTON 2015 FINAL Survey Point 10 emi N tate EP Info Survey E gt Bagin a Survey Interview for Point 10 3 or Cancel gt The form should load automatically If it doesn t click on Add Record from the Record List to open up the survey form and start a new survey Note it may take a moment to load inf Record List Fill in the questionnaire Once complete select Save Record at the top ERA Record viewer CAPTURE CODRDINATES T i 2014 ity Health Hello am and this is representing Durham County Department of Public Health Show badges We are here to ask you to participate in a health opinion survey for our county The purpose of this survey is to learn more about the health and quality of life in Durham County The Health Department and the Partnership for a Healthy Durham will use the results of this survey to develop plans for addressing major health and community issues The survey is completely voluntary It should take no longer than 20 minutes to complete If you have already completed this survey or if you don t live in Durham County please tell me now Would you be willing to participate Do you live in this
30. when conducting a survey e Incident Commander would be team leader e Public Information Officer send out communications about surveys to local papers law enforcement neighborhood associations e Liaison Officer may report Updates for overall CHA team or committee e Safety Officer would provide safety training and monitor safety of those going out to conduct interviews e Operations Section would perform the tasks needed to complete the survey such as going out and conducting the interviews e Planning Section may set the survey collection dates anticipate resources that will be needed for data collection determine interview teams and prepare survey packets and maps e Logistics Section would provide the resources needed to complete the surveys such as clipboards pens tablets chargers transportation snacks etc e Finance and Administration Section would provide administrative support to monitor and allocate staffing volunteers or money to complete the resources They may also reimburse teams for travel expenses or track staff time Above is an example of how the Texas Department of State Health Services organizes their teams for conducting CASPERs This approach integrates into the ICS structure where the strike team leader Implementing the survey Page 15 reports either as a group or branch to the Operations Section supervisor Interview teams are added as needed In some situations there may not be enough people
31. will want to e Explain how the community was engaged in the survey process selecting survey questions volunteering to conduct survey etc Allow community members and survey volunteers a chance to share their experiences collecting surveys e Describe the methods of when and how the survey was conducted including the number sampled and how you selected the sample e Provide the demographic results of the respondents compare the data with the latest census data for your county or region highlighting depending on the audience you may want to include the 95 confidence limits with data and describing any limitations to the survey It is important to convey what level the results can be generalized typically the results are only generalizable at the county or region level If the results are being included in a report include a copy of the questionnaire as an appendix Examples Wake County NC 2013 e Appendices D F Durham County NC 2010 e Summary presentation e Appendix G Lonnmemneeeeememeeeneeeeeeeeeeeeeeemeeeeeeeemeeemeeeeeeeemeeeeeeeemeeeneeneeeeeeeeeeeemeeeeeeeemeeemeeeemeeemeeeeeeeemeeened Analyzing and reporting results Page 68 APPENDIX Appendix Page 69 Appendix A Resources Basic Resources CASPER Online introductory module available at http tinyurl com XXXXX Epi Into M software download hittp wwwn cdc gov epiinto html downloads htm Epi Into M Community Health Assessment Tutorial
32. you are planning on using excel or some other database software for the compilation and analysis of your results The ep7 files can be imported back into your Epi Info database using the password a which is automatically created by the app EEE He GASTON 2015 FINAL37 04 29 2015 Survey Files Survey Name Type Size Date Modified clusterfile km KML File 221 4 KB Apr 28 2015 11 24 AM pointsfiie kmi KML File 351 2 KB Apr 28 2015 11 24 AM questionaire xmi XML File 236 7 KB Apr 28 2015 11 24 AM questionaire_NoteEdge_ 201505010546 epi7 EPI Survey Fie 13 6 KB May 1 2015 5 46 PM questionaire_NoteEdge_201505010546 xml XML Fite 10 2 KB May 1 2015 5 45 PM questionaire SamonGs5 201505010056 epi7 EPI Survey Fite 57 3 KB May 1 2015 1254 PM questionaire SimonGs5 201505010056 xm XML File 43 KB May 1 2015 12 54 PM guestionaire SimonGs5 201505010211 epi EPI Survey File 71 2KB May 1 2015 209 PM guestionaire SimonGs5 201505010211 xmi XML File 53 4 KB May 1 2015 2 09 PM guestionaire SamonGs5 201505010213 epi7 EPI Survey Fie 71 2 KB May 1 2015 2 12 PM guestioname SimanGs5 201505010213 jamt XML File 53 4 KB May 1 2015 2 12 PM guestionaire SimonGs5 201505010239 epi7 EPI Survey Fie 85 5 KB May 1 2015 2 37 PM questionaire SimonGs 201505010239 xmi XML File 64 1 KB May 1 2015 2 37 PM guestionaire SimonGs5 201505010306 epi7 EPI Survey File 99 5 KB May 1 2015 3 04 PM guestionalre SwnonGs5 201505010306 xmi XML File 746 KB May 1 201
33. 0 teams ideally the teams will represent a mix of individuals based on gender race ethnicity age students vs professionals and locals with non locals out of town volunteers or state personnel Also consider teaming folks Up based on skill and experience Ideally have at least one person on the team who would feel comfortable interviewing strangers Also make sure that team members feel comfortable with the technology and data collection methods before sending them off These considerations are for safety and to evenly distribute local knowledge and field experience Throughout the team pairing Implementing the survey Page 41 process remember to be flexible and allow for last minute changes to teams Appendix B Is an example of how of a team assignment list When managing teams be sure to collect information such as team number names cell phone numbers times checked out and in equipment whether checked out in the number of surveys completed each day and notes on troubleshooting for team if there were issues with getting to survey sites with the equipment etc This information can be tracked in a Word document Appendix C as an example or easily in a spreadsheet Field Safety The safety of team members is the top priority not completing the surveys and collecting the data Communication with law enforcement and community partners Is key in informing residents about the survey and what to expect which will ultimately incr
34. 5 3 04 PM guestionaire SimonGs5 201505010341 epi EPI Survey File 113 6 KB May 1 2015 3 39 PM guestonaire SrmonGs5 201505010341 xmi XML File B5 2 KB May 1 2015 3 39 PM guestionaire SimonGs5 201505010400 epi EPI Survey Fie 127 6 KB May 1 2015 3 59 PM questionaire_SimonGs5_201505010400 xmi XML File 95 9 KB May 1 2015 3 59 PM guestionaire SimonGs5 201505011046 epi7 EPI Survey Fie 14 4 KB May 1 2015 10 45 AM guestionaire SimonGs5 201505011046 xmi XML File 10 8 KB May 1 2015 10 45 AM questionaire_S nonGs5_201505011114_epi7 EPI Survey File 28 6 KB May 1 2015 11 12 AM questionaire SamonGs5 201505011114 xmi XML File 21 4 KB May 1 2015 11 12 AM questionare_SenonGs5_ 201505011152 epi EPI Survey Fite 43 1 KB May 1 2015 11 50 AM questionaire SimonGs5 201505011152 xmi XML File 32 3 KB May 1 2015 11 50 AM guestionalre SimonGs5 201505020009 epi EPI Survey Fie 184 9 KB May 2 2015 12 08 PM auestionaire SmonGs5 201505020009 xml XML File 138 6 KB May 2 2015 12 08 PM questionaire_SimonGs5_201505020040 epi7 EPI Survey File 199 KB May 2 2015 12 38 PM guestionaire SimonGs5 201505020040 om XML File 149 2 KB May 2 2015 12 38 PM guestionaire SrmonGs5 201505020116 epi EPI Survey Fie 2128 KB May 2 2015 1 14 PM guestionaire SimonGs5 201505020116 xmi XML File 1596 KB May 2 2015 1 14 PM questionaire_SenonGs5_201505020141 epi EPI Survey File 227 1 KB May 2 2015 1 39 PM Appendix J includes some common FAQs from teams while in the field collecting interviews Implementing the s
35. 8 and Survey 1 14 Ai Code in St ee KE SE 103 Main St FC Tr 107 Main St st sg gt TA Wn ae t lt po Main St Pp EE Min St a ee OE Each attempt is labeled as the cluster number this team was assigned to Cluster 2 followed by the survey number It took the team three tries to get the first survey completed so they keep using 1 until an attempt is successful For the first interview in this cluster 3 attempts were made at 101 Main St no one home 103 Main St interview refused and 107 Main St interview completed Then the next attempt is labeled 2 2 and the team was successful in getting an interview at 108 Main St Then they move on to 2 3 where at 112 Main St no one was home and they were successful in getting 2 3 at 115 Main St The next address to appear on the form would be labeled 2 4 If this team was unable to finish the cluster and another team went out the next day the other team would start at point 2 4 If 2 4 was attempted the previous day and only NOs not occupied no one home were noted the new team would go back to those same households to see if anyone was home this time They would not revisit any homes that were marked IR interview refused or IC Interview completed Sometimes a cluster is small enough that teams are essentially knocking on every door If this is the case you can return to homes that were not occupied 101 Main St for example until you reach the quota It can be diffic
36. ASPER Because everyone in the community has a chance to participate CASPER results are representative of the community and generalizable beyond the individual respondents This is not true of a convenience sample which only represents the individuals who responded to the survey Since CASPER is based on the rapid needs assessment framework results can be collected and processed in several days giving decision makers the information they need in a timely manner Door to door surveys like CASPER have higher quality data than other methods The response rates are usually stronger than mailed surveys the data is more complete and the interviewer can clarify questions that may be unclear In addition interviewers in the field can gather data by personal observation such as the physical condition of roads or structures If volunteers are used to conduct interviews CASPER can be significantly less expensive than other data collection methods By going into the community interviewers also promote increased community awareness of the organization e g health department or hospital conducting the assessment Interviewers can distribute fact sheets contact information or resource lists In this way the CASPER serves an educational function as well as its primary function to determine the health status of the community When to USE CASPER The CDC specifies three post disaster situations in which CASPER is especially useful when the effect of the
37. ASPER staff roles with their definition A Strike Team Leader F Responsible for conducting safely household interviews within assigned clusters B Assistant Strike Team Leader C Responsible for random sampling process creating and printing maps assigning teams to clusters and assigning equipment C GIS Specialist E Responsible for overseeing 5 interview teams of two people each approximately 10 people D Epidemiology Specialist B Assists the Strike Team leader E Interview Team Leads Responsible for overall survey collection organizing and directing the field command post and overseeing the tactical assignments given to the Strike Team F Interviewers D Responsible for data collection data analysis of survey results and report writing Implementing the survey Page 48 Chapter 3 Implementing the survey Implementing the survey Page 49 CHAPTER 3 IMPLEMENTING THE SURVEY Field Survey Methods At this point in the CASPER process you should have a thorough and efficient questionnaire in the hands of well trained volunteers ready to head into the safely informed community fo visit the 210 sites you have selected via a two step cluster method This next section of the manual will explain e How to implement a two stage cluster sample in the field e How to use survey start sites or points e Strategies for efficient data collection and identify special sampling situations e Data collection tr
38. BEL is the combination of cluster and point number i e 3 9 for cluster 3 point 9 POINT X and POINT Y are the geographic coordinates latitude longitude in decimal degrees You can download AN EXAMPLE POINTS KML FILE HERE e Questionaire Filename Choose File No file chosen THE NORTH CAROLINA gt Institute for Public Health in fo Upload Files Selecting Survey Sites Stage 2 Once you have completed the first stage of the sample selection you can move onto selecting the survey sites Choose a fixed number k of housing units per cluster to approach the most common approach is to select 7 housing units per cluster The magic number for cluster sampling is 210 30 clusters and 7 interviews per cluster There are several methods for selecting survey sites Choosing the method that is appropriate for your situation will largely depend on your available resources It is important to remember that survey sites Implementing the survey Page 32 should be random in all three options The first option is a simple random sample The second option is called the EPI method and requires the least amount of planning The third option is recommended by the CDC and Is called systematic random sampling Option 4 requires random waypoints fo be generated ahead of time Option 1 Simple Random Sample SRS To use this method you must have a census of your area or a complete list of housing units within the clu
39. Collect SMART User Guide for Project Managers Using CASPER THE NORTH CAROLINA Institute for Public Health Developed by the North Carolina Institute for Public Health Development Team Principal Investigator Rachel Wilfert MD MPH CPH Project Manager Matt Simon MA GISP Research Associate John Wallace PhD MSPH Research Associate Kasey Decosimo MPH Curriculum Developer Meredith Davis MPH Curriculum Developer Kristen Ricchetti Masterson MSPH Manual Editor Annika Pfaender Acknowledgements David Zane Tracy Haywood Dr Jennifer Horney and her PHEB 689 Methods in Field Epidemiology class epidemiology graduate students from Gillings School of Global Public Health and community members staff and volunteers from Durham County Department of Public Health Pender County Health Department Chatham County Public Health Department Forsyth County Department of Public Health Gaston County Department of Health and Human Services and Granville Vance District Health Department for providing feedback during software piloting For more information about this project contact North Carolina Institute for Public Health Gillings School of Global Public Health The University of North Carolina at Chapel Hill Campus Box 8165 Chapel Hill NC 27599 8165 919 843 556 1 collectSMART unc edu http sph unc edu nciph ta SMART Table of Contents Page 3 Table of Contents CHAPTER E INTRODUCTION cnonanensnondnecanongnnd
