Chapter 8: Using the WinBUGS interface in MlwiN
Contents
1. 97 5 0 588 5000 actual iterations storing every iteration Effective Sample Size ESS 4197 Update Diagnostic Settings 2 Help We can repeat all of the above procedures for the intercept parameter and the two variances and we will see that we get similar results for all four parameters with both MLwiN and WinBUGS MLwiN WinBUGS Bo intercept 0 005 0 042 0 002 0 040 B slope 0 563 0 012 0 563 0 013 ou between schools variance 0 097 0 021 0 097 0 020 o residual variance 0 566 0 013 0 566 0 013 Time for 5 500 iterations 21 seconds 81 seconds So why have a WinBUGS interface The example we have just gone through will give similar results using both software packages and to use WinBUGS we have to move back and forth between the two packages Also the estimation engine in WinBUGS is slower so you may be asking yourself the above question The interface was written originally as a testing tool to confirm that when new types of models are programmed into MLwiN we get the same answers as WinBUGS We recommend that while this version of the MLwiN software is still in development you check that both packages give similar answers t distributed school residuals The main advantage of having a WinBUGS interface however is to allow models that have not yet been developed in MLwiN to be fitted using WinBUGS We will illustrate this by considering alternative distributions for the2. 0 371 5000 actual iterations storing every iteration Effective Sample Size ESS 185 Update Diagnostic Settings Pher The results from the two methods can be seen in the following table The worst mixing parameter is the intercept B1 and to get an effective sample size of 5000 will take 18 5000 876 176 17 minutes 3 seconds using AR sampling while Metropolis will take 7 5000 185 35 15 minutes 53 seconds This means that even though the AR sampling produces a less correlated chain Metropolis is sufficiently quicker to give an effective sample of 5 000 in slightly less time Note that Metropolis here has a long burn in due the adapting method Parameter ESS AR Metropolis ESS Metro Bi a 544 0 091 0 540 0 089 185 Bo 0 009 0 005 4658 0 009 0 005 1209 Su 0 273 0 091 1050 0 273 0 092 433 Time 194s 18 176 42s 7 35 So we see that here it appears that Metropolis sampling in MLwiN is marginally better for this random effects logistic regression model Browne and Draper 2000 showed similar results for a couple of random effects logistic regression models It is however not guaranteed that this will be true for all models Multilevel Poisson models which can also be fitted using either method seem to give highly correlated chains using Metropolis and they may be models that it makes more sense to fit using AR sampling in WinBUGS References Best N G Cowles M K and Vines S K 19
3. 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 Again BUGS borrows its notation from S plus using the convention c to represent a vector of observations Here we see the first two constants N and n2 that define the number of level 1 units and level 2 units followed by a list of the school identifier for each observation To load the data into BUGS we need to highlight the ist identifier at the start of the data list and click on the load data button in the specification window If this is successful the message data loaded will appear at the bottom of the screen Next we have to combine the data and model definition by clicking on the compile button Again if this operation is successful a message appears at the bottom of the screen this time stating that model compiled Finally as BUGS uses MCMC methods all unknown parameters will need starting values These are included in the initial values part of the file that for our example is as follows Initial values file list beta c 0 002391 0 563371 u2 c 0 373760 0 502043 0 503888 0 018131 0 240431 0 541395 0 379001 0 026173 0 135181 0 337021 0 179300 0 061863 0 149648 0 165593 0 182923 0 409983 0 172781 0 084464 0 011510 0 214462 0 244
4. 501 2000 4000 iteration As can be seen this parameter does not mix that well with large autocorrelation in the chain We can see from the summary statistics below that on this small sample of 5000 iterations we cannot reject the possibility of a heavy tailed distribution node mean sd MC error2 5 median 97 5 start sample df 64 29 35 78 4 19 9 757 62 94 129 5 501 5000 In order to investigate the potential of starting value dependence we started three chains with identical parameter starting values except for df which was set to 2 10 and 200 respectively for the 3 runs To do this in WinBUGS is fairly easy as on the specification window there is a num of chains box that we edit to 3 immediately after checking the model is syntactically correct Then we have to load the 3 sets of initial values but this simply involves editing the df 0 line of the initial value file before loading each set We will increase the burnin to 2000 by changing the beg box to 2001 on the sample window when we input the parameters we wish to monitor We will also increase the updates to 12000 so that we have a monitoring run of 10000 iterations after the burnin Pressing update the three sets of chains will be run concurrently and so this will take longer df chains 1 3 200 0 150 0 100 0 50 0 0 0 2001 5000 7500 10000 iteration If we look at plots of the 3 sets we see that although all 3 chains are heavily autocorrelated there is overlap sugg
