SAS Code Repository for Managing ILD
SAS Code Repository
Working with intensive longitudinal data (ILD) can be challenging because of the size, complexity, and nested structure of the data. We have created several files of SAS syntax that can facilitate handling and pre-processing of ILD and multilevel data. Click on files names to download. The code below uses the sample data set democ.sas7bdat.
Questions? Email firstname.lastname@example.org.
This code can be used to center data. Two centering techniques are demonstrated,
- grand-mean centering, in which values are centered on the grand mean across all people and observations and
person-mean centering, in which values are centered on each person’s mean which allows estimation of within-person associations.
If data are organized in wide format, this code will help you convert them into long format (i.e., one row per subject to one row per observation). This organization is necessary for running multilevel or mixed models (e.g., PROC MIXED or PROC GLIMMIX), time-series models (e.g., PROC ARIMA) and the %TVEM SAS macro.
This code identifies the intraclass correlation in multilevel data. It allows you to identify how much of a variable’s variance is attributable to within-person differences (how much people differ from themselves over time) and how much is attributable to between-person differences (how much people differ from each other on average).
This code enables users to identify intraindividual standard deviation (iSD) and mean of squared successive differences (MSSD). These measures allow researchers to determine how much each individual’s responses on a specific item (e.g., their mood) fluctuate across time, with larger values indicating greater instability.
This code allows users to adjust data to insert lags, leads, and successive differences. This is useful for researchers who are trying to examine how exposure at one time point affects a variable at a later time point, as well as how stable versus unstable exposures and outcomes are over time.