40. MART Data Collection FAQ rrnnnnnrnnnvnnnrnrnnnnnrernnnnnrernnnnnnernen 8 Table of Contents Page 5 Chapter I Introduction Introduction Page 6 CHAPTER 1 INTRODUCTION Project Background Collect SMART The Survey Management and Response Tool is a suite of software designed to help users manage and implement data collection efforts including the Center for Disease Control and Prevention s CDC Community Assessment for Public Health Emergency Response CASPER Through collaboration with local health departments and their community and hospital partners Collect SMART was designed developed and pilot tested to be an easy to use mobile app and project management dashboard to facilitate field data collection efforts Collect SMART allows for the rapid collection of community level data for decision makers based on validated sampling methods improving the timeliness of data collection while reducing the cost of purchasing or maintaining mapping or data collection software and equipment Collect SMART was designed to overcome some of the technical barriers of primary data collection using GPS units and CASPER including Difficultly training volunteers in GPS units Lack of wireless connectivity Inability fo upload data immediately upon capture to a cloud based database Cost prohibitive nature of GPS units and proprietary mapping software Inability to track results in real time Given the increasing use of smartphones and ta
41. Time d Kanga Alcohol ab 7 Checkbox g Ves No fa Oiscnminsbon rans m Opton e Yes No 98 Command Button Domestic viol lence ial age Option ord oe GE New Field Group Grid es roghanna ab D Tahi bp Command Button Comment vol Z Legal Gang involvem E Page Check Code tail Image E Raate Homelessness ay Mirror 07 es Lack of cle g i 3 ay Py 3 Grid Felde Lack of epmmunky support TILT NE ita Pages Iz 5 Promects Lack of exercise ee Legal Values Lack of hy food choices or affordable healthy food ed Codes s Sta Lack of places to exe on Tab es Comment Legal Lack of transiabon or mterpreta when needed for daily activities lect sb Relate ome poverty glect and ab Pollution or r wat d ee Ready enUS 29 NUM Hi Demonstration Videos Epi Into 7 Create Forms Part 1 9 minutes Epi Into 7 Create Forms Part 2 11 minutes 22222222 EEE EEE Implementing the survey Page 21 ACTIVITY 2 1 DEVELOPING A GUESTIONNAIRE Time 10 minutes Instructions You are working with the local hospital and other community partners to design a questionnaire for a community health assessment One of the partners would like to add a guestion about diabetes since there currently is no local data providing prevalence for adults Answer the following questions Additional CDC BRFSS Questionnaires website Resource http www cdc gov brfss questionnaires index htim 1 What is a potential que
42. a skip pattern to avoid asking subjects irrelevant questions Unless missing data is due to errors in data entry it usually cannot be corrected without re contacting the subject which is not possible for anonymous surveys Instead there are several other approaches that can be taken when dealing with missing data e Often the best approach is to continue to analyze the results with the missing data but do not include the missing category in your analysis This will reduce the total number of respondents for any questions with missing data but it will maximize the total amount of data that you have for other questions Coding the data in this manner allows you to keep the record with missing data as opposed to deleting in entirely from your dataset e Another common approach is called complete case analysis which is usually only valid when the amount of missing data is small less than 10 for each variable This method deletes subjects with missing data trom the analysis so that only subjects with complete data are analyzed Another method is to delete variables that have a large amount of missing data Of course if the variable is important for your analysis and is related to your hypothesis then this method should not be used And presumably you would not have asked the question if you did not think it was important so dropping variables may not be a good solution e Another method that was common practice for many years was to tr
43. aining and project management using the Collect SMART app and website Scheduling If you have 10 teams of two volunteers each and you are planning for 3 days of surveying we recommend that you survey Thursday through Saturday As a general rule knocking on someone s door before 10 00 a m might be considered rude and could adversely affect your response rates The recommended time to catch residents at home is between 2 00 p m and dinner time If your teams have tried all the homes in their assigned areas without success tell them to take a break and return at 2 00 p m Interview teams should not go door to door after dark Depending on what area of the country you are in and what time of year if is will determine what time you request teams to return to the base station Be sure to set that time up front during the training and tell teams they MUST return to the base station at the designated time Also keep in mind that this can be exhausting work for your teams so avoid asking them to stay too late Finally if your teams are out too late they will likely be interrupting residents during dinner time which is not advisable If you are planning for just 2 days of surveying It s important to include a weekend day to make sure you reach people who have to work during normal business hours so a Friday and Saturday survey plan is recommended To complete the surveying in two days you might need to recruit more than 10 teams As we discussed earlier
44. and be aware of the surroundings If there is a bad feeling leave e When approaching doors be sure to maintain a comfortable distance First knock and then step back After you try 2 or 3 times knocking and using the doorbell it s time to move on e Regarding gates screened porches or homes with side or back doors use your judgment If there is a gate check for signs of dogs before entering You probably shouldn t assume you are welcome at someone s back door so be courteous and always approach the front door e Be aware of dogs cats and other animals even if they seem friendly Avoid petting or picking Up pets Even if you are a dog person everyone should be cautious about dogs e lf at all possible complete the interview outside of the home Note that if an elderly or disabled resident comes to the door it might be difficult or uncomfortable for them to stand at the door 10 15 minutes Offer to sit on their porch If they invite you in use your best judgment It would be good to come up with a code word or sign ahead of time to share with your interview partner for these situations Base Station Check In Once team members wrap up their final survey for the day teams should check back into the base station no later than 6 00 p m These times can be revised based on project requirements but It Is recommend that teams report back fo the base station before dark and before dinner time When teams arrive back to the base collect
45. and point lists Include current aerial imagery or street maps cluster boundaries and survey start sites This will helo teams orient themselves and stay within the selected cluster s boundaries Although field maps aren t necessary they can be a helpful addition to the digital map data displayed on the GPS units or tablets 31 f p e 2 vau Man teen Teer te Team 5 Team t tear t Tear 8 ler 8 Survey i Pender County NC fae suara UAGHLINA Community Health Opinion Survey March 26 29 2014 Sample Overview Map Showing Clusters Sample Aerial Field Map 6 A Point List is a list of the households seven in the 30 7 method within the cluster that have been identified to visit fo conduct the survey Information included should be the full address possibly including GPS coordinates This is not the same thing as a tracking form which will include additional details about what happened at each household visited But this point list is the guide for where each team will be heading Implementing the survey Page 44 CLUSTER 1 Staging Area Health Department Point Address City i Latitude Y 1 1 123 Main St Smallville 77 777 1 2 101 Main St Smallville 66 666 1 3 100 Hwy 101 5 Smallville 77 777 1 4 111 HWY 1015 Smallville 77 777 1 5 333 Broad St Smallville 1 6 555 Oak Ave Smallville 22 222 1 7 1001 HWY 101 N Smallville 7 Atechnology cheat sheet with basic instructions on how to use the technology This can be
46. being conducted This step should be done in consultation with a Statistician Weight Total of households in the Sampling Frame of households interviewed within the cluster x Total of clusters selected Below is an example in a sampling frame of roughly 50 000 Out of the 30 clusters within 28 of them all 7 expected interviews were conducted In cluster 29 6 interviews were conducted and in cluster 30 5 were conducted In the raw data create a weight column and apply the calculated weight to each interview as shown below In this example the weight for clusters 1 28 will be same while 29 and 30 have a larger weight since fewer surveys were collected in each cluster CLUSTER SURVEY WEIGHT 28 1 238 10 28 2 238 10 238 10 50 000 housing units 7 x 30 238 10 which will be applied ii e 00 to the 28 clusters where 7 interviews were completed 28 5 238 10 28 6 238 10 28 7 238 10 29 1 277 78 29 2 277 78 29 3 277 78 50 000 housing units 6 x 30 333 33 which will be applied 29 4 277 78 to the one cluster where 6 interviews were completed 29 5 277 78 29 6 277 78 233 33 50 000 housing units 5 x 30 277 78 which will be applied 30 2 333 33 to the one cluster where 5 interviews were completed 30 3 333 33 30 4 333 33 30 5 333 33 Once weights are assigned frequencies can be calculated for each of the interview questions To calculate frequencies in Epi Info be sure to specify the weight variable calc
47. blets even in resource poor environments data collection on these platforms makes this methodology more accessible to a wider audience This user guide will walk project managers through all of the steps to implementing a CASPER using Collect SMART Collect SMART Application Purpose The purpose of Collect SMART is to help users manage and implement data collection efforts including CASPER Fig 1 A project manager can access Collect SMART to pull their random sample of clusters and then upload this spatial data to their mobile devices installed with the Collect SMART app Data is stored securely on the cloud and can be analyzed in real time by the project manager Sample Selection Project Dashboard Figure 1 Collect SMART software suite Introduction Page 7 Audience The audience for the Collect SMART application includes state and local public health professionals in the United States who require a rapid and simple way to assess the public health needs of their population In addition Collect SMART can be used by hospital health care and other community agency staff who are involved in conducting community health assessments i e Health Impact Assessments preparedness assessments or other community surveys Application The set of tools provided by Collect SMART enables users to 1 Share digital questionnaire files 2 Select a random sample of neighborhoods using U S Census data 3 Locate the selected neighborhoods
48. blic health issues and we will use this information to better inform the health department strategies for providing services to the community After the introduction if the resident is willing to do the interview interviewers should go ahead and conduct the interview ideally outside of the home but they should trust their best judgment If the resident is firm they do not want to do the interview they should be thanked or their time and the interviewers should discreetly record their address on the tracking form so they do not get bothered again However it is also important to train interviewers that they shouldn t take no for an answer right away Some residents might say no but really just need a little bit of encouragement Interviewers should remind residents of the value of their perspective and input If the resident refuses twice the interviewers should move on to the next site It is important not to coerce participation The Interview When training interviewers on the questionnaire be sure that they follow these guidelines 1 Ask the questions properly Don t put the questions into your own words or skip words 2 Ask questions in the order presented The questions are ordered in a particular way and often involve skip logic 3 Ask every question that is specified Don t skip around or skip questions entirely 4 Record the responses accurately If you have to read al list of responses read them all Implementing th
49. choices and allow for a text entry to specify what other means Keep in mind that allowing open ended responses will add to data analysis time as you will have to attempt to categorize or group these responses to make sense of them Limit the number of open ended responses whenever possible Questionnaire Layout The overall goal of good questionnaire format is to make questionnaire completion as straightforward as possible Questionnaires should include detailed instructions scripts and definitions To format a questionnaire that minimizes burden for the resoondent and the interviewer format the page layout with a vertical flow from question to question and from response option to response option Here are some tips for questionnaire layout e Questions should be formatted in a logical order It is best to group questions on related topics together e Place open ended or sensitive questions towards the end of the survey as these tend to be more taxing on the respondent It is also important to build rapport with the resoondent before asking sensitive or personal questions e Number every question and if there are skip patterns If the response Is no skip to question be sure to use clear cues either in text or with arrows to guide respondents and interviewers e Response option codes should be consistent e g responses to questions should either be labeled A B C Dor 1 2 3 4 but do not use letters for one question and number