5. in WinBUGS the fixed effects beta and all other vectors always start with index and not 0 so that there will probably not be direct correspondence between the MLwiN and WinBUGS indexing Next the code defines the random effects u2 as being Normally distributed before finally giving the priors for the fixed effects and the variances WinBUGS has two types of relationship distributional relationships that are described by the symbol and deterministic relationships that are described by the lt symbol which is also used in the S plus package Note that the normal distribution definition in WinBUGS dnorm has two parameters that are the mean and the precision NOT the variance hence the deterministic relationship used to calculated the variance The prior distributions are identical to those used in the MCMC options in MLwiN Before running a model in WinBUGS we first need to read in the particular elements of the model using the Specification window available from the Model menu After selecting the window containing the model by clicking on it clicking on the check model button should give the message model is syntactically correct at the bottom of the screen Next we need to load in the data for the model Due to the fact that the data is generally the largest part of the file generated by MLwiN it is included after the initial values For the tutorial bug example the data section begins as follows rhb ly ly bab EE S G a 2 2 2
6. median 97 5 start sample beta 1 0 002979 0 03995 0 002516 0 07465 0 004435 0 07912 501 5000 beta 2 0 5634 0 01264 1 997E 4 0 5388 0 5636 0 5882 501 5000 These results are similar to those obtained from MLwiN which is reassuring and we can also get similar results for the other parameters as shown below node mean sd MC error2 5 median 97 5 start sample sigma2 u2 0 09662 0 02019 3 256E 4 0 06415 0 09446 0 1426 501 5000 node mean sd MC error2 5 median 97 5 start sample sigma2 0 5661 0 0127 1 653E 4 0 5416 0 5662 0 5905 501 5000 We can also get trace plots and kernel density plots via the history and density buttons respectively Below we see the trace plot for the parameter beta 2 beta 2 0 625 0 64 0 575 0 55 0 525 0 5 501 2000 4000 iteration and the kernel plot beta 2 sample 5000 AS 0 5 0 525 0 575 0 6 Currently WinBUGS does not allow the smoothing parameter to be changed for the kernel plots so that they look rather crude but this will be changed in later versions WinBUGS currently produces limited summary statistics and plots itself Historically the plots and MCMC diagnostics were provided via a suite of S plus functions called CODA Best et al 1995 and WinBUGS also has the option to produce the input files that CODA requires MLwiN can also use these files to input the parameter chains from WinBUGS into columns in MLwiN Here we will only consider the fixed parameter vector beta
7. so select this parameter in the node box and press the coda button on the sample window This will produce two windows that are labelled CODA index which contains the variable names and CODA for chain I which contains the values for the parameter chains We will now save these files as text files by clicking on the respective windows and then choosing Save As from the File menu We will need to save the files in plain text txt format We will store the CODA index file as beta ind and the CODA for chain I file as beta out in the same directory as tutorial bug Note that these are the extensions that the classic BUGS used for these files but as we have selected the plain text format WinBUGS will add an additional txt to each filename and so the files are actually saved as beta ind txt and beta out txt Now back in MLwiN if you want to input the beta traces return to the BUGS options window that we used earlier available from the MCMC advanced options window BUGS Options oO x Saving Options Save current model in BUGS format Loading Options Input out file beta out txt Input ind file beta ind txt Start column number c300 7 ome Here we will need to modify the out and in file fields to beta out txt and beta ind txt respectively Note that if you did not put these files in the current directory you will have to include their full path names in the respective boxes The window should t
