1

My study involves a pre- and post-test knowledge score.

However, due to participants not understanding the unique identifier code generation, some of the data collected is not able to be paired.

Some successfully used the unique identifier code generation, and we have some paired data.

How can i incorporate both of these into analysis?

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
user333462
  • 11
  • 1
  • If the mistakenly unpaired data dominate, I would analyze all the data as unpaired. Otherwise I would analyze only paired data and throw out the unpaired. – ttnphns Aug 29 '21 at 12:30
  • There is also a way through split-plot analysis in ANOVA with respondents factor nested in the pre-post factor. I've never tried it myself, though, as a way to cope with the situation you describe. – ttnphns Aug 29 '21 at 12:36
  • 1
    See https://stats.stackexchange.com/questions/25941/t-test-for-partially-paired-and-partially-unpaired-data, https://stats.stackexchange.com/questions/215769/how-to-analyse-paired-data-lacking-pairing-information?noredirect=1&lq=1 solves maybe the problem for t-test. Please augment your question, what whould you have used if pairing was complete? – kjetil b halvorsen Aug 30 '21 at 15:55

0 Answers0