50. choices may be stored as text which may impede analysis efforts Data cleaning Is an ideal time to check for these issues and update the coding before analysis In addition some statistical analysis packages require a standard type of numeric coding which might be applied after the data has been collected You may also choose to combine variables or answer choices For example if each answer choice for a question was coded independently you may wish to create a new summary variable that combines all of this information For questions with an other please specify answer choice you may wish to review the text entries to organize those answers into new or existing numeric categories which will give you precise 95 confidence intervals during weighted analysis instead of lumping all text entries into the other category Some of the most common sources of errors are errors in the design of the data collection instruments such as poorly worded questions or improper use of skip patterns Other sources of error may stem from improper functioning of equipment as well as errors made by interviewers offen due to inadequate training The survey participants themselves may also introduce errors if they provide the incorrect response which can occur if they do not read or interpret a question properly or if they intentionally provide a false answer Errors can also be introduced after the data has been collected most often by data entry errors or
51. d HONEN NGN niina 7 Fe 7 EE TANN 7 DON 7 PAUCI CS ERE 8 RIC OU e E ma E ENE 8 MSR 9 Advantages OF CASPER vere 9 MOVES 7 ACTIVITY 1 45127710 ERE se Error Bookmark not defined ACTIVITY 1 KNOWLEDGE CHECK ANSWER KEY Error Bookmark not defined CHAPTER 2 PREPARING FOR YOUR CASPER cccssscosssccescccesccnssccescecesccscceescs 14 OCC THOTT MM EN 14 TN 14 Using the Incident Command Syste PE a aai 14 ENN 16 ENN NNN 17 Questionnaire LO OUT MMA a a e t 19 alale ide TME US STOO RR 20 CHE CIING tom ABD SEN PP 20 ACTIVITY 2 1 DEVELOPING A QUESTIONNAIRE Error Bookmark not defined ACTIVITY 2 1 DEVELOPING A QUESTIONNAIRE ANSWER KEYError Bookmark not defined TE NT 24 MTT 25 se KSO Se EEE EEE OE 27 Selecting the NNN 28 Selecting Survey MEL 32 Reeddila ope Mo 5 EAA EEE SE 35 Uzelalg ek 1 Ga SJENERT JE 36 TN 4 ae ok ei stats ca rte sot A terra EIA AA A AA 42 Maroa ana AA FERNER 44 TEONE CEN ENE 44 ACTIVITY 2 3 FORMING A SURVEY TEAM Error Bookmark not defined ACTIVITY 2 3 FORMING A SURVEY TEAM ANSWER KEY Error Bookmark not defined CHAPTER 3 IMPLEMENTING THE SURVEY csccoscccssccesccssccssccsscesccssccesccesceeecs 50 Feld NNN 50 DC FS SI e EGET SE 50 PETON SG JE i i eee SI Gong Door 16 DOO RN 00 secosou ti li 52 SIDS CIC Sampling 106710 ge 59 Table of Contents Page 4 POI FOC CN EEE EEE 54 Monitoring Teams and Analyzing Results in Real time
52. d in each county for a total of 560 interviews Implementing the survey Page 25 eet Ni A sampling frame can also be defined by geographic boundaries e g houses in a specific direction from a landmark such as a road or a river or by use of a subpopulation e g the Hispanic or Latino population in your community The map below shows how a 100 and 500 year floodplain shown in dark and light blue was used to delineate a sampling frame Since the assessment was most concerned with areas likely to have flood damage sampling was restricted to clusters where at least 50 of the area was within a flood prone region Remember that if you choose to draw a sample from a sub population you can only generalize your results to that entire sub population if every member of the sub population has a chance of being selected For example if you wanted to conduct a survey with only the Hispanic or Latino population you could draw your sample from only census blocks where at least 50 of the population identified themselves as Hispanic in the last census This can be a more efficient way to reach this population than a convenience sample but keep in mind that you CANNOT generalize the results to ALL Hispanics and Latinos since your sampling frame was limited to a smaller set of census blocks However CASPER sampling can still be used to assess the needs of sub populations so long as you understand how to interpret your results Implement
53. e Each tablet is locked with a code To access the tablet draw the following pattern Square starting in middle drawing to the right don t complete Getting Around At the bottom of every screen no matter what you re doing you ll always find these three navigation buttons Back opens the previous screen you were working in even if it was in a different app Home if you re viewing a left or right Home screen opens the central Home screen ol Recent apps opens a list of thumbnail images of apps you ve worked with recently GPS and Battery The CASPER App requires connection to a server Make sure the GPS button is ON and keep WIFI off Quick Settings Access by swiping down from the right top corner of any screen twice to adjust Wi Fi settings manage display brightness and view battery charge e Charge your tablet during lunch using the car charger Getting Around EpiInfo Click on the Epilnfo icon on the home screen Then click on Map View An automatic login screen will briefly flash on the screen EPT IN FGY StatCalc Collect Data Analyze Data Collect SMART lt From the Map View menu onthe lower right select Clusters and select the cluster number you are assigned to Tap onthe blue dots to pull up the call out of the point number and address OR you can select Points to view all starting points for that cluster Then select the specific point on the list to zoom
54. e Question At a Time Ask one question at a time Least Useful Better In the event of a large In the event of a large scale scale disaster would you disaster would you evacuate evacuate and where a Yes would you turn to for b No shelter information If Yes where would you turn to for shelter information a 911 b Television c Radio d Internet c Other Implementing the survey Page 18 For closed ended questions which are those that list response options make sure that the response options are mutually exclusive In other words make sure that the response option categories don t overlap Mutually Exclusive Responses Avoid overlapping of response categories Least Useful Better What is your age What is your age a 18 34 years a 18 34 years b 34 64 years b 35 64 years c Over 65 years c Over 65 years In the first example the first list of answer options would be confusing for a respondent who was 34 years old as this number is included in both of the options The second set is better as it is more clearly defined All questions should have a response marked Always provide an option for don t know and refused Mark refused if the resoondent does not want to answer the question mark don t know if they don t know the answer This will also help you distinguish between missing data or non responses Use the other choice to capture any responses that don t fit in the
55. e Training Introduction to CASPER 25 minutes lt a i i pp i pp e ip D S i a S n s Table of Contents Page 10 ACTIVITY 1 KNOWLEDGE CHECK Time 5 minutes Instructions Answer the following multiple choice questions Compare your answers to the answer key on the following page Additional None Resources Required 1 What is the primary purpose of CASPER a To gather disease surveillance data on a recurring basis b To offer disaster assistance including providing for basic needs such as food water and shelter c To determine the health status and basic needs of an affected community d To identify individuals who qualify for public health programs Answer 2 How are households chosen to participate in a CASPER survey a Simple random sampling in the geographic area of interest O Two stage cluster sampling starting with Census blocks d Convenience sampling of nearest households Q All households in the geographic area of interest are included Answer 3 Which of the following is NOT a potential advantage to using CASPER over another data collection method a Low cost O High quality data d Few technical requirements Q Rapid results Answer ee ll el ll lt SSG Table of Contents Page 11 ACTIVITY 1 KNOWLEDGE CHECK ANSWER KEY 1 What is the primary purpose of CASPER a To gather disease surveillance data on a recurring basis b To offer disaster assis
56. e and you should interpret your results with caution Tracking form codes can be entered into a database to count the numbers of attempts made completed interviews refusals and ineligible participants Success Rate Also called the completion rate the success rate is a measure of how close you were to reaching your goal For example if your goal was to complete 210 surveys and you completed 200 then your success rate would be 200 210 or 95 2 Success Rate of Completed Interviews Goal Number of Interviews Success rates below 80 168 out of 210 result in an unacceptably low number to represent the sampling frame Efforts should be made to ensure that clusters with a low response are revisited to ensure sufficient numbers Cooperation Rate The cooperation rate is the proportion of housing units where contact was made and an interview was conducted Cooperation Rate of Completed Interviews All housing units where contact was made The denominator includes all completed interviews incomplete interviews and refusals You should also include homes where a language barrier prevented the successfully completion of an interview This cooperation rate is often call the response rate and typically falls between 60 80 for household interviews Rates below 60 may indicate the potential for bias in the results but they do not necessarily mean that bias exists Contact Rate The Contact rate is the proportion of housing units whe
57. e first row in red shows the selected cluster based on the first random number of 84 Cluster 8 is selected because 84 Is within the cumulative housing units range of 75 86 The next random number is 102 Cluster 10 would be selected since 102 is in the range 99 110 Repeat this procedure until you have all 30 clusters and then renumber the clusters from 1 30 It is possible that some clusters will be selected more than once If a cluster is selected twice then you will need to collect twice as many surveys In that cluster Selecting sample using GIS If you are using ESRI s then you have a couple options to automate your sample selection A toolbar that is compatible with ArcGIS 9 x is available upon direct request This toolbar was created by the North Carolina Public Health Regional Surveillance Team 5 also called PHRST 5 the NC division of Public Health s office of Public Health Preparedness and Response and the Department of Agriculture This work was envisioned and guided by Dr Mark Smith Guilford County Health Department and was partially funded by an ESRI industry Homeland Security grant back in 2002 This is an ArcGIS toolbar that runs in ArcMap 9 x and below It is NOT compatible with ArcMap 10 x The toolbar performs 3 primary functions including 1 Automates the procedure of selecting both the first stage and second stage samples including reverse geocoding points 2 Assigning survey sites to multiple survey teams 3 Generation of
58. e survey Page 38 During the training have interviewers work in pairs and practice conducting the survey to each other using a copy of the questionnaire Below are some suggestions for common interview techniques that can help make the interview run smoothly and capture the best quality data e Pause After reading the question pause briefly to wait for a response Don t rush to help them find an answer Be patient and allow the individual a moment to consider If you sense confusion you may offer to repeat the question If you are asked to repeat the question please do so e Clarify If appropriate you may clarify the question This can get tricky because this is where you might influence the response by putting your own spin on the question This is considered surveyor bias The best course of action Is to repeat the question and offer clarification sparingly However clarification can be important when the participant does not understand the question You may need to clarify what sick means or what constitutes an emergency In these cases be careful to simplify the question using neutral words that do not influence the participant s answer e Practice active listening Practice active listening by looking for non verbal cues For instance if someone shakes or nods their head rather than saying Yes or No you will need to repeat the participants reply verbally to confirm their answer This is especially important if t
59. e which ones have a high concentration of Spanish speakers or any Clusters that have additional safety concerns Appendix Page 72 Appendix C Staff and Equipment Check Sheet TEAM Staff amp Equipment Check Sheet Date please circle 1 4 ae gt Dmy oo Team Member Names Leave start time AM PM Arrival finish time AM PM Driver Name Mileage Sheet Submitted Yes No N A Equipment Check in check all that apply Binder Laminated cards Tablet Tablet car charger and USB cord GPS car unit and mount GPS car unit charger Incentives Base Station Notes concerns problems encountered comments Call In Status Notes safety status surveys complete expected time return to base station Time of call Notes Time of call Notes Time of call Notes Appendix Page 73 Appendix D Survey Press Release Template Community Health Assessment Press Release Template FOR IMMEDIATE RELEASE Date NAME COUNTY INVITES PARTICIPATION IN COMMUNITY HEALTH OPINION SURVEY To identify and respond to the health needs of residents Name County is partnering with a number of local organizations to conduct a Community Health Assessment The assessment will include a health opinion survey of approximately 210 residents across the county Starting date surveyors will be conducting randomly selected door to door surveys to get resident s opinions on the m
60. ease team safety It is imperative that team members follow safety guidelines Remind team members to use good judgment and common sense when it comes to ensuring their personal safety Communication At least 1 week before the surveying is scheduled to begin work with agency communication staff to alert residents that surveyors will be in their neighborhood Communications can start with issuing a press release and subsequent media interviews about the data collection project survey and what residents can expect Include a 1 page information flyer about the survey process that survey teams can take with them if residents have questions Notification should also be sent through project partners and their networks to let folks know about the survey as well as calling local law enforcement and neighborhood associations so that the surveyors will be anticipated You can also provide these partners a copy of your informational flyer If additional funds and staffing are available you can also mail letters or postcards about the survey to the census blocks that are selected for the sample An example press release is included in Appendix D and information flyer in Appendix E Base Station When setting up the survey base station consider having training and staging efforts in one central location This would need to be a room that is large enough to accommodate training and base station materials including audio visual equipment available for trai