8. 016 0 435732 0 489244 0 209408 0 230472 0 023543 0 023121 0 610002 0 240626 0 158475 0 033280 0 006457 0 029589 0 137882 0 128634 0 181341 0 189077 0 153068 0 130317 0 234439 0 211543 0 092819 0 089927 0 247556 0 109729 0 352728 0 042628 0 045058 0 042845 0 302413 0 051373 0 381929 0 723313 0 547252 0 503474 0 009972 0 031894 0 1381 15 0 658368 0 225656 0 039551 0 054029 0 535641 0 087691 0 165765 tau 1 767625 tau u2 10 854517 This gives the estimates from the IGLS run for the fixed effects and precisions and an MLwiN RESI command for the initial values for u2 that are exactly what the MCMC routine in MLwiN uses as starting values To use these values in WinBUGS we need to highlight the ist identifier at the start of the initial values and click on the load inits button on the specification window This will then give the final message initial values loaded model initialized Note that WinBUGS will generate starting values for any parameters that have not explicitly been given starting values but here we have given all parameters starting values Note that using the IGLS estimates as starting values is generally more efficient than allowing WinBUGS to choose its own We are now ready to run the Gibbs sampler in the WinBUGS package Before we start the estimation engine we have to tell WinBUGS which parameters we wish to monitor We will choose the same parameters as MLwiN uses From the Inference menu selec
9. 1 and rename the columns accordingly We can now look at the diagnostics for the two parameters via the Column diagnostics window z Column Diagnostics OF x Column bean Apply Done P Her Select the Column diagnostics window from the Basic Statistics menu Select beta 1 from the Column pull down list Click on the Apply button The diagnostics will then appear as shown below Here we see that the effective sample size for B is 876 for this sample of size 5 000 due to the autocorrelation in the chain We can repeat this procedure for Bz and also we can save the chains for o and find its effective sample size The information for all these parameters will be summarized in the table at the end of this section MCMC diagnostics 2 ae a E 7 z T m a a x Accuracy Diagnostics Raftery Lewis quantile Nhat 9696 8441 when q 025 975 r 005 ands 95 Brooks Draper mean Nhat 7954 when k 2 sigfigs and alpha 05 Summary Statistics Column beta 1 posterior mean 0 544 0 0032 SD 0 091 mode 0 539 quantiles 2 5 0 739 5 0 699 50 0 542 95 0 395 97 5 0 370 5000 actual iterations storing every iteration Effective Sample Size ESS 876 Update Diagnostic Settings Pher We can now run the same model using Metropolis sampling in MLwiN The model should currently be set up and MCMC should already be selected and so all we need to do i
10. 95 CODA Convergence Diagnosis and Output Analysis software for Gibbs Sampler output Version 0 3 Cambridge Medical Research Council Biostatistics Unit Browne W J 2002 MCMC Estimation in MLwiN in preparation Browne W J and Draper D 2000 Implementation and performance issues in Bayesian and likelihood fitting of multilevel models Computational Statistics 15 391 420 Goldstein H 1986 Multilevel mixed response model analysis using iterative generalized least squares Biometrika 73 43 56 Rasbash J Browne W J Goldstein H Yang M Plewis I Healy M Woodhouse G Draper D Langford I and Lewis T 2000 A User s Guide to MLwiN Version 2 1 London Institute of Education University of London Seltzer M H 1993 Sensitivity Analysis for Fixed Effects in the Hierarchical Model A Gibbs Sampling Approach Journal of Educational Statistics 18 207 235 Spiegelhalter D J Thomas A and Best N G 2000 WinBUGS version 1 3 user manual Cambridge Medical Research Council Biostatistics Unit 20
11. Using the WinBUGS interface in MLwiN Dr William Browne Centre for Multilevel Modelling Institute of Education London 22 October 2001 Background to MLwiN MLwiN is a statistical software package designed to fit multilevel models and developed at the Institute of Education It is the latest in a line of multilevel modelling packages developed over the last 15 years and forerunners include ML2 ML3 and MLN It is produced by a team of academics at the Institute under the leadership of Professor Harvey Goldstein It is sold for a small fee and currently has a user community of around 3 000 users worldwide primarily academic researchers in the social and medical sciences The project team consists of 3 research officers Jon Rasbash who is the package s main programmer Min Yang who is in charge of user support and William Browne who is the programmer of the MCMC options in the package There are also two lecturers at the Institute Fiona Steele and Ian Plewis who are associated with the project and a group of project fellows from other institutions around the UK MLwiN has different aims and attracts a different user base from WinBUGS It consists of a user friendly Windows interface written in Visual Basic on top of fast estimation engines written in C It is designed to fit particular classes of statistical model and so is less ambitious than WinBUGS which attempts to fit almost all statistical models However more attention i
