Hi everyone, grad student in need of help!
I have distributed two surveys, they are very similar but one was for teachers and one for students, as part of a needs assessment for e-learning. I wrote the survey to have variables assessing readiness, enthusiasm, and accessibility.
1) How do I properly assess the reliability of my surveys...the participation rates were low for one, and okay for the other, which makes me wonder whether or not EFA is going to be effective. Alternatively, in SPSS you can run the reliability analysis and get your Cronbach's alpha. How does EFA and the reliability analysis differ?
There are myriad manipulations that can be done in statistics, so I won't address any technical differences of assessments - that may be part of your dissertation. HOWEVER, a caveat in statistics is that the methods ONLY apply to the POPULATION, and not the individual. In practical terms, that means that you really CANNOT generate really meaningful results UNLESS you DO have sufficient random sampling. Small sample sizes must be corrected, not ignored or "calculated" around. If we accept calculations and analyses based on improper bases, we really don't need experiments or surveys at all!
Short answer: Get more data (samples/survey responses)!
Thanks for your reply! I think I knew this when I began and was hoping there was a salvageable option.
Metrics like Cronbach's alpha and McDonald's Omega measure the consistent internal of questionnaires. But, a robust reliability measure is more naturally obtained via Factor Analysis and SEM.
When the sample is small, traditional statistical methods are not recommended. But other methods like Discourse Analysis or Content Analysis.
- Badges
- Science topic