We make available the data sets and some of the associated output from
Rogosa, D. R., and Saner, H. M. (1995). Longitudinal data analysis examples with random coefficient models. Journal of Educational and Behavioral Statistics, 20, 149-170. Also: Reply to Discussants, 234-238.

The data analysis and generation programs are also available. timepath tpsim

Data examples

1. Rat Data shown in Exhibit 1, p.153 (also from 1989 HLM manual)
2. Artificial Data also shown (in part) in Exhibit 1, p.153 created by tpsim comes in three versions--see description p.159
   • Full Data aka YFZF
   • Y-missing, Z-full which was the data used for Table 2 results (miss=999)
   • Y-missing, Z-missing (miss=999)
   • The data sets with missing values were constructed by applying a small utility program tpmiss to the YFZF data
3. Artificial Data (n=100) used in the Reply (described pp.236-7) having different observation times for each individual
   • Full Data (Y-values and t-values listed separately)
   • Y-missing, Z-full (Y-values and t-values listed separately)

HLM Output

We provide utilities to translate data sets from the rectangular form of Exhibit 1 to the between-unit and within-unit data files required by the HLM program.

HLM output provided includes sets of runs (CNBN, CYBN, CNBY, CYBY) for:

• Runs for the Rat Data summarized in Table 1 (p.157)
• Runs for the YMZF Artificial Data summarized in Table 2 (p.160)
• Runs the North Carolina Data summarized in Table 3 (p.161)
• Previous runs for the North Carolina Data in 1989 discussed on p.163 and in footnote p.237.
• Runs for the Artificial Data summarized in the Reply pp. 236-7.