61. eat missing data as another potential response category of course this can only be done for nominal variables Epidemiologists have since recognized that this method can lead to biased results and is therefore not recommended e For datasets with larger amounts of missing data imputation techniques are often used to make informed guesses as to what the missing value would have been based on the range of responses observed in other subjects and the other characteristics of the subject with the missing value With imputation you are insinuating what the resoonse might have been Imputation can be somewhat complex and requires more advanced statistical methods and software Duplications Duplications occur when there is more than one record in a dataset that contains the exact same information Duplications can arise in several ways most often when the same person s information is entered into the same database more than once Less commonly a person may be asked to complete the same interview or questionnaire twice Good record keeping and organization will usually prevent these types of errors from occurring Sometimes duplications can arise when multiple databases are merged together into one Duplications can usually be identified by looking for multiple records that have the same identifying information i e ID number A frequency listing of ID numbers can be used to make sure that each ID number only has one record If you are using Epi I
62. ecord Viewer 8 CAPTURE COORDINATES F SAVE 2014 Community Health Opinion Survey Hello I am and this is representing Durham County Department of Public Health Show badges We are here to ask you to participate in a health opinion survey for our county The purpose of this survey is to learn more about the health and quality of life in Durham County The Health Department and the Partnership for a Healthy Durham will use the results of this survey to develop plans for addressing major health and community issues The survey is completely voluntary It should take no longer than 20 minutes to complete If you have already completed this survey or if you don t live in Durham County please tell me now Would you be willing to participate Do you live in this household Are you at least 18 years old START INTERVIEW Latitude Longitude Date Time Record List Record ID 1 ADD RECORD Cluster Survey Datea Create Sync File Preview 3 4 29 2015 Generate Map Record ID 2 Analyze Data i Cluster Survey DateandTime Preview 1 4 29 2015 1 22 58 PM Record ID 3 Cluster Survey DateandTime Preview 2 4 29 2015 2 14 11 PM Record ID 5 Cluster Survey DateandTime Preview 1 3 4 29 2015 2 44 46 PM Record ID 6 i Cluster Survey DateandTime Preview 16 5 5 2 2015 10 38 00 AM Record ID 7 Cluster Survey DateandTime Preview 3 One of the most important things to stress is to save the
63. ect data A typical home screen for Android devices the home page for the Epi Info and the Collect SMART login screens are shown below Map View Login User EPI INFGY Enter your EPI account StatCalc Collect Data msimon Auto Login Password Visible Analyze Data Collect SMART Next demonstrate the mapping capabilities of the Collect SMART Start by going over the 4 main navigation buttons shown below on the bottom right of the screen 1 Clusters 2 Points 3 Map Type and 4 Staging Area When a user clicks the Clusters button they are shown a list of the numbered clusters on the bottom of the screen Then if the user clicks on a cluster number in the list they are Zoomed to that cluster and shown the list of available points in that cluster Implementing the survey Page 54 AG 9 MAN 11 57 Map View GASTON 2015 FINAL Survey y SB sl Ranlo 7 Te Cluster 10 PG e E Staging Area 0 BASE 990 W Hudson Bivd GASTONIA T o X v NS AST E Sx 7 n M i SoS an XY Ni lt e x SU Q E 10 1 1121 W DAVIDSON AVE GASTONIA COMPLETED 10 2 1041 RIDGE AVE GASTONIA COMPLETED 4 0 605 N RANSOM SI GASTONIA COMPLETED 10 5 401 APT N HIGHLAND S GASTONIA COMPLETED 10 6 405 APT N HIGHLAND S GASTONIA COMPLETED lt O 0 From the image above on the right choosing a point from the list will Then zoom the user to that point in the cluster and pull up the addre
64. eing the most up to date survey counts click the Refresh Points button or refresh the browser page Gaston Coun Art Central M Probation amp Parole Departmen e W Mauney Ave W Rankin Ave Cluster ID 1 TOTAL 201 1 10 2 10 3 20 4 10 5 10 6 20 7 10 8 10 9 10 10 10 11 10 12 10 13 10 14 10 4 15 20 16 10 17 10 HOME CONTACT US LOGOUT Surveys Complete 180 complete 90 90 90 60 90 100 85 100 100 100 80 100 100 100 70 100 90 100 As surveyors send their sync files to the server results are stored in the under survey files under the Admin tools on the Collect SMART website Choose Admin from the main screen and then Manage Projects HOME CONTACT US LOGOUT Administrator Tools Implementing the survey Page 57 HOME CONTACT US LOGOUT Manage Projects Matt Simon is Administrator Project information ID Survey Name Create Date 24 GASTON 2015 FINAL Update Data Delete Survey View Survey Files From the Manage Project page choosing View Survey Files will bring up a listing of all the files associated with the project Software Tip Choosing Delete Survey will permanently delete ALL survey files associated with your project so use caution SSS SSS On the listing of files you will notice both ep7 files and xml files Use the xml files if
65. esident of household was home those where there was a language barrier and those who refused to be interviewed This level of detail will be important for the project manager to use when calculating response rates that determine whether there is a chance that bias could exist in the sample More information about how to calculate response rates will be covered in the analysis section of the manual It is also very Important to remember that a different interview team may complete the cluster the following day and you don t want those team members to be approaching households that already refused or completed an interview The tracking forms are also used to be aware that they may be addressing the households with language barriers Below is an example of a tracking form used by Team 1B The starting point for this team s cluster was 101 Main St A tracking form template can be found in Appendix G Implementing the survey Page 45 Cluster Assessment Tracking Form Assessment Tracking Form Date Oct ist Interview Team 18 Date Interview Team ee Status Codes Status Codes e gt oa A IC Interview Conducted NE Not Eligible IR Interview Refused NO Mat Qecupied No One Home LB Language Barrier Stat Cluster 1 28 and Survey 1 14 Address Cod IC Interview Conducted NE Not Eligible IR Interview Refused 101 Main St NO Not Occupied No One Home LB Language Barrier Status PETER Cluster 1 2
66. ey 1 2 months in advance and then immediately start recruiting volunteers Recruitment materials should include an explanation of the purpose of the survey the dates and times of the survey training location and what to bring snacks water comfortable shoes and photo ID As you collect volunteer information be sure to include agency community affiliation cell numbers dates of participation encourage full participation rather than partial days whether or not the volunteer can drive a personal vehicle possible language skills such as Spanish and any dietary restrictions if you are providing food Another strategy is fo designate 10 team leaders and have the team leader be responsible for finding their partner for the 3 to 4 days of surveying You may find that providing volunteers with some incentives and recognition will go a long way toward encouraging their participation Incentives could include providing breakfast or lunch for at the start of each day Since volunteers will be out Conducting surveys from 10 a m to as late at 7 00 p m a to go lunch should be provided at a minimum Most often volunteers eat their lunch in their car or out in the field so keep that in mind when considering lunch vendors Ask a local restaurant sandwich shop deli or diner if they would consider donating gift cards to participants You can also ask local businesses to consider donating gift cards to volunteers If county vehicles are not available f
67. fall within that plus or minus 10 range Remember in any sampling situation each sample will produce only an estimate of the value of interest 95 Confidence interval is a range above and below the observed estimate where we would expect the true estimate to be 95 of the time since the observation is influenced by random error While results obtained through this two stage cluster sampling can be generalizable to the sample frame level e g the county as a whole results CANNOT be broken down with this same precision into race ethnic gender or income categories If you want to gather more information about a particular sub population consider conducting a focus group or key informant interview e 9g homeless Hispanic certain geographic areas Sampling Frame Before you address Stages 1 or 2 you must first determine the sampling frame for your CASPER which is the largest boundary of the population that will be included in your assessment Consider the community that you are studying for your assessment Is if currently defined as a city county or service area The purpose of Conducting a survey is to describe key characteristics of the population under study To derive an accurate estimate the survey sample must be representative of the overall population If the community being assessed is very large or dispersed over a large area the survey sample should be taken from as wide an area as is practical and not restricted to a
68. he resoondent offered a response that you didn t quite understand or didn t quite fit one of your categories For example a question asks the respondent to pick a choice on a scale of strongly agree to strongly disagree A common response might be Yeah I think so to which you should reply So would that be a Strongly Agree or Agree ma am e Maintain neutrality It is important to stress that interviewers maintain neutrality and avoid leading questions A good example of this might be a question that asks whether or not a resoondent smokes A leading way to ask this question would be You don t smoke do you This phrasing is giving the respondent the answer that you are expecting No I don t smoke This is another reason why It is important to ask the questions exactly as they are worded e Maintain pace Interviewers will need to maintain an efficient pace in the field Interview teams will get to meet all types of people going door to door Some people will be impatient and want you to rush Some people might want you off their porch as soon as possible and be ina hurry Try to keep a steady pace but don t skip over questions or paraphrase Some people might be slow to respond to a question If they are really struggling with comprehending a question and are still struggling to answer a question after further clarification offer to move on and record their response to that question as don t know or No response Refused to
69. here one side of the street is in one census block the opposite side is in another This is a rather common occurrence so be on the lookout for clusters that look similar to the example provided Donut holes are not a problem with Census tracts or block groups Implementing the survey Mobile Application Interview teams will be using only the Collect SMART mobile application The application is compatible with most Android tablets and phones To ensure that teams feel comfortable using the Collect SMART mobile application it is important to first go over the tablet features including how to turn the devices on and off how to monitor the cellular signal strength and how to turn the GPS and Wi Fi on or off Allow at least 30 minutes of demonstration time for the Collect SMART mobile application allowing teams the opportunity to practice Here is a recommended outline for demonstrating technology e Watch on screen demonstration Basic introduction to Android OS Epilnto app Other device features e Check for data signal e Email communication e Hand out devices to teams e Demonstrate again walking everyone through each step During the demonstration first show the home screen and ask how many people are familiar with Android phones or tablets with a show of hands This will give you a good sense of how much time you need to spend on the basics of Android Then identify the Epi Info app as the app we will be using to coll
70. ho easily find rapport with clients patients or community members Participants that are familiar to the community can be a benefit Outgoing friendly people with stamina and a tolerance for rejection can nicely balance younger individuals with the technical savvy to interpret maps and work with mobile technology Epidemiologists nurses environmental health and administrative staff at your local health department can be a great place to start recruitment efforts Also consider students in nursing medical or public health programs at local colleges or universities who often take courses In community health Many of these programs require students to complete practicum projects working with communities Don t overlook the importance of recruiting volunteers from among agency partners and members of your Community Health Needs Assessment committee Past surveys have included volunteers from hospitals Federally Qualified Health Care Centers medical society United Way agencies and non profit or community agencies such as American Red Cross local Latino advocacy organizations and Hospice Other volunteers to consider are those registered with the local Medical Reserve Corps staff in your local county s emergency services and medical ministries or nurse parish programs from local faith based organizations Implementing the survey Page 35 It is strongly recommended that you start recruiting early in the process Set a date for your surv