12. ations Pala Rs use Binomial denom z use te 9 OCons logit T cons 8 age Fonts Subs Name Add Term Estimates Nonlinear P Heip Clear Nota We will firstly run IGLS and set up the model for AR sampling in WinBUGS Click on the Start button Click on the Estimation Control button Click on the MCMC tab Click on the Advanced MCMC Options button on the Estimation Control window Click on the Bugs Options button on the Advanced MCMC Options window Click on the Save button Save the file as bang bug We now need to start the WinBUGS program and load the file bang bug as a text file from the directory it has been saved Note that again we will need to change the Files of type box to All files to see the file bang bug When the file is loaded the model definition will look as follows WINBUGS code generated from MLwiN program MODEL Definition model Level 1 definition for i in 1 N use i dbin p i denom i logit p i lt beta 1 cons i beta 2 age i u2 district i cons i Higher level definitions for j in 1 n2 u2 j dnorm 0 tau u2 Priors for fixed effects for k in 1 2 beta k dflat Priors for random terms tau u2 dgamma 0 001 0 001 sigma2 u2 lt 1 tau u2 Here we see that our response variable use is defined as Binomially distributed and we have the logit link function to the predictor
13. esting that sensitivity to the starting value is not a problem If we look at the summary statistics for the three chains combined we get node mean sd MC error2 5 median 97 5 start sample df 91 9 50 36 2 87 6 241 88 59 189 4 2001 30000 This summary information suggests that a very small degrees of freedom df parameter and hence an extremely heavy tailed distribution is not feasible but that a value of df of less than 10 is not out of the question As a sensitivity analysis we will instead try fitting a model where the degrees of freedom is assumed known and has value 8 which suggests a slightly long tailed distribution at level 2 Seltzer 1993 gives Gibbs sampling algorithms for exactly this scenario of a known df parameter We will need to simplify our model definition as follows WINBUGS code generated from MLwiN program MODEL Definition model Level 1 definition for i in 1 N normexam i dnorm mu i tau mu i lt beta 1 cons i beta 2 standirt i u2 schoolfi cons i Higher level definitions for j in 1 n2 u2 j dt 0 tau u2 df Priors for fixed effects for k in 1 2 beta k dflat Priors for random terms tau dgamma 0 001 0 001 sigma2 lt 1 tau tau u2 dgamma 0 001 0 001 sigma2 u2 lt 1 tau u2 df lt 8 and we will also need to remove the initial values for parameter df as it is now a constant Again we will monitor beta sigma2 and sig
14. g and if WinBUGS needs to use Metropolis Hastings sampling whether to adapt There is also the option to use a technique called over relaxation that improves the mixing of the MCMC chains but takes longer per iteration The current version of WinBUGS 1 3 will choose the MCMC routines it uses for you depending on the form of the conditional distribution see section 1 3 of the WinBUGS manual for details Now that we have set the number of iterations press the update button to start the sampler After a few minutes depending on the speed of your processor and how many traces you are viewing the update counter will reach 5500 and the sampling will be finished WinBUGS has the nice feature that it will give you a message at the bottom of the screen for example updates took 88s stating how long the sampling took which is useful for comparing model run times etc Generally WinBUGS is slower for models that can also be run in MLwiN but as we will see later it has greater flexibility in the models it can fit and sometimes the MCMC methods it uses are more efficient than the Metropolis Hastings methods used in MLwiN for binomial and Poisson response models Once the sampling has finished we can now look at the estimates plots and other information again via the sample window To get summary information select beta in the node box and click on the stats button A node statistics window will appear giving the following node mean sd MC error2 5
