0

I have two tests on the same groups for students, may be evaluating an education program. If we can paired the same groups of students, this would be paired t-test.

But for privacy reasons, so that I cannot associated the test result of individuals and hence I cannot pair them.

Statistically what test for comparing mean scores can I do for these type of samples? Thank you

PS. I checked previous questions, seems that they can still paired, but just the matter of interoperation.

PS2. I checked again as suggested. I think the following post partly answer my question. Analysing pre-and-post intervention study with anonymous responses

That is, the mean is still the same, and I am thinking to use bootstrap to generate the CI for difference.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Mahali Sindy
  • 101
  • 1
  • [This is a FAQ](https://stats.stackexchange.com/questions/25941/t-test-for-partially-paired-and-partially-unpaired-data). There see at the right margin, headers **Linked** and **Related**, where you find many related Qs. – kjetil b halvorsen Mar 26 '21 at 11:10
  • I am not sure how these two questions are linked. 1) mixed pair and unpaired data and 2) paired data but missing pair information. – Mahali Sindy Mar 27 '21 at 00:36
  • This [stored search](https://www.google.com/search?channel=fs&client=ubuntu&q=site%3Ahttps%3A%2F%2Fstats.stackexchange.com+paired+data+with+missing+pair+information) might be useful! for instance https://stats.stackexchange.com/questions/109731/paired-t-test-when-pairing-is-not-known-and-some-subjects-responded-only-in-on and https://stats.stackexchange.com/questions/370953/analysing-pre-and-post-intervention-study-with-anonymous-responses/373979#373979 – kjetil b halvorsen Apr 03 '21 at 18:11

0 Answers0