<CENTER>
<H1>BSTT513 Longitudinal Data Analysis - Fall 2004</H1></CENTER>
<CENTER>
<H1>Instructor: Don Hedeker</H1></CENTER>
<CENTER>
<H1>Datasets and class notes</H1></CENTER>
<HR>

<H3>Class notes and handouts in PDF or ASCII form</H3>
<P>
<B><I>Week 1: Monday August 23, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/bio513ab.pdf" target="_blank" >Course syllabus and information sheet</A></B> </P>
<P><B><I>Reading material</I>: Hedeker, D. and Gibbons, R.D. "Longitudinal Data Analysis" (in progress).</B>
<B>Chapter 1: Introduction (<a href="http://tigger.uic.edu/%7Ehedeker/chap1.pdf" target="_blank" >pdf file</A>)</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/intro.pdf" target="_blank" >PDF file - introduction overheads</A></B>
  </P>
<P><B><I>Week 1: Wednesday August 25, 2004</I></B> </P>
<P><B><I>Reading material</I>: Hedeker, D. and Gibbons, R.D. "Longitudinal Data Analysis" (in progress).</B>
<B>Chapter 2: ANOVA approaches to longitudinal data (<a href="http://tigger.uic.edu/%7Ehedeker/chap2.pdf" target="_blank" >pdf file</A>)</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/anova.pdf" target="_blank" >PDF file - univariate ANOVA overheads</A></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/bockv.txt" target="_blank" >bockvoc</A> - ASCII file with SAS code and output from analysis of vocabulary data from Bock (1975) using univariate mixed-effects ANOVA model that is equivalent to a randomized-block ANOVA.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/Bio513p1.pdf" target="_blank" >Problem Set 1</A></B>
  </P>
<P><B><I>Week 2: Monday August 30, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/pothoff0.txt" target="_blank" >pothoff0</A> - ASCII file with SAS code and output from analysis of Pothoff & Roy data using univariate mixed-effects ANOVA model that is equivalent to a split-plots ANOVA (two groups: boys and girls). Includes a plot of the means across time for the two groups.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/pothoffb.txt" target="_blank" >pothoffb</A> - ASCII file with SAS code and output from analysis of Pothoff & Roy data using univariate mixed-effects ANOVA. This analysis is for a main effects model (no group by time interaction).</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/pothoff2.txt" target="_blank" >pothoff2</A> - ASCII file with SAS code and output from analysis of Pothoff & Roy data using multivariate ANOVA. Includes test of sphericity & univariate results.</B>
  </P>
<P><B><I>Week 2: Wednesday September 1, 2004</I></B> </P>
<P><B><I>Reading material</I>: Hedeker, D. and Gibbons, R.D. "Longitudinal Data Analysis" (in progress).</B>
<B>Chapter 3: MANOVA approaches to longitudinal data (<a href="http://www.uic.edu/%7Ehedeker/chap3.pdf" target="_blank" >pdf file</A>)</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/manova.pdf" target="_blank" >PDF file - multivariate analysis of variance for repeated measures overheads</A></B>
  </P>
<P><B><I>Week 3: Monday September 6, 2004 - LABOR DAY - no class</I></B>
  </P>
<P><B><I>Week 3: Wednesday September 8, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/bockvoc2.txt" target="_blank" >bockvoc2</A> - ASCII file with SAS code and output from analysis of vocabulary data using MANOVA.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/bockvoc3.txt" target="_blank" >bockvoc3</A> - ASCII file with SAS code and output from analysis of sleep data using IML to do MANOVA.</B> </P>
<P><B><a href="http://tigger.uic.edu/%7Ehedeker/prozwtc.txt" target="_blank" >prozwtc</A> - ASCII file with SAS code and output from analysis of Prozac weight data using multivariate ANOVA. Includes use of CONTRAST statement to obtain estiates of a-priori contrasts for the multi-group situation.</B>
  </P>
<P><B><I>Week 4: Monday September 13, 2004</I></B> </P>
<P><B><I>Reading material</I>: Hedeker, D. and Gibbons, R.D. "Longitudinal Data Analysis" (in progress).</B>
<B>Chapter 4: Mixed-effects regression models for continuous outcomes (<a href="http://tigger.uic.edu/%7Ehedeker/chap4.pdf" target="_blank" >pdf file</A>)</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/long1.pdf" target="_blank" >PDF file - introduction to mixed models overheads</A></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/RIESBYM.txt" target="_blank" >RIESBYM</A> - ASCII file with SAS code and output from analysis of Riesby dataset using a few different MIXED models. Includes grouping variable and curvilinear effect of time.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/bio513p2.pdf" target="_blank" >Bio513p2.pdf</A>- second problem set.</B>
  </P>
