0

I have two groups control and test. For both groups data was collected at two time points : One is Baseline and other End of Study. Data is not normally distributed. However, only 50% of data is available for End of study as compared to baseline. So to compare my baseline data with End of study data... it is mix of unpaired and paired data and also it is not normally distributed. I want to know which statistical test can I use?

Jenny
  • 1
  • [here](https://stats.stackexchange.com/questions/25941/t-test-for-partially-paired-and-partially-unpaired-data) is an answer for the same situation, but normal assumption. Same thinking can be applied – kjetil b halvorsen Mar 30 '19 at 08:39
  • Possible duplicate of [t-test for partially paired and partially unpaired data](https://stats.stackexchange.com/questions/25941/t-test-for-partially-paired-and-partially-unpaired-data) – kjetil b halvorsen Mar 30 '19 at 08:39
  • I don't think that's really a duplicate, unless the title of the other question is made more general. – Peter Flom Mar 30 '19 at 18:33

1 Answers1

0

  1. Separate the data into 4 groups: Baseline-control, end of study-control, baseline-test, end of study-test. Then check if your measurements are normal distributed or not in each group. If not, try to do some transformations until they look like normal.

  2. Fit a mixed model with baseline vs end of study and control vs test as fixed effect, and the subject specific intercept as random effect.

  3. After model fitting, construct the null hypothesis you want to test.

user158565
  • 7,032
  • 2
  • 9
  • 19