15. hen look as above Pressing the Input data button will now load the chains into columns c300 and c301 To confirm this bring up the Names window from the Data Manipulation window and scroll down to c300 and you will see the following S Names OF x Refresh Categories P Help 0 1503 0 1601 0 5235 0 6045 beta 1 0 0 0 0 0 0 0 beta 2 e302 c303 c304 c305 c306 c307 c308 c309 c310 c311 e312 c313 c314 e315 c316 c317 c318 c319 e320 e321 e322 c323 ana We can now use the MLwiN MCMC diagnostics on the BUGS output for example for the slope parameter Select the Column Diagnostics window from the Basic Statistics window Select the column labelled beta 2 The window should then look as follows Column Diagnostics OF x Column beaa Apply Done Pher Note that this parameter is the fixed effect for the slope that is labelled B in MLwiN Clicking on Apply will give the following diagnostics screen which is very similar to that given by the MLwiN MCMC sampler in earlier chapters SE MCMC diagnostics parameter kernel density sored ipte Accuracy Diagnostics Raftery Lewis quantile Nhat 3957 3577 when q 025 975 r 005 and s 95 Brooks Draper mean Nhat 30 when k 2 sigfigs and alpha 05 Summary Statistics Column beta 2 posterior mean 0 564 0 0002 SD 0 010 mode 0 564 quantiles 2 5 0 539 5 0 542 50 0 564 95 0 584
16. iN are fairly new and we currently fit only models of particular types although we are constantly extending the number of models that can be fitted If however a user wishes to fit a model that cannot be currently fitted for example fitting an alternative distribution for the school level random effects there are three main options Firstly wait for a later version of MLwiN that will fit their model secondly write their own code to fit their model or thirdly try an alternative software package for example WinBUGS WinBUGS Spiegelhalter et al 2000 freely available from http www mrc bsu cam ac uk bugs in its earlier guise of BUGS was one of the first Bayesian software packages and is a more general purpose Bayesian estimation engine than the MCMC engine in MLwiN They work on a different philosophy of fitting models that can be represented by directed acyclic graphs DAGs BUGS has a compiled language which allows the user to specify their model through statements of two types logical and distributional which between them describe the structure of the DAG and hence the model Then BUGS compiles this user code and constructs an MCMC estimation engine for the user s model that can be run to give chains of estimates in a similar way to the MLwiN engine We will firstly consider a normal response variance components model and show how to fit this model in WinBUGS We will then go on to consider a model that MLwiN cannot fit which has
17. igma2 u2 lt 1 tau u2 df dunif 2 200 We will also need to give a starting value for df in the initial values file and so we will choose arbitrarily df 10 Our initial values file then looks as follows Initial values file list beta c 0 002391 0 563371 u2 c 0 373760 0 502043 0 503888 0 018131 0 240431 0 541395 0 379001 0 026173 0 135181 0 337021 0 179300 0 061863 0 149648 0 165593 0 182923 0 409983 0 172781 0 084464 0 011510 0 214462 0 244016 0 435732 0 489244 0 209408 0 230472 0 023543 0 023121 0 610002 0 240626 0 158475 0 033280 0 006457 0 029589 0 137882 0 128634 0 181341 0 189077 0 153068 0 130317 0 234439 0 211543 0 092819 0 089927 0 247556 0 109729 0 352728 0 042628 0 045058 0 042845 0 302413 0 051373 0 381929 0 723313 0 547252 0 503474 0 009972 0 031894 0 1381 15 0 658368 0 225656 0 039551 0 054029 0 535641 0 087691 0 165765 tau 1 767625 tau u2 10 854517 df 10 This time we will monitor the same four parameters as before plus the df parameter which we will set in the sample window Note that the adapting box on the update window is ticked because for this model WinBUGS needs to use Metropolis Hastings to update the df parameter Note that the tick disappears when adapting has finished We again run for 5000 iterations after a burnin of 500 iterations and get the following trace for df df 150 0 100 0 50 0 0 0
18. ma2 u2 After running for 5000 iterations after a burnin of 500 we get the following estimates node mean sd MC error2 5 median 97 5 start sample beta 1 0 001065 0 04125 0 002364 0 08322 7 714E 40 07792 501 5000 beta 2 0 5633 0 0124 1 979E 4 0 5391 0 5636 0 5872 501 5000 sigma2 u2 0 07657 0 01805 4 247E 4 0 04746 0 0745 0 1171 501 5000 sigma2 0 5662 0 01279 1 787E 4 0 5421 0 5659 0 5916 501 5000 14 Here we see that the fixed effects and level 1 variance are little changed in terms of point estimate and standard errors suggesting the analysis is robust to different level 2 distributions The level 2 variance parameter is not directly comparable as the variance of the t distribution is a function of both the sigma2 u2 and df parameters In fact the variance is 8 6 0 07657 0 102 which is slightly higher than for the Normal case Using WinBUGS on a Binary response model We will now consider a second example this time from demography We are interested in the use of contraceptives by women in Bangladesh Our dataset consists of 1934 women who are grouped into 60 districts and we will consider just one predictor the age of the women Multilevel binary response models are interesting in that we do not have conjugate priors for all parameters and the two packages use different approaches MLwiN uses random walk Metropolis sampling for these parameters whilst WinBUGS uses adaptive rejection AR sampling Our model is as follows S Equ