<P><B><I>Week 4: Wednesday September 15, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/RIESBYM2.txt" target="_blank" >RIESBYM2</A> - ASCII file with SAS code and output from analysis of Riesby dataset. This handout shows how empirical Bayes estimates can be output to a dataset in order to calculate estimated individual scores at all timepoints.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/RIESBY5.txt" target="_blank" >RIESBY5</A> - SAS code and output from analysis of Riesby dataset. This handout has the analysis considering the time-varying drug plasma level effects.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/riesby5b.txt" target="_blank" >RIESBY5b</A> - SAS code and output from analysis of Riesby dataset. This handout has the analysis considering the time-varying drug plasma levels, separating the within-subjects from the between-subjects effects for these time-varying covariates.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/riesby5d.txt" target="_blank" >RIESBY5d</A> - SAS code and output from analysis of Riesby dataset: subjects with complete data at the last 4 timepoints. This handout has the analysis considering the time-varying drug plasma levels, separating the within-subjects from the between-subjects effects for these time-varying covariates, and showing the relationship between ordinary regression analysis of the means and the between-subjects drug effects.</B>
  </P>
<P><B><I>Week 5: Monday September 20, 2004</I></B> </P>
<P><B><I>Reading material</I>: Hedeker, D. and Gibbons, R.D. "Longitudinal Data Analysis" (in progress).</B>
<B>Chapter 5: Mixed-effects polynomial regression models for continuous outcomes  (<a href="http://tigger.uic.edu/%7Ehedeker/chap5.pdf" target="_blank" >pdf file</A>)</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/orthpoly.pdf" target="_blank" >PDF file - orthogonal polynomials in mixed models overheads</A></B>
  </P>
<P><B><I>Week 5: Wednesday September 22, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/riesby4b.txt" target="_blank" >riesby4b</A> - ASCII file with SAS IML code for generating orthogonal polynomial contrast matrix.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/riesby3.txt" target="_blank" >riesby3</A> - ASCII file with SAS code and output from analysis of Riesby dataset. This handout contrasts models using the raw metric of week versus orthogonal polynomials.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/RIESBY4.txt" target="_blank" >RIESBY4</A> - ASCII file with SAS IML code and output that lists the calculation of estimated means and the variance-covariance matrix based on several different models.</B>
  </P>
<P><B><I>Week 6: Monday September 27, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/RandCoef.PDF" target="_blank" >RandCoef.PDF </A>- PDF file with overheads for random coefficients model article "Estimating Individual Influences of Behavioral Intentions: An Application of Random-effects Modeling to the Theory of Reasoned Action," Hedeker, Flay, & Petraitis (1996) Journal of Consulting and Clinical Psychology, 64:109-120.  
<a href="http://www.uic.edu/%7Ehedeker/rrmtra.PDF" target="_blank" >pdf file of article</A> (optional reading)</B></P>
<P><B>
<a href="http://www.uic.edu/%7Ehedeker/bio513p3.pdf" target="_blank" >bio513p3.pdf</A> - third problem set.</B>
  </P>
<P><B><I>Week 6: Wednesday September 29, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/relapse.pdf" target="_blank" >relapse.pdf</A>- PDF file for analysis of relapse data article "Application of Random-Effects Regression Models in Relapse Research," Hedeker & Mermelstein (1996) Addiction (supplement), 91:S211-S229.</B>
<B><a href="http://www.uic.edu/%7Ehedeker/Addict.PDF" target="_blank" >pdf file of article</A> (optional reading)
</B></P>
<P><B><I>Week 7: Monday October 4, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/estim.PDF" target="_blank" >PDF file - mixed model estimation overheads</A></B>
  </P>
<P><B><I>Week 7: Wednesday October 6, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/RRM2RIES.txt" target="_blank" >RRM2RIES</A> - SAS code and output from analysis of Riesby dataset. This handout includes SAS IML code for a random-intercepts model. A comparison to analysis using PROC MIXED is also included.</B>
  </P>
<P><B><I>Week 8: Monday October 11, 2004</I></B> </P>
<P><B><I>Reading material</I>: Hedeker, D. and Gibbons, R.D. "Longitudinal Data Analysis" (in progress).</B>
<B>Chapter 6: Covariance pattern models for continuous outcomes  (<a href="http://tigger.uic.edu/%7Ehedeker/chap6.pdf" target="_blank" >pdf file</A>)</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/VarCov2.pdf" target="_blank" >PDF file - variance-covariance structure models overheads</A></B>
  </P>
<P><B><I>Week 8: Wednesday October 13, 2004</I></B> </P>
<P><B><I>Reading material</I>: Hedeker, D. and Gibbons, R.D. "Longitudinal Data Analysis" (in progress).</B>
<B>Chapter 7: Mixed regression models with autocorrelated errors (<a href="http://tigger.uic.edu/%7Ehedeker/chap7.pdf" target="_blank" >pdf file</A>)</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/ACerr2.pdf" target="_blank" >PDF file - mixed effects models with autocorrelated errors overheads</A></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/Bockrrm.pdf" target="_blank" >ACerrors: extra page</A>- a PDF file with an extra page on mixed-effects models with autocorrelated errors</B> </P>

