Welcome to the Home Page for

Comparing Three Modern Approaches to Longitudinal Data Analysis:

An Examination of a Single Developmental Sample

A Symposium presented at the 1997 Biennial Meeting of the
Society for Research in Child Development, Washington, D.C.
 
     This is the home page for the 1997 SRCD symposium that focused on modern approaches to longitudinal data analysis. The goal of this symposium was to compare and contrast three recently developed methods for analyzing developmental change over time. A single developmental data set was provided to all of the symposium participants with the instructions to analyze the data in any way they wished using a data analytic approach of their own choosing. The three analytical approaches that were selected were latent growth modeling (by Jack McArdle), hierarchical linear modeling (by Mike Seltzer and Steve Raudenbush), and exploratory data analysis (by Mark Appelbaum). Patrick Curran organized the symposium, created the data set and wrote the documentation, and Hendricks Brown was the discussant.
 
     The general topical area of interest was the development of aggressive behavior and reading ability in children over time, and the relation between individual differences in development in these behaviors and several time specific child and family characteristics. The participants were provided a general introduction to the theoretical questions of interest as well as a set of substantive research hypotheses that might be of interest to study further. The sample consisted of N=405 children drawn from the Children of the National Longitudinal Survey of Youth, about half of which were missing one or more of the repeated measures on aggression or reading ability. The measures of interest included four repeated measures of aggressive behavior taken at two-year intervals, four repeated measures of reading achievement taken at two-year intervals, Time 1 measures of cognitive stimulation and social support of the child in the home, and background measures of child gender, child age, and mother age.
 
     This page was created in hopes that anyone who was interested could download the same data set that was used by the symposium participants to try their own hand at analyzing the data and to compare their results to those of the participants. This page will be periodically updated with the specific analyses presented by the participants as well as anyone else's analyses who would like to share their results with others.
 
     What follows are the supporting documentation for the symposium, the raw data file, the necessary SAS code for manipulating the raw data, and various approaches that have already been used to analyze this data. For all the following files, the file contents can be saved directly to your disk using the "save as..." option under the "File" menu, or "select all", then "copy", and then "paste" the contents of the file into your own processor (e.g., SAS editor, MSWord, etc.), then save that file to your disk.


Full Supporting Documentation, either in HTML  or PDF  format, written by Patrick J. Curran, Duke University. Note that the free Adobe Acrobat Reader is required for opening the PDF format file.
 
 

Summary of contents
Raw data file of N=405 cases in ASCII format

SAS code  that inputs the ASCII raw data file from above, computes descriptive statistics on all variables, and reads out the raw data for use in other programs

SAS log and SAS output that result from running the SAS code listed above.
 

The above files provide the actual raw data and the SAS code needed to input the data, name the variables, and read out new data sets for input into other statistics programs. The following files are sample programs written in both EQS and HLM that test various questions about growth. To use these files, you must have access to the full software packages of HLM or EQS. The following files simply contain code used to define various types of growth models -- these files can not be executed by themselves. Also, note that these files simply represent a small sampling of ways in which growth could be modeled using this data. They are in no way meant to reflect "the" way to test growth.
This page is very much a 'work in progress', and I plan on adding further resources as time allows. Please check back occasionally for these changes. Also, if you have analyses of these data that you might like to share with others, send me a note and let me know.
 
This page was created and is maintained by Patrick Curran in the Department of Psychology at The University of North Carolina. Please feel free to send any comments or suggestions to curran@unc.edu.
 
Last Updated 7-28-97