19. ndow from the Inference menu Change the beg value to 501 Type beta in the node box Click on the Set button Type sigma2 u2 in the node box Click on the Set button We have now set up the burnin of 500 iterations and the parameters we wish to store to run the updates we need to use the update window Select the Update window from the Model menu Change the updates value to 5500 Click on the update button We now need to wait a few minutes for WinBUGS to run On a Pentium 733Mhz machine the updates take 194 seconds In order to measure the efficiency of the sampler we will import the chains back into MLwiN and look at the effective sample size ESS measure To do this we need to do the following Select the Samples window Select beta from the node pull down list Click on the Coda button Click on the window labeled Coda Index Choose Save As from the File menu and choose plain text txt format Save file as beta ind Click on the window labeled Coda for chain 1 Choose Save As from the File menu and choose plain text txt format Save file as beta out Having saved the two files for the fixed effects we can return to MLwiN and use the BUGS Options window we used earlier to input the data Change Input out file to beta out txt Change Input ind file to beta ind txt Click on Input Data button This will store beta in column c300 and beta 2 in column c30
20. plers available in MLwiN and looks as follows SE Advanced MCMC Options OF Xx Estimation Method m Fixed Effects Gibbs C Univariate MH C Multivariate MH Random Effects Residuals Gibbs C Univariate MH C Multivariate MH m Higher Level Variance Matrices Updated by Gibbs sampling Level 1 Variance Updated by Gibbs sampling Random Number Seed 1 Haphe eg m Metropolis Hastings settings Scale factor for proposal variances 5 8 block dimension Use adaptive method v Desired acceptance rate 50 Desired tolerance 10 MH Cycles per Gibbs iteration 1 r Default Diffuse Priors for Variance Parameters Gamma priors C Uniform on variance scale Reset Done Informative Priors 9 Help Classifications Missing Data BUGS options Measurement Error Factor Analysis Correlated Residuals From this screen we can bring up the BUGS options by pressing on its button and the following screen appears 2 BUGS Options of x Saving Options Save current model in BUGS format Loading Options Input out file fest out Input ind file ftest ind Start column number c3200 v Input data Done From this screen we can save the BUGS code for the currently set up model or read in the output files that contain parameter traces from BUGS for use in MLwiN see later For now we will save our cu
21. rrent model in BUGS format so Click on the large button at the top of the screen This will bring up a file save window similar to those for inputting and saving worksheets For now we will save the file in the default directory as tutorial bug This will create a file that contains the BUGS model definition initial values and data For users of classic BUGS who are used to having three separate files in WinBUGS the file tutorial bug contains the information from these three files with dividing comment lines between them For background information on using WinBUGS it is strongly suggested that the user reads some of the user manual and examples documentation that comes with the package in particular to become familiar with the WinBUGS notation For now to fit our model in WinBUGS we must start the WinBUGS program and read in the file tutorial bug from the directory it was saved in as a text file Note that you will have to change the Files of type box to All files to see the file tutorial bug Having read in the file a window headed tutorial bug will appear containing the information needed by BUGS for this model As mentioned earlier the WinBUGS code is split into 3 sections and we will consider these here in turn Firstly a model definition is required and this consists of a description of the structure of the current problem The code for our simple variance components problem is as follows WINBUGS code generated from ML