<P><B><I>Week 9: Monday October 18, 2004</I></B> </P>
<P><a href="http://www.uic.edu/classes/bstt/bstt513//bockrrm3.sas.txt" target="_blank" ><B>BOCKRRM3.SAS</B></A><B> - SAS code for analysis of Bock dataset. This handout lists syntax for several PROC MIXED analyses including (a) mixed-effects models, (b) covariance structure models, and (c) mixed-effects models with autocorrelated errors.</B>
</P><B><a href="http://www.uic.edu/%7Ehedeker/bock2b.txt" target="_blank" >BOCK2</A> - SAS IML code and output from analysis of Bock dataset. This handout uses IML to provide estimated means, variances, and correlations across time based on several mixed-effects models with autocorrelated errors.</B>


<P><B><a href="http://www.uic.edu/%7Ehedeker/bockrrm.dat.txt" target="_blank" >BOCKRRM.DAT</A> - Bock dataset. The variable order and names are indicated in the previous two handouts.</B>
  </P>
<P><B><I>Week 9: Wednesday October 20, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/classes/bstt/bstt513/bio513p4.pdf" target="_blank" >bio513p4.pdf</A> - fourth problem set.</B> </P>
<P>
  </P>
<P><B><I>Week 10: Monday October 25, 2004</I></B>  

<B><I>Reading material</I>: Hedeker, D. and Gibbons, R.D. "Longitudinal Data Analysis" (in progress).</B>
<B>Chapter 9: Mixed-effects regression models for binary outcomes. <a href="http://www.uic.edu/classes/bstt/bstt513/chap9.pdf" target="_blank" >(pdf file)</A></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/Long3.pdf" target="_blank" >PDF file - mixed logistic models overheads</A></B>
  
  </P>
<P><B><I>Week 10: Wednesday October 27, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/schizblr.txt" target="_blank" >schizblr </A>- Using the NIMH Schizophrenia dataset, this handout provides SAS (PROC LOGISTIC, NLMIXED, and PROC IML) code for showing the relationships between (a) probit versus logistic fixed-effects estimates and (b) fixed-effects versus mixed effects estimates.  This handout also shows how estimates of the dichotomous outcome can be obtained from the various models.</B>
  
<B><a href="http://www.uic.edu/%7Ehedeker/SCHIZX1.DAT.txt" target="_blank" >SCHIZX1.DAT</A> - ASCII datafile for example above.

