Repeated measures with censoring


New Member
I am conducting an experiment to test spin times of tops.

I have 5 aluminum tops and 5 wooden tops. I spin each top 3 times with a maximum spin time of 2 minutes (after 120 seconds I stop the top if it is still spinning and record a value of 120s). I wish to test any differences in spin time that might exist between the 2 types of tops.

Issue 1: In writing the SAS program, I'm having trouble figuring out how to denote censoring (120 second values) when reading in the data.

Issue 2: I'm planning to use PROC GLM and pay attention to the between subjects analysis, ignoring all else (this isn't a time series). Any problems with this?




Dark Knight
I guess you are doing analysis the duration of spinning time. Here you can treat the data as survival data.
Issue1: You can create a flag variable for censor. ( if censored, flag =1or 0)
Issue2: Check the censor % and if it is negligible you can go with proc GLM


New Member
Thanks for the reply vinux. I am indeed measuring duration of spin time (latency to stop spinning).

1: Your suggestion of using a censoring variable is a good one. I can figure out how to set that up when each independent variable has only one dependent variable, but with repeated measures, I'm having trouble.

2: For most experiments, censor % is in the neighborhood of 20-25%. I'm guessing that's not negligible.