22. s given to creating fast algorithms for the particular classes of models that MLwiN can fit and more documentation is provided MLwiN allows the user to perform both frequentist and Bayesian analysis of multilevel and other models The frequentist analyses are performed using least squares based methods such as IGLS Goldstein 1986 and RIGLS whilst the Bayesian analyses use both Gibbs sampling and Metropolis Hastings sampling depending on the problem Users specify their models via an equation based interface and running the analysis involves simply clicking on a start button MLwiN to WinBUGS interface A new development release of MLwiN is due out early next year and contains many new MCMC features In particular the new version will contain the ability to convert any model set up in MLwiN into WinBUGS code that can be run using WinBUGS version 1 3 Models available include Normal response linear and multilevel models Multilevel logistic and probit regression models Multilevel Poisson regression models Measurement error models Cross classified and multiple membership models Spatial CAR models Multivariate normal response models Multilevel factor analysis models The new release of MLwiN will coincide with the release of a detailed manual MCMC Estimation in MLwiN by W J Browne 2002 and the remainder of this poster is an adaptation of a chapter on the WinBUGS interface taken from this book The MCMC features in MLw
23. s start the estimation MLwiN does not by default time estimation but we can set this by enabling the Smileys option The Smileys option gives the user a smiley face on the screen that shows when the program is performing an operation or not and as a side effect the timer is enabled Select Close All Windows from the Windows menu Select Smileys from the Options menu Click on the Start button When the estimation finishes a message box will appear saying how long the estimation took in hours minutes and seconds For this example the estimation took 42 seconds on a 733MHz Pentium Click on the OK button to remove this window We can now get Effective sample sizes and other information from the MCMC diagnostics window Select Trajectories from the Model Menu Click in the B chain box Select Yes to the Calculate MCMC Diagnostics box The diagnostics for B using Metropolis sampling will then be displayed as shown below Here we have higher autocorrelation and so we see that the effective sample size is only 185 E MCMC diagnostics olx g a E D z T m a a x Accuracy Diagnostics Raftery Lewis quantile Nhat 27779 22369 when q 025 975 r 005 ands 95 Brooks Draper mean Nhat 22592 when k 2 sigfigs and alpha 05 Summary Statistics param name f posterior mean 0 540 0 0054 SD 0 089 mode 0 531 quantiles 2 5 0 724 5 0 692 50 0 537 95 0 399 97 5
24. school level residuals in the tutorial example we considered earlier In the MLwiN manual Rasbash et al 2000 we look at plots of residuals against normal scores to confirm that the normal distributional assumption is a good fit to the data The normal distribution is a member of the t distribution family The t distribution family has an additional degrees of freedom df parameter and the normal distribution is the limiting case when this parameter reaches infinity We will here consider replacing the normal distribution at level 2 with a t distribution where the df parameter has itself got a prior distribution For this we will use a uniform prior and allow the df parameter to take values in the range 1 to 200 This allows both small values where the distribution has very long tails and large values which are indistinguishable from the normal distribution To include this prior we will need to edit the model definition in the file tutorial bug The new version is as follows edits in bold font MODEL Definition model Level 1 definition for i in 1 N normexam i dnorm mu i tau mul i lt beta 1 cons i beta 2 standirt i u2 schoolfi cons i Higher level definitions for j in 1 n2 u2 j dt 0 tau u2 df Priors for fixed effects for k in 1 2 beta k dflat Priors for random terms tau dgamma 0 001 0 001 sigmaz2 lt 1 tau tau u2 dgamma 0 001 0 001 s
25. t distributed residuals at the school level It should be noted at this point that most multilevel models are large and so cannot be run using the educational version of WinBUGS and so you will need to have the release version of WinBUGS Note that you will need to update from the educational version of WinBUGS to run the multilevel models listed here Variance Components models in WinBUGS We will consider here an educational dataset where our response is a normalized total exam score at age 16 for 4059 children in 65 schools in the UK We have set up a variance components model with one explanatory variable which is a standardized London reading test taken by all the children at age 11 and acts as an intake ability indicator The model was run using the IGLS MLE method in MLwiN On convergence we get the following estimates Equations Ioj Xx normexam N XB Q normexam cons 0 563 0 012 standlrt Boy 9 002 0 040 ug 2 gy xq TNO 20 Qu 0 092 0 018 en TNO Q 2 0 566 0 013 2 loglikelihood IGLS Deviance 9357 242 4059 of 4059 cases in use We could now fit this model using MCMC in MLwiN but here we will consider instead using WinBUGS To get to the BUGS options in MLwiN we need to do the following Click on the Estimation Control button Select the MCMC tab Click on the Advanced MCMC Options button This will bring up the Advanced MCMC Options screen that shows the possible MCMC sam
