0

I am stuck with this question and a reasonable answer will be well appreciated. Suppose a numerical data is obtained from the same people by two different investigators. However, we could not link the people, so the pairing is unknown. In addition, nor the investigators could get a hold of all people. For example, investigator 1 could take the measurement of 100 people whereas investigator 2 had it for 79 people. So there are some missing observations. Can I do an independent t-test with these two samples? Thank you in advance!

curiousmind
  • 551
  • 1
  • 5
  • 12
  • 1
    Your question seems to be self-contradictory: if the data are obtained "from the same people" (and are appropriate data for a t test) then the sample sizes must be the same, so how could they vary between the investigators? Evidently your situation differs from your description. To prevent misleading or irrelevant answers from being posted, I have voted to close your question until you can clarify the situation. – whuber Apr 17 '17 at 16:38
  • 1
    @whuber thanks, I have made some edits, hope it is clearer now. – curiousmind Apr 17 '17 at 16:44
  • I am not sure this really counts as `self-study` ("A routine question from a textbook, course, or test used for a class or self-study.") – Glen_b Apr 17 '17 at 16:46
  • I believe this has been asked before; I'm just trying to find a previous version with suitable answers. – Glen_b Apr 17 '17 at 16:47
  • @Glen_b I looked into similar questions. However, those were without an answer. – curiousmind Apr 17 '17 at 16:48
  • Other possibe dup: https://stats.stackexchange.com/questions/25941/t-test-for-partially-paired-and-partially-unpaired-data – kjetil b halvorsen Dec 06 '18 at 23:20

0 Answers0