</B></P>
<P><B><I>Week 11: Monday November 1, 2004</I></B>
  </P>
<P><B><I>Week 11: Wednesday November 3, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/mlogest.PDF" target="_blank" >PDF file - estimation of mixed logistic models overheads</A></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/mixbin.sas.txt" target="_blank" >mixbin.sas</A> - SAS IML macro program for random-intercepts binary regression</B>
  </P>
<P><B><I>Week 12: Monday November 8, 2004</I></B> </P>
<p>
<P><B><I>Reading material</I>: Hedeker, D. and Gibbons, R.D. "Longitudinal Data Analysis" (in progress).</B>
<B>Chapter 8: Generalized Estimating Equations (GEE) Models. <a href="http://www.uic.edu/classes/bstt/bstt513/chap8.pdf" target="_blank" >(pdf file)</A></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/gee.PDF" target="_blank" >PDF file - GEE models overheads</A>
</B></P>
<P><B><a href="http://www.uic.edu/classes/bstt/bstt513/geetab.pdf" target="_blank" >PDF file - GEE models - extra page</A></B></P>
<P><B><a href="http://www.uic.edu/classes/bstt/bstt513/bio513p5.pdf" target="_blank" >bio513p5.pdf</A> - last problem set</B> </P>
<P><B><a href="http://www.uic.edu/classes/bstt/bstt513/arthrit.sas.txt" target="_blank" >arthrit.sas</A> - program with some SAS code for problem set 5
</B></P>
<P><B><a href="http://www.uic.edu/classes/bstt/bstt513/arthritb.dat.txt" target="_blank" >arthritb.dat</A> - data for problem set 5

</B></P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/schizgee.txt" target="_blank" >schizgee</A> - Using the NIMH Schizophrenia dataset, this handout has SAS PROC GENMOD code and output from several GEE analyses varying the working correlation structure.</B> </P>
<P><B><a href="http://www.uic.edu/classes/bstt/bstt513/robingeb.txt" target="_blank" >robingeb</A> - SAS PROC GENMOD code and output from analysis of Robin Mermelstein's smoking cessation study dataset. This handout illustrates GEE modeling of a dichotomous outcome and also includes SAS IML code to perform a multi-parameter Wald test.</B> </P>
<P><B><a href="http://www.uic.edu/classes/bstt/bstt513/robing3c.txt" target="_blank" >robing3c</A> - SAS IML code and output from analysis of Robin Mermelstein's smoking cessation study dataset. This handout illustrates computation of estimated probabilities based on the parameter estimates from the GEE analysis.</B>
  </P>
<P><B><I>Week 12: Wednesday November 10, 2004</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/mixorfi3.txt" target="_blank" >mixorfi3</A> - SAS IML macro program for getting marginal probabilities from a random-intercepts binary logistic regression.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/mixorfp3.txt" target="_blank" >mixorfp3</A> - SAS IML macro program for getting marginal probabilities from a random-intercepts binary probit regression.</B>
  </P>
<P><B><I>Week 13: Monday November 15, 2004</I></B> </P>
<P><B><I>Reading material</I>: Hedeker, D. and Gibbons, R.D. "Longitudinal Data Analysis" (in progress).</B>
<B>Chapter 10: Mixed-effects regression models for ordinal outcomes. <a href="http://www.uic.edu/classes/bstt/bstt513/chap10.pdf" target="_blank" >(pdf file)</A></B></P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/Long4.pdf" target="_blank" >PDF file - mixed logistic models for ordinal data overheads</A></B> </P>
<P><B><U>
</U></B><B></B></P>
<P><B><I>Week 13: Wednesday November 17, 2004</I></B>
  </P>
<P><B><I>Week 14: Monday November 22, 2004
</I></B></P>
<P><FONT size=+1><B><I>Reading material: </I></B>Hedeker, D., & Gibbons, R.D. (1997).  Application of random-effects pattern-mixture models for missing data in longitudinal studies.  <I>Psychological Methods, 2,</I> 64-78. (<a href="http://tigger.uic.edu/%7Ehedeker/RRMPAT.pdf" target="_blank" >pdf file</A>)</FONT> </P>
<p>
<P><B><a href="http://www.uic.edu/%7Ehedeker/missing.PDF" target="_blank" >PDF file - missing data overheads</A></B> </P>
<P><B><I>
Week 14: Wednesday November 24, 2004</I></B>
<B>
<a href="http://www.uic.edu/%7Ehedeker/Long2.pdf" target="_blank" >PDF file - pattern mixture models overheads</A></B>  
</P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/schizpm2.sas.txt" target="_blank" >schizpm2.sas</A> - ASCII file with SAS code from analysis of NIMH Schizophrenia dataset to perform a pattern-mixture analysis. Includes IML code to do the mixing over the pattern results.</B>

