I am conducting a Cronbach Alpha calculation for 9 participants on a 96 item Stroop Test and received this warning from SPSS. The alpha value reported was 0.83. But with that warning does that mean the value is not acceptable? The program itself had omitted many items because they were all answered the same. Would KR-20 be more suitable as an alternative to Cronbach since the participants are scored how many items they answered in 45 seconds and must correct their answers before proceeding. So it's correct in 45 seconds, and there are the remaining questions not answered in 45 seconds.
With only 9 participants you will have difficulty with any method trying to estimate the "reliability". Reliability in quotes since Cronbach's alpha is often used for many purposes (beyond what it actually does). Assuming you get a lot more participants, you might consider if this is the measure that you want. I assume your data are binary. You might think about types of IRT.
- Badges
- Science topic
- Similar topics
- Correction