It is established that questionnaires reliability is tested & defined on SPSS by the Cronbach's alpha. my question is: how do i test the reliability of a test which has correct and incorrect answers and not a just scale (1-5)?
I initially tried doing it using SPSS (the same way i do for questionnaires) but that gave me a very low negative answer, so i am trying to explore different options.
Thank you in advance.
#SPSS #Statistics #ExperimentalStudy
In class one element could be a chapter on experimental design. As a learning activity, you give the class an exam. However, you make five different versions of the exam (different questions, different phrasing, different order). On each of the following five days all students take all exams. One would hope that the distribution of scores in the class is independent of the exam, and that the score for each student is the same for all versions of the exam. That would be a fun experiment.
I would expect that some students will have high variability in test scores. They way the questions are asked and the order that they are asked has a strong influence on that student's performance.
Hello Abdelbar,
If your interest is that of internal consistency reliability, Cronbach's coefficient alpha works the same way regardless of whether scores are dichotomous (e.g., right/wrong) or polytomous (1, 2, 3, etc.). A low estimate can arise from: (a) item sets that cover more than one dimension or attribute; (b) items in need of modification or deletion; (c) respondents who do not pay attention to the items in giving their answers; (d) little or no variance across respondents; or (e) some combination. Do understand that negative estimates of Cronbach's alpha are possible (though theoretically no different from zero), as the computation formula involves subtraction.
In any event, a low estimate of reliability would suggest that it is unwise to consider respondent behavior/performance/endorsement as being consistent across items.
Good luck with your work!
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS