I am doing my PhD on community empowerment initiatives of faith- based NGO using mixed methods research design. My total sample is 165. The overall Cronbach's alpha score of data collected through structured questionnaire is 0.934. There are nine parts in my questionnaire. In three parts the scores are 0.618, 0.652 and 0.693. Is it acceptable in social science research? If it is then please give me some references.
Shofiqur - best to see Chronbach's as individual scores. Two of your scores are at low levels. Not sure how the sum of all three ranks as high as 0.9?
Dean Whitehead has a good point. You'll want to look at the alpha levels of each scale individually. The reason why the scale with all of the items is so high may be because of the number of items. Cronbach's alpha is a function of the correlation between all of the items and the number of items in the scale. If you have a scale with a very high number of items, then they'll tend to have a higher alpha level just because of this. I recommend looking at the alphas if items are removed and seeing if there's a way to optimize it by removing them. Alternatively, you could use an exploratory factor analysis to see how the items go together. It could be that the scales that you think you've created would be better off reorganized.
Those indices of reliability seem to be in the limit, I would recommend the coefficient of maccdonald that takes into account the relation of the factors that indicates a more adequate value of reliability, there are articles that mention reliability greater than .60 also I would recommend to make a factorial analysis to confirm the structure of the instrument.
I would begin with an Exploratory Factor Analysis using an oblique (correlated factors) rotation and either Principal Factors or Maximum Likelihood extraction. The goal there is to determine whether your observed data actually matches the three factors that you proposed.
If so, I would use the options in the procedure for alpha to examine "alpha if item deleted" to determine whether some of the items were lowering the alpha for a scale (which is equivalent to an item having a low loading on the corresponding factor).
Assalamualaikum,
Please try to read these articles which will enrich your understanding about the minimum C'alfa scores
Article The change and stability of NEO scores over six-years: A Bri...
Article Development and validation of the unifying elective orthopae...
generally, you want alpha of 0.70 or higher. With those, I would say that they are marginal and perhaps more scale development work is needed.
- Badges
- Science topic
- Similar topics
- Philosophy
- Ethics