71. idemiologist GIS specialist and a volunteer coordinator Once the team is formed begin to schedule survey dates preferably when the chance of inclement weather is low Depending on the situation this may occur days or months in advance of the survey It is necessary to plan for volunteer training by developing or obtaining training materials Using the Incident Command System One way to organize the survey team is to use the Incident Command Structure ICS ICS is a standardized incident management approach that is typically structured to facilitate activities in five major functional areas Command Operations Planning Logistics and Finance Administration Even with limited staff the ICS structure can be used to conduct a community survey One of the advantages of using ICS is that it is scalable and flexible All of the functional areas do not have to be used depending on the scope of the survey ICS is also extremely useful not only does if provide an organizational structure for conducting a survey but it also guides the process for planning the survey as well We recommend that you use ICS for a CASPER to help hone practice and maintain skills needed for a potential disaster Your agency might be able to use this opportunity as an exercise that can meet your agency s disaster planning requirements Implementing the survey Page 14 Using Incident Command Systems to Organize Team Here are some examples on how it can be used
72. idity in the design of a particular question Validity Does question measure what it s supposed to measure Least Useful Better Where do you go most Where do you go most often often when sick when sick a Clinic Doctor s office b Hospital b Urgent Care c Community Health Center d Pharmacy Clinic d Emergency Room e Hospital By using this question we are trying to determine where residents access health care most often In response to the question Where do you go most often when sicke the first example gives options of clinic and hospital At first glance this may seem adequate but if we think further there are a number of options that could be encompassed by the one response option of clinic do they mean a doctor s office pharmacy or community health clinic In the second set of response options which includes a doctor s office urgent care community health center pharmacy clinic and emergency room you will be able to gain a much more accurate measurement of where residents access health care most often It s also important in questionnaire design to ask only one question at a time Instead of asking the respondent if they would evacuate in a large scale disaster and where they would go for shelter information all in one question first ask if the resoondent would or would not evacuate Then if they indicated that they would evacuate ask where they would turn fo for shelter information On
73. imilar to the example above you will need to supply a survey name and specify the state and date Then choose your cluster point and questionnaire file The spatial data has to be a kml with the fields specified on the webpage Click the link to download an example to get started Collect SMART requires that the Survey Name be in each of the files that you upload and the cluster point and questionnaire file be named in a specific manner For example if you were conducting a survey in Gaston County in 2015 you might enter Gaston 2015 for the survey name Then the cluster name would need to be named clusters Gaston 2015 your points need to be names points Gaston 2015 and the questionnaire should be named survey Gaston 2015 Implementing the survey Page 31 HOME CONTACT US LOGOUT Upload New Survey Please complete all the following fields then hit Upload Files button e Enter New Survey Name up to 64 characters e Select Survey State Please Choose v Enter Survey Date e Cluster File Choose File No file chosen Cluster file needs to be a kmi formated file with the pat ay fields OCCUPIED SAMPLES CLUSTER TEAM where CLUSTER is the cluster number usually 1 30 not the fips code You can download AN EXAMPLE CLUSTER KML FILE HERE e Points File Choose File No file chosen The point file also needs to be a kmi format file with the following fields CLUSTER PNT NUMB LABEL POINT X POINT Y where LA
74. in that time is about 40 60 questions total Creating Questions The way questions are formatted and phrased and the type of responses the questions solicit impacts how respondents interoret and respond to the questions You want all resoondents to interoret a question the same way and you want them to be able to easily answer the questions If questions or response options are too long complex or difficult to interoret you may not get the data that you need Question design also impacts the type of data you will collect and how you will be able to analyze data Formatting of the questionnaire itself can impact the design of an on screen data entry form and or analysis of variables in your software program Questions should be specific and simple For specific questions avoid ambiguity in question wording For example if you ask Did you ever have problems getting the health care you need you could get a variety of responses when the respondent was a child when the respondent lived in another community etc A better way to ask the question Is to be specific about the timeframe In the past 12 months Try to write questions as simply as possible The comprehension level should be appropriate for the community Instead of asking if the respondent if they ve ever been told by a health professional that they have hypertension use the term high blood pressure instead A question that is designed to be reliable will a
75. in the form of a pocket guide for teams who run into glitches out in the field An example Is provided in Appendix H 8 Instructions for non English speaking households In neighborhoods where you anticipate encountering non English speaking households be sure to have some materials translated into the language you are most likely to encounter You can use U S Census data to determine what languages are spoken in your selected census blocks One suggestion Is to have a card in the appropriate languages that asks the resident to fill in their phone number if they wish to be surveyed it can be arranged at a later time to survey the household by phone with the appropriate translator 9 Laminated answer choices sheets in English and Spanish for questions that have a long list of responses usually more than 5 possible responses When the interviewer reads the question they can hand over the response list to the respondent to read over the choices 10 Educational materials and services offered by your agency or sponsoring organization Don t overwhelm households with too much information and unwanted literature A checklist of items for team packets Is included in Appendix I Tracking Forms When interview teams are working in an assigned cluster they need to document each household they reach beyond just those where an interview was conducted It is important fo document unoccupied households those where no one eligible over 18 and actual r
76. ing the survey Page 26 If the CASPER is being conducted a part of a Community Health Needs Assessment it is important to note that according to the Affordable Care Act the Treasury and IRS expect non profit hospitals to assess communities defined by a geographic location or service area e g a particular city county or metropolitan region However in some cases the definition of a hospital tacility s community may also take into account target populations served e g children women or the aged and or the hospital facility s principal functions e g focus on a particular specialty area or targeted disease More information can be found at the IRS website Primary Sampling Units Once you have selected your sampling frame you will need to determine your primary sampling unit PSU the cluster Cluster sampling can be conducted with any number of geographic delineations so long as they don t overlap i e are mutually exclusive Ideally these clusters are also geographically exhaustive meaning that they cover the entire area within your sampling frame The ideal data set for cluster sampling is U S Census Data since It is exclusive and exhaustive Another advantage of the census data is that it is nested at different sizes This means you could use Census Tracts Census Block Groups or Census Blocks Census Tracts can have anywhere from 1 500 to 8 000 people in them and the optimum size is 4 000 Block groups can have anywhere
77. keep in mind that teams should only be conducting interviews with residents who live in the selected cluster This means avoiding industrial commercial or any non residential neighborhood This can be identified ahead of time using tax parcel data building point or address data or through Interpretation of aerial imagery Instruct teams to avoid these areas and focus on residential neighborhoods or addresses In the case of multiple family units such as single high rise buildings or condos first randomly select a floor in the building by using a random number generator For example in a 5 story building randomly select a number between 1 and 5 Once on that floor count the number of units on that floor and randomly select the first unit to interview Similarly in the case of multiple apartment buildings within a complex if there are units A E assign the numbers 1 5 to the units and randomly choose a number between I and 5 Alternatively survey teams could be instructed to contact the base station with an estimate of the number of floors and number of units per floor Two random numbers should be then be chosen one for the level and one for the unit The teams should start at this random location and then proceed in a census like manner until an interview is completed If you are sampling in 2010 U S Census blocks teams should be careful to avoid the donut holes in which blocks are nested within blocks In the example shown
78. lude topics such as list topics Surveyors will be in your neighborhood Insert Dates 10 00 a m 6 00 p m To ensure the safety of residents all surveyors can be identified by a name badge For more information contact Office phone number web address Appendix Page 75 Appendix F Base Station Layout PR ample of CASPER Strike Team Operations Room Laydiif Sign in Sign Out Table 1 Equipment GIS radios Table 4 4 Health Education Materials Table 3 Water Food Supplies Agency Responder Kits Table 2 Appendix General Assembly Area Chairs OOOO DO OOOOO0 nana OOOO0 OOOO0 OOOUOQ Questionnaire Tracking Sheet and Referral Sheet Review Table 5 EPI DESK Data Entry and Laptop Computers Station Large Reference Maps on Walls Printers and Photocopier Station Page 76 Appendix G Tracking Form Template Cluster Assessment Tracking Form Date Interview Team Status Codes IC Interview Conducted NE Not Eligible IR Interview Refused NO Not Occupied No One Home LB Language Barrier Appendix Page 77 Appendix H Collect SMART Pocket Guide Google Android Tablet Turning it On On the side of the tablet you will see the power lock key to turn it on and the volume key If you cannot get the tablet to power on hold the power key for 5 seconds SIDE Microphone Power lock key Volume key Lock Cod
79. method can also be used to assess public health needs in non disaster scenarios such as community health needs assessments required for non profit hospitals by the Patient Protection and Affordable Care as well as to answer other research or preparedness questions in non disaster settings While CASPER has proven to be very effective for assessing health needs in a number of different settings it still requires expertise in sampling access to expensive GIS and statistical software and often it requires extensive technical support from the CDC state health departments or academic partners Collect SMART overcomes some of these technical barriers making the CASPER sampling method more accessible to a broader audience This sampling method is a type of rapid needs assessment that was originally developed in the 1960s to assess Immunization coverage and was refined to assess health needs following a disaster According to the CDC CASPER is a specific set of tools designed to quickly obtain a snapshot of health needs and concerns for decision makers Using valid sampling methods CASPER divides populations of interest into randomly selected geographically distinct clusters such as census blocks Within the clusters a random sample of households is selected to be interviewed This method has been validated for rapid assessments and the estimation of population level health needs producing valid and precise estimates of 10 Advantages of C
80. ndix I Team Packet Checklist Team Packet Checklist One packet per team Item Description nad p in Packet 1 Contact List of all team member s cell phone numbers field numbers post staff phone numbers 2 Training slides Copy of training slides for reference 4 Tracking forms 2 tracking forms per cluster 5 Field maps Overview map of entire area plus a map for each Cluster with cluster boundary and survey start sites 6 Technology Pocket guide for using tablet technology 3 Paper copies of At least one paper copy for interviewer include questionnaire handful as back up cheat sheet 7 Instructions Instructions in languages likely to encounter on how handouts for non those households can participate in the survey English households 8 Survey answer Laminated answer choices English and Spanish for choices questions with more than 5 response categories 9 Informational Educational materials or services offered by handouts sponsoring organizations Appendix Page 80 Appendix J Collect SMART Data Collection FAQ The random start point house has a no trespassing sign What do we do Write the address down and note that on the tracking form Go to the next closest houses until you get an interview No one is home at our random start point Do we go to the next random start point Write the address down and note NO on the tracking form Go to the next closest houses until you get an interview Once you get
81. ned for logic errors or logically impossible combinations of responses You can often detect errors in data simply by looking at each record to see if the responses are logical For example a question was asked about current pregnancy and the reply is marked yes but you notice that the respondent is male There are some software programs that also have the capability to incorporate automated logic checks so that improbable values are flagged for the investigator to examine The simplest and best way to correct outliers or improbable values is to look at the original data source the questionnaire and compare it with the database to check for possible data entry errors If this is the case you simply correct the data with the true value If this is not the case or if you do not have access to the original data source you have several options One option would be to recode the value as missing and then deal with it the same way you handled other missing data You could also choose to delete the record from the analysis Or you may decide to leave it as it is because some outliers are not errors and often true outliers can provide some important information on the characteristics of people at the extremes of the distribution Overall knowing how to deal with outliers and improbable values will depend on your knowledge of the subject area and how important the variable is fo your analysis Weighted Analysis CASPER data can be analyzed in n
