I'm in JMP Pro 15.2.1.
I have 60 thousand observations, where the response is true/false and two categorical predictors: job, and group (assume they're not correlated). Many thousands of observations. I was about to do a logistic regression under the assumption that every observation was independent. But then I realized that each person had three consecutive response opportunities (one each quarter) and that their successive responses are not entirely independent from earlier responses. I have a person ID for each response, so I know which responses belong to each person ID.
My first attempt was simply to use the 'fit model' platform with a nominal logistic personality. But I wasn't sure how to handle the repeated measures generated by the 'case' variable. I could ignore case, and treat every observation as independent, but that's probably invalidating my test statistics.
For repeated measures, I tried the multiple response module in analyze -> consumer research --> categorical -> related -> repeated measures. 'response' is my repeated measure, 'person' is my ID, and 'job' and 'group' are my X values. But when I run the model, I don't see any of the tests that I'd expect to see with logistic regression. Am I in the wrong module after all?
| person | case | response | job | group |
| a | 1 | T | exec | blue |
| a | 2 | F | exec | blue |
| a | 3 | T | exec | blue |
| b | 1 | F | staff | blue |
| b | 2 | F | staff | blue |
| b | 3 | T | staff | blue |
| c | 1 | F | supt | orange |
| c | 2 | T | supt | orange |
| c | 3 | T | supt | orange |
