There were ten questions in a pre intervention test given to 40 members. The correct answer was marked as 1 and the wrong one 0. The content validity was tested by a subject expert. Can anyone suggest ways to improve the value. The output as a word document is attached herewith. Thank you
Hi,
A negative correlation indicates the need to recode the item in the opposite direction. or you need check whether your sample size is sufficient. Follow the below link you have some more information.
https://www.researchgate.net/post/Why_do_I_get_a_negative_value_of_Cronbachs_Alphaa
Kind regards
Thushara
When the item total score correlations are examined, the results are quite low. For all items, this value is under .30, which is the rule. This indicates that some of the items need to be removed from the scale and others must be redacted. My first impression is that students may have responded randomly. Or the reverse scored items may have been calculated without conversion.
I agree with Ibrahim. Because your items are very poor. item discrimination indexes have negative value. So this means that your items discrite people negative way. Hence succeed student will have low score than not succeed students.
In addition, Cronbach alpha calculates via variance of items and test scores. If you have nagative value of Cronbach alpha, You can evaluate it as zero. But this means that the items are not related with each other. If you sure students may have responded randomly or the reverse scored items may have been calculated without conversion as @Ibrahim Uysal said you have really bad items. May be you can re-write items or use other scale for your purpose.
Thank you very much for your replies. Please excuse me if my question is silly. The factor analysis is NOT on scale items. It is either they know the answer or they don,t. Recoding the question will not help here, I think. Yes I agree, there is reverse score pattern but that is because of item difficulty at which my intervention will be focussed. As I have mentioned it is a pre test with dichotomous results. Is there any other reliability test for such type of questions?
- Badges
- Science topic
- Similar topics
- Psychology
- Psychometrics