82. nfo for your analysis the database will automatically exclude records with the same ID The solution for dealing with data duplications is simple first make sure that both records contain the exact same information and then delete the duplicate record Analyzing and reporting results Page 64 Outliers or Improbable Values Outliers and improbable values are also common types of potential errors that can be found in data Outliers are values that fall outside the range of values of the majority of the responses Outliers are often defined statistically such as values that fall outside 2 standard deviations of the mean Improbable values are values that are highly unlikely or impossible Often they are identified when you discover logically impossible combinations of resoonses such as answers for questions that should have been skipped Outliers are most offen detected by examining descriptive statistics for each variable including the minimum and maximum values measures of central tendency and frequency distributions which are represented graphically by histograms Variables based on continuous data such as age usually have most of the values clustered around the mean so if you see one or two people who are very old or very young compared fo the rest of the study population these people might be considered outliers However outliers in this case should be kept with the database and not deleted Wherever possible data should be exami
83. ning and adequate and easily accessible parking for survey team staff or volunteers Examples of base station venues include the health department community centers hotels and libraries gt Example of CASPER Strike Team Operations Room Layoiit Operations and technical assistance usually provided by the strike team leader GIS specialist or epidemiology specialist ma should always be available for the survey team through the _ General Assembly Ares base station Keep that in mind when booking a room the m space will be needed not just for the training but to staff the OOOO0 base station operations and technical staff during the entirety 00000 of the survey onanan OOoO0d This example of a base station layout was prepared by the OOO00 Community Preparedness Section of the Texas Department of a OOOO0 State Health Services who routinely implement CASPERs after disasters using the ICS structure Appendix FJ Implementing the survey Page 42 Equipment Safety equipment includes vests or agency t shirts to maximize visibility in the field and every team member should wear identification badges with at least the agency logo and name Official county or employee ID badges can work too Surveyors should also carry their driver s license If two way radios are available consider using them as part of a preparedness exercise Cell phones are also sufficient in most areas but expect that some teams will be in areas with little or
84. of the screen e Responses tool Analysis of results can also be conducted on the fly or in real time The Response Tool shown on the left side of the image below tabulates responses as sync files are sent to the server Questions that are coded as check boxes yes no or multiple choice are tabulated in a basic table Open ended questions can t be summarized with this tool Implementing the survey Page 56 e Progress tool Shown on the right side of the screen below tracks the number of completed surveys in each cluster Both of these tools can be turned on and off using the button at the top of the screen and moved around and resized to fit anywhere on your screen 47 30 A 17 10 lt 117450 M Surveyor60 Surveyor5Y Surveyor61 Surveyor62 F1 Surveyor63 Surveyor64 Surveyor65 m Surveyor67 Surveyor68 Surveyor6 Surveyor70 Surveyor7 10 ood Hand Surveyor72 Surveyor73 Surveyor74 1A ow do you prefer to get information about 0 ID GASTON 2015 FINAL project 17 70 A Hide Responses tool Hide Progress tool Y Refresh points Hide Clusters your community Select al Responses 21 1 Through social media such as Facebook and Twitter 21 2 21 SEN 21 JER Through your house of worship 21 Through seminars workshops and classes 21 Through printed materials such as flyers 21 Downloading Records Software Tip If you are not se
85. om sampling According to the CDC s CASPER Toolkit this method involves selecting a random starting point by using a detailed map upon arriving at a given cluster such as the one below SOURCE CDC CASPER TOOLKIT SECOND EDITION A printed map can be used to count or estimate the number of housing units within the cluster That total can be divided by 7 which will be the N that can be used starting at the house nearest the randomly selected starting point to consistently count every Nth house Interviews should be conducted until every Nth house is complete Option 4 Random Start Points The fourth option will likely produce a sample with the least amount of bias as it requires less surveyor training and reduces the chances of surveyor error implementing the above methods In this method you must generate seven random points using mapping software It is recommended that staff with knowledge of geographic information systems GIS perform this task To increase the likelinood of success using this method consider accessing local address point data center line data or tax parcel data from the local GIS office This data is often much more accurate than national address databases and contain attribute information such as residence type that you can use to identify residential addresses only Assign an address to the points reverse geocoding and instruct surveyors to use these points as random start sites If no one is home at the point or
86. omeone to participate in an interview Participant Motivation Source Kahn RL amp Cannel CF 1968 Interviewing In Sills DL ed International encyclopedia of the social sciences 8 153 New York NY Crowell Collier and Macmillan Interviewing someone in person requires careful attention on the part of the interviewer Here are some tips that can be covered during a training session that can help improve response rates and open doors e Dress appropriately Business casual or professional and neutral Wear comfortable clothing and shoes Always wear an identification badge Refrain from eating drinking or chewing gum during an interview and remove sunglasses e Be neutral yet attentive Face the participant at a comfortable distance and maintain appropriate eye contact e Be accepting of the participant Do not express any judgment of the participant through your body language tone or facial expression e Concentrate on what the participant is saying Do not let your mind wander or interrupt the participant Try to not show a reaction if the respondent says something unexpected Implementing the survey Page 37 The Introduction When training interviewers on the introduction the first thing interviewers should do when approaching a resident is immediately identify yourself and affiliation Remind interviewers to make eye contact and smile and avoid reading from a script Also have interviewers practice their introd
87. onymous During the training stress that the interviewer should never give out details about the participant s home or what was said in the interview Once the interview is complete interviewers should thank participants and give them a thank you gift for their participation Past gifts that have been effective incentives for participation include first aid kits water bottles and reusable shopping bags This is also an opportunity for participants to receive any informational handouts about the agency conducting the survey services or the assessment project without overwhelming them with paper If the resoondent wants more information about the survey an interviewer should have a telephone number for him her to call Using the Mobile Application Interviewees will also need to be trained on the Collect SMART mobile application including how to open the app finding their cluster opening a point entering data and sending sync files to the Collect SMART cloud space See page 54 for instructions It is recommended that you demonstrate the features of the app using screen mirroring software first before you hand out the equipment so that you have the full attention of the interviewers Then hand out the tablets or smartphones and be sure to allow sufficient time for interviewers to become comfortable with the technology before sending them off in the field Team Management When you are considering pairing up your staff or volunteers into 1
88. oose 8 generate clusters to pull your sample HOME CONTACT US LOGOUT Create New Project Use this page to medi a random sample of clusters in a sampling frame specified by one of more counties or zipcodes Please Enter New Project Name up to 64 characters Please Enter Survey Date 2015 0617 _ Please select sites for survey from the following options Select County Counties Dare Coun Selected Region Population HUs Durham lt sess 267587 120217 i Enter the Number of Clusters you Select Zipcode s default 30 Select Zipcode s A 27006 E 27007 EE Require minimum of 7 Housing 27009 Units per cluster me use Block Groups To select multiple options JBM CLick HERE TO GENERATE CLUSTERS Hold down the control ctrl button and click For Mac Hold down the command button and click Davidson County Davie County Duplin Coun 27012 v Keep in mind that if you exclude clusters you are limiting the generalizability of your sample This needs to be balanced against your ability to collect enough interviews for a valid sample typically 80 of total selected In practice it is often difficult to conduct enough interviews in blocks with fewer than 10 housing units Your other option for using Collect SMART is to supply your own spatial data From the first New Project window choose Use this option if you have used a GIS site selection tool to create your sample S
89. or use volunteer teams will have to use their personal vehicles so consider a plan to reimburse them for mileage or provide gas cards Lastly it s also important to find a way to recognize the survey volunteers Include survey teams in the acknowledgements section of the survey report or presentation take a team photo to share in the report or agency website and provide volunteers with a certificate of acknowledgement Training Volunteers To ensure data quality and safety of teams it is important to conduct a 2 3 hour mandatory training for anyone who will be helping conduct the survey Even though the survey will likely being conducted on a fast timeline interviewer training is not an optional step that can be skipped for the sake of time A lack of training may lead to lower quality data which in turn may lead to data that is not useful for planning efforts particular as it pertains to Community Health Assessments Also lack of training on how fo use the survey equipment tablet GPS units or the survey instrument can lead to missing data or data that was completed incorrectly No one can control all the factors influencing community data collection but a good interviewer training process will help start things out on the right foot Training sessions can be hosted the day before or morning of the first day of surveying This training should ideally cover the following topics see link for interviewer training template e The purpose of
90. ost pressing health needs in the community Quote about importance of survey for the assessment process from CHA committee chair Hospital public health staff and volunteers will be conducting the surveys in survey area city county region on dates from approximately 10 00 a m 6 00 p m To ensure the safety of citizens surveyors will be wearing name badges Survey participants will be asked questions about list survey topics The Community Health Assessment is conducted every years to identify the health status concerns and resources in Name County to as a part of a community wide strategic health planning process The results will be used by health and service organizations to develop priorities and action plans for improving services Quote encouraging residents to participate and share their opinions from lead agency director List project partners are providing support for this project For more information on the Community Health Assessment contact phone number or visit website HH Appendix Page 74 Appendix E Community Flyer for Data Collection HELP IMPROVE THE HEALTH OF YOUR COMMUNITY NAME COUNTY HEALTH SURVEY e Name County is partnering with a number of local organizations to conduct a Community Health Assessment e Surveyors will be conducting randomly selected door to door surveys to get resident s opinions on the most pressing health needs in the community e Survey questions inc
91. re contact was ATTEMPTED who successfully completed an interview Contact Rate of Completed Interviews All household units where contact was attempted Analyzing and reporting results Page 62 The denominator includes all completed interviews incomplete interviews refusals and non responses including where a home was abandoned or unoccupied You want your contact rate to be above 20 The contact rate can vary greatly particularly if a CASPER is being conducted after a disaster or emergency If damage impact is extensive you will likely experience very low contact rates Contact rates are offen used as a measure of the amount of effort involved in carrying out the survey Data Cleaning Once data have been entered into whatever platform will be used to analyze it must be thoroughly reviewed or cleaned to check for inconsistencies implausible entries missing data and skip pattern errors Before the cleaning begins always save a new copy of the database file so that any errors that may occur during cleaning and analysis do not corrupt the original raw data Your goal during data cleaning Is to prepare the raw data for analysis A thorough review of all of these considerations will improve the quality of your results and reduce problems down the line In addition to dealing with possible errors and missing data data cleaning can also include changes to the data to prepare for weighted analysis For example some numeric answer