</P>
<P><B><I>Week 15: Monday November 29, 2004</I></B> </P>
<P><B><a href="http://tigger.uic.edu/%7Ehedeker/Pow2grp.pdf" target="_blank" >PDF file - sample size calculations overheads</A></B>
  </P>
<P><B><I>Week 15: Wednesday December 1, 2004</I></B>
  </P>
<P>-----------------------------------------------------------------------------------------------------------------------------
<B><I>extras:</I></B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/bockgee.txt" target="_blank" >bockgee</A> - SAS PROC MIXED & GENMOD code and output from analysis of Bock dataset. This handout compares results from mixed-effects modeling to GEE modeling for this dataset with no missing data across time and a continuous outcome variable.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/riesgee2.txt" target="_blank" >riesgee2</A> - SAS PROC MIXED & GENMOD code and output from analysis of Riesby dataset. This handout compares results from mixed-effects modeling to GEE modeling for this dataset which does have missing data across time and a continuous outcome variable.</B>
  </P>
<p>
<HR>

<H3><B>The following datasets are in ASCII form and can be downloaded</B></H3><B><a href="http://www.uic.edu/%7Ehedeker/vt.dat.txt" target="_blank" >vt.dat</A> has data on 22 subjects from a study of affective facial expressions (Vasey and Thayer, 1987).</B>
<B>In this study several pieces of music were played to each subject in an attempt to elicit selected affected states.  Trial 1 was a baseline, relaxing music condition.  Trial 2 was designed to produce positive effects, and trial 3 was designed to produce agitation.  Each trial lasted 90 seconds and the response variable at each trial was the mean electromyographic (EMG) amplitude from the left brow region.  The variable order in the datafile is subject number</B>
<B>followed by the three trial EMG measurements.  For problem set 1.</B>
<P><B><a href="http://www.uic.edu/%7Ehedeker/box.dat.txt" target="_blank" >box.dat</A> has body weight data for 30 rats measured at baseline and at weekly intervals for 4 weeks. There are three groups (1=control, 2=thiouracil, 3=thyroxin) of 10 rats each.. The datafile contains, in order, GROUP, RAT, BW0, GAIN1 (BW1-BW0), GAIN2 (BW2-BW1), GAIN3 (BW3-BW2), and GAIN4 (BW4-BW3). For problem set 1.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/SCHIZREP.DAT.txt" target="_blank" >SCHIZREP.DAT</A> has severity of illness scores on 437 schizophrenics measured across time. Subjects were randomized to one of four treatments: placebo, chlorpromazine, fluphenazine, or thioridazine. Here the drug groups have been combined into one group. The datafile contains, in order, Patient ID, IMPS79 (7-point severity scale), week (week 0 to week 6, though most of the measurement occured on weeks 0, 1, 3, & 6), treatment group (0=placebo, 1=drug), and sex (0=female, 1=male). For problem set 2.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/SCHIZX1.DAT.txt" target="_blank" >SCHIZX1.DAT</A> - Same dataset as above with the following variables, in order, Patient ID, IMPS79 (7-point severity scale), IMPS79b (binary version of IMPS79), IMPS79o (ordinal version of IMPS79), intercept (a column of ones),  treatment group (0=placebo, 1=drug), week (week 0 to week 6, though most of the measurement occured on weeks 0, 1, 3, & 6), square root of week (helps to linearize the relationship of IMPS79 over time), and treatment group by square root of week (interaction term).</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/RIESBYT4.DAT.txt" target="_blank" >RIESBYT4.DAT</A> has Hamilton Depression Rating Scale (HDRS) scores on 66 depressed subjects across time. Subjects were given imipramine for four weeks and their drug plasma levels were obtained during each week. Drug plasma levels of desimipramine (imipramine's metabolite) were also obtained. The variables, in order, are HDRS change from baseline score, a field of ones, week (coded 0 to 3), sex (0=male, 1=female), diagnosis (0=non-endogenous, 1=endogenous), imipramine plasma levels (in ln units), and desimipramine plasma levels (in ln units). For problem sets 3 and 4.</B> </P>
<P><B><a href="http://www.uic.edu/%7Ehedeker/RIESORD3.RRM.txt" target="_blank" >RIESORD3.RRM</A> has Hamilton Depression Rating Scale (HDRS) scores on 66 depressed subjects across time. The variables, in order, are HDRS score, HDRS dichotomized score, HDRS trichotomized score, a field of ones, week (coded 0 to 3), DMI dichotomized value, DMI dichotomized value by week, DMI centered value (in ln units), and DMI centered value (in ln units) by week.</B> </P>