26. t the Samples options and a window will appear that allows the user to specify which parameters to monitor In this window we will firstly select the fixed effects by typing beta in the node box Note that when a correctly typed parameter is input the set button will become enabled We will also want to use a burnin of 500 iterations so we will also modify the beg value from 1 to 501 After this press the set button and the parameter will be set for monitoring We now need to repeat this procedure with the two variance parameters sigma2 and sigma2 u2 It should be noted that it is possible in WinBUGS to get dynamic traces of the parameters like those in the trajectories window in MLwiN via the sample window If we either type beta again in the node box or use the scroll button at the side of the box to select beta you will see that now all the buttons become enabled Clicking the trace button will give 2 empty trace plots for beta 1 as shown below and beta 2 which will become dynamic when we start updating Similar traces can be brought up for the two variance parameters beta 1 1 0 0 5 0 0 iteration We are now ready to set the estimation engine running and this is done via the Update window found in the Model menu We need to specify the number of updates including the burn in and so we will replace the 1000 here with 5500 as is used in MLwiN As in MLwiN you can also specify how often to refresh the screen whether to use thinnin
27. variables We will need to repeat the setting up procedure that we used for the normal response model Select the Specification window from the Model menu Click on the Check Model button Highlight the ist identifier at the start of the data list Click on the Load Data button Click on the Compile button Highlight the ist identifier at the start of the initial values list Click on the Load Inits button This will have set up our model and we can now pick our variables to store Before we do this we will introduce an interesting feature of WinBUGS not mentioned in the first example If we wish to find out what methods WinBUGS is using to fit the various components of the model we have defined we can use the Node Info tool available from the Info menu If we then type our node name for example beta into the node box on this window we can then hit the method button and get the following in the log window beta 1 UpdaterDFreeARS StdUpdater beta 2 UpdaterDFreeARS StdUpdater This can be translated to mean that both the fixed effect are being updated by the standard AR sampler If on the other hand we type the node name tau u2 into the node box and hit the method button we will get tau u2 UpdaterGamma OptUpdater This is because the precision parameter tau u2 is being updated using Gibbs sampling from a Gamma full conditional distribution So we now need to tell WinBUGS which parameters we wish to store Select the Samples wi
28. wiN program MODEL Definition model Level 1 definition for i in 1 N normexam i dnorm mu i tau mul i lt beta 1 cons i beta 2 standirt i u2 schoolfi cons i Higher level definitions for j in 1 n2 u2 j dnorm 0 tau u2 Priors for fixed effects for k in 1 2 beta k dflat Priors for random terms tau dgamma 0 001 0 001 sigma2 lt 1 tau tau u2 dgamma 0 001 0 001 sigma2 u2 lt 1 tau u2 WinBUGS is a more general modelling package and so there is no standard order to the model description although when the code is generated from MLwiN it will generally have a similar structure We firstly define the relationship between the response in this example normexam and the fixed and random predictor variables Note that the column names from MLwiN are used as the variable names in WinBUGS WinBUGS does have some differences in what it allows as a variable name so if the WinBUGS code will not work it may be that some of your variable names are illegal for example a column name like 995 will be interpreted as a number in WinBUGS so it is worth renaming such columns in MLwiN So we see here that our response is normally distributed and that we have two fixed effects beta 1 and beta 2 always defined as beta by the code generator and one set of random effects u2 always defined as u where is the level classification indicator Note that
Download Pdf Manuals
Related Search