92. routes to be uploaded to ArcPad using networking algorithms Implementing the survey Page 29 If you are using ArcGIS 10 1 or 10 2 the CDC has a toolbox which is available as well This toobox automates sample selection in the first stage It is available upon direct request from the CDC s CASPER group Selecting using Collect SMART If you would like to generate a sample using Collect SMART you must first sign up for a free account HOME CONTACT US LOGIN Collect amp SMART Survey Management and Response Tools Then choose Create from the Home Page You have two options for using Collect SMART shown below Collect SMART HOME CONTACT US LOGOUT Welcome to Collect SMART The power of Collect SMART sampling is in its ability divide the population of interest into randomly selected geographically distinct clusters using U S Census blocks or block groups The Collect SMART sampling tool will draw a random sample with probability proportional to population based on the number of housing units in your sampling frame The sampling frame can be defined as a single county multiple counties or by zip code or multiple zip codes Within the clusters a random sample of households should be selected to be interviewed Refer to the User Guide for Project Managers on acceptable second stage sample selection methods This method has been validated for rapid assessments and the estimation of population level health needs producing
93. s for others e Always note the units for the response for each question such as select one select all that apply and choose up to 3 Implementing the survey Page 19 Finalizing the Questionnaire After the initial questions have been developed meet with stakeholders to prioritize and revise the questionnaire as necessary This is critical to ensure that stakeholders are invested in the project and have their key questions addressed in the survey Depending on the area being surveyed if may also be necessary to have the questionnaire translated in to other languages Lastly it s very important to pilot your questionnaire before finalizing the survey Test out your survey questions with staff to make sure it flows and makes sense and to also determine how long the survey takes Remember your goal should be to make the interview less than 20 minutes Ideally the interview should only take 10 15 minutes to avoid survey fatigue If the interview takes longer than that you risk respondents losing interest and not fully paying attention to your questions This can lead to lower quality data and can possibly introduce bias into your results Resoondents may also feel like they have answered enough of your questions and want to end the interview prematurely which results in missing data Below are important elements for you to evaluate as you pilot test the questionnaire Pilot Test Do respondents understand all wording
94. sals and non respondents Source CDC CASPER Toolkit Second Edition Implementing the survey Page 51 Going Door to Door Each interview team should have a list of randomly selected households 7 in the 30 7 sampling method including addresses and perhaps GPS coordinates These seven households are referred to as start sites or points The CDC recommends that the most scientific and representative way would be to select only 7 households and to continue returning to the selected 7 households until an interview is complete at each one If a household is selected and no one answers the door the team should plan to try again later in the day or even another day This will keep the contact rates low and reduce the chance that a convenience sample will be result Teams should attempt to revisit previously sampled households up to three times This is also the gold standard in survey research However this approach is much more expensive and time consuming and the need to obtain scientifically valid results needs to be balanced with the time constraints of completing the survey So some replacement households will likely need to be found if surveyors are met with empty households or with individuals who are not appropriate candidates to complete the survey e g someone under 18 someone who Is not a permanent resident of the home like a babysitter etc It is also known in survey research that replacements from within the
95. same cluster will be similar to each other The most realistic method is to simply try the next closest house continue on to the next closest home until you complete a survey In densely populated areas this should be easy enough to find rural locations may require some assistance from a map or aerial imagery Teams who are unsure of what to do should contact base station for guidance Be sure the team notes their movements and the resoonses met at each household on the tracking form By using the randomly selected start site the home where the interview ultimately takes place will still be random using the method described above This method does have the potential to bias results slightly towards the elderly and female population who tend to be home more often Remember the goal is to keep contact rates above 20 in order to avoid bias in your survey results In small clusters it is possible to exhaust all of the homes In the example shown here the cluster only has about 40 total A p SUL homes to choose from A survey team A NU i se pe Ta R ERAUR DR do a might start on point 7 1 and end up at 7 2 if no one is home or agrees to the i interview If this happens then the Sines A Plan teams are effectively conducting a j fe census This is fine teams should be instructed to continue in a census like manner knocking on the door of each adjacent house until they conduct the two interviews for 7 1 and 7 2 UT g
96. se the teams came to collecting the goal number of surveys 210 usually also called the success rate YoU want completion rates over 80 or your survey will not be representative of the sampling frame Cooperation Rate The cooperation rate is often what people think of when they hear response rate This rate describes how eligible and willing the community was to participate in the CASPER The lower this rate is the more the sample leans toward convenience and away from random and indicates the possibility of bias Higher cooperation rates will mean a survey that is better representative of the sampling frame Aim for cooperation rates above 60 Contact Rate The contact rate describes essentially the effort of the survey the proportion of alll households visited where a survey was completed Try to keep your contact rates above 20 one house in every five attempted completes a survey If your contact rates are below 20 you will need to interpret your results carefully as the demographics of your sample may not align well with the demographics of your sampling frame Number of completed Interviews Completion rate Number of interviews goal usually 210 Number of completed interviews All HUs where contact was made including completed interviews incomplete interviews and refusals OE Number of completed interviews ARE Number of HUs where contact was attempted including completed interviews incomplete interviews refu
97. ss on the screen in a white callout box Tapping on the callout box will bring up a prompt to either begin an interview for that random start point or cancel When a user clicks begin interview the record list will come up and the form will be loaded Instruct interviewers to be patient at this step as sometimes the forms are large and take up to 5 seconds to load A Map View GASTON 2015 FINAL Survey info EPI Info Survey Loading Form for Cluster Point 3 3 Begin a Survey Interview for Point 10 3 or Cancel Please wait 10 1 1121 W DAVIDSON AVE GASTONIA COMPLETED 10 2 1041 RIDGE AVE GASTONIA COMPLETED 10 3 1022 HOLLAND AVE GASTONIA NOT STARTED 10 4 605 N RANSOM ST GASTONIA COMPLETED 10 5 401 APT N HIGHLAND S GASTONIA COMPLETED 10 6 405 APT N HIGHLAND S GASTONIA COMPLETED lt O o Once the form is loaded give interviewers time to scan through it and compare this digital version of the survey to the paper one If you have built in code to capture the date and time and exact location of the survey be sure to tell interviews how this code is activated In Epi Info this code can be called up most easily with a button or simply upon loading the form Keep in mind that a button will allow the surveyors to adjust the time and coordinates if they end up pulling the survey up early and then knock on multiple doors before actually beginning the interview Implementing the survey Page 55 info R
98. ssure that the words are interoreted the same way in any setting and that resoondents answer the same way In any setting To assure that your questions are reliable avoid ambiguous terms or words that might be interpreted differently by different people If you ask if the respondent has had a routine health check up in the past year you may need to define what you mean by a routine health check up Does that include OB GYN check ups Or if you ask a resoondent if they are caring for a friend or family member with a disability define what you mean by disability Closed ended questions with pre determined categorical resoonse options may provide more reliable responses than an open ended question which can leave room for misinterpretation of the question It s also important to consider the answer choices for your questions Limit lists to only what is necessary and analyzable Consider how data will be analyzed to help determine what variables are Implementing the survey Page 17 needed and thus what questions should be included If you are asking residents to list their 3 top concerns about the community and there are 20 possible answers be sure to provide a laminated card of the choices so that the participant can look them over which saves time from the interviewer reading them all out Always offer to read the answer choices though since you should not assume the participant is literate Here is an example that shows val
99. ster From this list you then randomly select seven housing units to conduct interviews Alternately you could choose one random house to begin with and then visit every nth housing unit e g house 1 5 10 15 According to the CDC s CASPER Toolkit while this method does ensure that each household has an equal chance of being selected and is least likely produce a sample with bias it may be less feasible and inefficient due to the difficulty of training volunteers and the time needed to count all housing units in a selected cluster SOURCE CDC CASPER TOOLKIT FIRST EDITION Option 2 EPI Method When a census of your cluster is not available one method you can use to make sure you are surveying a representative sample of the affected population is to First go to the center of the selected cluster selected in Stage 1 Then randomly choose a direction by spinning a bottle pencil or pen on the ground and note the direction in which it is pointed when it stops Finally walk in the direction indicated by the object and select the first house closest to the point where you are standing Proceed in that direction until you have completed 7 interviews of occupied households This is called the EPl Method and Is the easiest to implement if the center of the cluster can be readily located SOURCE CDC CASPER TOOLKIT FIRST EDITION Implementing the survey Page 33 Option 3 Systematic Random Sampling Another option is systematic rand
100. stion that has already been validated that would assess adult diabetes prevalence 2 One of your partners came up with their own question to assess diabetes What is wrong with the following question Has a health professional EVER told you that you had diabetes and did not refer you to nutritional counseling Yes No GaSe SS a ee A ee ee Se ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee EEE Se Se ee eS ee ee ee ee ee ee See See ee Implementing the survey Page 22 ACTIVITY 2 1 DEVELOPING A QUESTIONNAIRE ANSWER KEY 1 Has a doctor nurse or other health professional EVER told you that you had diabetes Yes Yes but female told only during pregnancy No No pre diabetes or borderline diabetes Don t know Not sure Refused Source 2014 BRFSS survey 2 The question is actually asking 2 questions Also the responses should include an option for don t know and refused Implementing the survey Page 23 Survey Site Selection Collect SMART uses CASPER s two stage cluster sample method This sampling method is preferred as an alternative to simple random sampling and convenience sampling By staging your selection in two steps first by soecific geographic areas you can reduce travel and administrative costs by working In clusters versus random points soread out all over the your entire sample frame To achieve similar results in a convenience
101. tance including providing for basic needs such as food water and shelter c To determine the health status and basic needs of an affected community d To identify individuals who qualify for public health programs Answer C 2 How are households chosen to participate in a CASPER survey a Simple random sampling in the geographic area of interest 7 Two stage cluster sampling starting with Census blocks 2 Convenience sampling of nearest households All households in the geographic area of interest are included Answer B 3 Which of the following is NOT a potential advantage to using CASPER over another data collection method a Low cost O High quality data 2 Few technical requirements Rapid results Answer C Table of Contents Page 12 Chapter 2 Preparing for your CASPER Preparing for your CASPER Page 13 CHAPTER 2 PREPARING FOR YOUR CASPER Introduction This chapter describes the basic steps necessary to conduct a CASPER for project managers Although the steps are presented here sequentially some may overlap or occur in a different order depending on the situation Basic steps of conducting a CASPER 1 Gather your team 2 Develop a questionnaire 3 Select the sample 4 Recruit and train interview teams 5 Conduct interviews 6 Analyze and report Staffing The first step in preparing for your CASPER Is to get your team in place Suggested staff include an ep
102. there will be no shortage of eligible participants Selecting the Clusters Stage 1 Now that your sample frame Is defined you are ready to draw the survey sample There are three methods for selecting your clusters 1 selecting by hand using a spreadsheet 2 using Geographic Information Systems GIS software or 3 using Collect SMART s website If you are using your own GIS software you can find pre joined demographic data via TIGER Line Shapefiles or geodatabases Keep in mind that block data spatial and demographic Is only updated every 10 years so the most up to date census block data is from the 2010 Census Block group data is now updated as part of the American Community Survey on a yearly basis and is reported as a 3 5 year average Both blocks and block groups are now available from TIGER line pre joined with demographic data If you want to sample by hand you can download the tabular data from the US Census Bureau s American Fact Finder Selecting by Hand You can create a population weighted two stage cluster sample with Microsoft Excel or spreadsheet software The goal is to randomly select a sample of clusters using a probability proportional to the estimated number of housing units or total population in each cluster This is what is commonly referred to as PPS sampling probability proportional to size Most commonly a two stage cluster sample selects 30 clusters in the first stage and 7 interview locations in the second stage
103. they get an interview Two Stage Cluster Sample Stage ONE Clusters Population weighted random selection from geographically distinct clusters vy Stage TWO Points Within each randomly selected cluster select random interview locations The total sample size for this 30 by 7 cluster sample is 210 households The 30 by 7 sample is most common but other sizes are also an option For rural areas where many of the clusters have less than 10 households each you might consider conducting a 40 5 sample That would mean selecting 40 clusters and then 5 interviews per cluster Although you are interviewing 10 fewer residents you are actually conducting a more representative survey since you are visiting more clusters The 30 7 or 40 5 method of sampling around 200 households is adequate for getting results within 10 margin of error with 95 confidence For example the estimated prevalence of current smokers is 16 If the margin of error is 10 your survey would predict that the true result lies within 6 and 26 Another example with which you may be familiar most political polls mentioned in the news refer fo a margin of error 60 of voters said yes to Proposition Z with a margin of error of 5 Implementing the survey Page 24 If a data point is also given a confidence interval often 95 it means that if we repeated the sampling and interview process 100 times 95 of the time the true value would
104. u ever have a problem getting the health care you needed including filling a prescription or getting dental care Irrelevant Response I didn t vote for our mayor in the last election because I think he doesn t care about health Unclear Response In my life I ve always had problems getting health care When training interviewers be sure to cover how to probe The best probe is simply to repeat the question it is a neutral method that an interviewer can consistently rely on There are other neutral ohrases that can be used such as e What do you mean e How do you mean e Which would be closer e If you had to choose one which would you pick Probing versus Leading Probing versus Leading Example Example Interviewer About how long has it been since you last visited a doctorfora routine checkup This does Interviewer Probe correct Which would be closer 1 2 years ago 3 5 years ago or more than 5 years ago notinclude any times you visited the doctor because you were sick pregnant or for chronic disease a 1 2yearsago d I have never had a routine checkup b 3 5yearsago e Don t know c More than 5 years ago Interviewer Leading incorrect w z a DT Respondent So would you say 1 2 years ago Oh gee I didn t go very often maybe a few times In the example above the correct probe reiterates every possible response category It allows the respondent a bit more time to consider
105. uction using these tips e Introductions should be brief Interviewers should state the purpose of the survey and tell how the information will be used Itis important to state that information will be compiled with other responses from throughout the survey area and that their identity will remain anonymous Terms to avoid are research study and survey e The introduction should also define what is expected of the participant For example that participants can skip any questions they want and that participation is voluntary and will greatly help your organization understand what health concerns it should focus on and work to improve e Itis encouraged that interviewers be empathetic This is offen easier said than done but let interviewers know that with this point and all the others made thus far with practice and experience they will improve Below is an example of an appropriate introduction that an interviewer could use Hello My name is Matt and this is Kristen We are working volunteering with the Chatham County Health Department talking to people folks residents in your neighborhood about community health issues Do you have some time to speak with us today If the resident seems hesitant the following assurances could be offered Your answers are anonymous and will be kept completely confidential We will not record your name and your participation is voluntary We are asking for resident s opinions on a variety of pu
106. ulated in the previous step If you are using the Classic Analysis tools in Epi Info choose the complex sample frequencies option under Advanced Statistics Analyzing and reporting results Page 66 Frequency of All 7 Except Q30_ Insurance 12 ewercise In addition to weighting the data analysis you should also calculate 95 confidence intervals These confidence intervals will indicate the reliability of the weighted estimate Be sure to set the weight properly and assign the primary sampling unit to the CLUSTER variable This option is available in Analysis mode if you select Advanced Statistics and then Complex Sample Frequencies The demonstration video below will walk you through other analysis options within Epi Into M that you might want to explore as you analyze your dataset im Demonstration Video Epi Into M 7 Analyze Data 5 minutes Analyzing and reporting results Page 67 Reporting When it comes time to report your CASPER findings first decide who will benefit most from the results You will want to present the survey findings to your partners participants and other key stakeholders Keep the reporting brief with a focus on elements that are actionable Based on your target audience be sure to tailor the presentation or report to that audience For example when presenting findings to community members you will want to explain the process in simple non technical terms At a minimum you
107. ult for interview teams to remember fo fill out these forms especially while out on foot but emphasize their importance to the teams this tracking is essential for determining response rates and ensuring that teams finishing up clusters have the information they need to approach the right households Implementing the survey Page 46 ACTIVITY 2 3 FORMING A SURVEY TEAM Time 5 minutes Instructions Match the possible CASPER staff roles with their definition Compare your answers to the answer key on the following page A Strike Team Leader Responsible for conducting safely household interviews within assigned clusters B Assistant Strike Team Leader _ Responsible for random sampling process creating and printing maps assigning teams to clusters and assigning equipment C GIS Specialist Responsible for overseeing 5 interview teams of two people each approximately 10 people D Epidemiology Specialist ___ Assists the Strike Team leader E Interview Team Leads Responsible for overall survey collection organizing and directing the field command post and overseeing the tactical assignments given to the Strike Team F Interviewers ___ Responsible for data collection data analysis of survey results and report writing ge ga E E 5 1 1 2 10 1 2 S ga 5 2 i 2 2 Ea ng ll na ja EE ea SE Ed Eb E m Implementing the survey Page 47 ACTIVITY 2 3 FORMING A SURVEY TEAM ANSWER KEY Match the possible C
108. umerous software platforms such as Epi Info SAS or other statistical analysis programs Epi Into M is a useful program for producing statistics particularly for the weights that need to be calculated SAS and other statistical analysis programs are often more complex but they allow for advanced statistical techniques In general advanced statistical analysis programs are only recommended if your analysis team is already familiar with the software In addition Excel or any other spreadsheet software can be used to present the results in visually pleasing tables and graphs A properly conducted weighted frequency analysis should yield reasonably valid and precise estimates of plus or minus 10 of the proportion of the affected population Additional guidance for proper analysis of CDC s CASPER tool in Epi Info can be found on the CDC website and in the CASPER toolkit some of which is described below Unlike a simple random sample of the entire sampling frame area households selected in cluster sampling have an unequal probability of selection To avoid biased estimates all data analyses Analyzing and reporting results Page 65 should include a mathematical weight for probability of selection A weight must be added to each of the surveyed households using the formula shown here Applying the weights when running analyses provides estimates that can be generalized to every housing unit in the geographic boundary for which the CASPER is
109. urvey Page 58 Day of Preparations Below is a summary steps that the project manager can do for day of data collection preparations 1 Check Team Packets binders should be assigned team numbers e Verify contents tracking forms maps point lists e Include 5 7 paper copies of questionnaire e Include laminated answer choices if using e Provide 10 15 Spanish referral forms 2 Tablet Preparations label each tablet by team number e Ensure tablets are fully charged e Turn all tablets on e Open Collect SMART application and ensure proper login information is stored in project e Verify the following mobile device tablet settings if applicable below is an example using a Google Android tablet Mobile data is on GPS is OFF Wi Fi and Bluetooth settings are OFF Display is set to auto Battery saver on battery options battery saver Battery fully charged Clean screens from any smudges e Ensure each team has a car charger and USB cord for each mobile device tablet label by team number 3 Match tablet chargers and team packets by team number 4 If providing include GPS units including charger and windshield mount 5 Provide incentive boxes for each team include at least 7 12 incentive bags in each box 6 Verify team pairings and location assignments 7 Brief teams before they leave e Remind them of check in procedures and safety including weather forecast e Reminders on questionnaire edits
110. urvey counts looking at synched files and Team Packets that have the maps and tracking forms o On the Collect SMART website go to the Project View dashboard on the Collect SMART website You can review completed surveys in each cluster along with reviewing the tracking forms o Update counts for surveys to be complete the next day Review next day team assignments making sure to pair experienced inexperienced and local non local o Check in with staff or volunteers to ensure they will be there the following day Check Team Packets so that each team has at least 10 surveys needed for next day add clean up clusters that have remaining interviews to team assignments if necessary to make sure each team has enough interviews assigned Implementing the survey Page 60 Chapter 4 Analyzing and reporting results Analyzing and reporting results Page 61 CHAPTER 4 ANALYZING AND REPORTING RESULTS Introduction Once the surveys are collected the next stage Is to analyze and report the results Analysis of Tracking Forms In summarizing the main outcomes of your CASPER there are three different response rates to calculate which were discussed briefly earlier in this manual success rate also known as the completion rate cooperation rate and contact rate Response rates helo determine if sample adequately represents the study population For example if your rates are very low then you might not have a very representative sampl
111. valid and precise estimates of 10 Before you create your sample you will need to create your questionnaire using the desktop version of Epi Info 7 You can download that here NEW PROJECT Or if you only want to use Collect SMART to manage your survey and you have already pulled your sample go here and upload your spatial data and your questionnaire UPLOAD FILES SKY highways info You can use the selection algorithm built into the platform that randomly selects blocks or block groups using the most current U S Census data or you can supply your own spatial data First we ll describe the process of using Collect SMART to select your sample From the home page select new project and the page that says Welcome to Collect SMART choose new project again Before proceeding with the sample selection you should have your questionnaire designed in Epi Info and exported as an xml ready for upload to Collect SMART On the new project page see figure below you should complete the following 1 Assign your project a name and 2 survey date 3 4 define your sampling frame by choosing your state then choose either one county or multiple counties or zip codes 5 choose the number of clusters you would like to select 6 choose whether Implementing the survey Page 30 to use blocks or block groups 7 establish a threshold for the minimum number of housing units for a cluster to be included in the sampling frame Finally ch
112. your survey efficiency will depend on a number of factors The length of the survey instrument and the competency of your staff are the largest contributing factors It is certainly possible to finish in 2 days with 10 teams but it s always best to plan for a third clean up day where you can complete any last remaining surveys Keep in mind that some teams may struggle to only get 4 to 5 surveys a day If this is happening try to figure out what are contributing factors Do they lack confidence approaching strangers If so split the team up to make sure someone on the team is comfortable doing the interview Are they getting lost or wasting time trying to find survey locations Remind them how to use the field maps or data collection devices and give them the opportunity to ask questions If can helpful to do this away from the entire group since teams are sometimes embarrassed that they are not getting interviews when everyone else Is Finally if you are sending teams out only in the mornings on weekdays that might be another reason you are having low survey numbers Typically survey numbers are higher on the weekends and during the hours of 2 00 p m 6 00 p m on the weekdays Analyzing and reporting results Page 50 Response Rates The goal for a successful CASPER survey Is that it should be representative of the population within the sampling frame In order to determine whether the survey results are actually representative you will

Collect SMART User Manual for Project Managers

Contents

Download Pdf Manuals

Related Search

Related Contents