1
$\begingroup$

I'm working on data analysis for my dissertation, and I'm banging my head against the wall when it comes to Cronbach's alpha.

I just completed a pretest; after data cleaning, I have 119 valid and complete responses. These 119 participants answered the same six-item Likert-type scale for nine different scenarios.

However, I am not sure at all how to calculate Cronbach's alpha in this case. Do I calculate alpha for each of the nine scenarios then average them together? Or should I have SPSS calculate new variables, with each of the nine instances of the six items compiled together, and then calculating Cronbach's alpha from that?

...or is there a better calculation altogether?

$\endgroup$
7
  • $\begingroup$ Are the six Likert-type items designed to measure the same underlying dimension/construct? What do you mean by 'nine different scenarios'? $\endgroup$ – Ayalew A. Jan 26 '15 at 7:18
  • $\begingroup$ Good question. The underlying construct relates to perceptions of privacy in specific physical environment, so the same six questions were asked about nine different environments. $\endgroup$ – Eaners Jan 26 '15 at 7:26
  • $\begingroup$ That means you have 6*9=54 items? $\endgroup$ – Ayalew A. Jan 26 '15 at 7:40
  • $\begingroup$ Meaning a single item is asked nine times: for environment x, for environment y, etc? $\endgroup$ – Ayalew A. Jan 26 '15 at 8:58
  • $\begingroup$ Correct--the same six items are asked nine times each. $\endgroup$ – Eaners Jan 26 '15 at 16:14
0
$\begingroup$

It seems that your intention is to measure 'perception of privacy' under each specific physical environment separately. In that case, your use of Cronbach's alpha would be to measure the internal consistency of items used to measure the 'perception of privacy' of your respondents in each environment. Hence calculate alpha for each physical environment and report your findings separately for each environment. You cannot average the alphas obtained from each scenario because that average doesn't make sense. It is also not possible to sum the ratings for all items across all possible environments to come up with a 'global perception of privacy' score because basically the scale you used is a unidimensional scale applied under different circumstances.

Also look at this post on CV.

$\endgroup$
1
  • $\begingroup$ Makes sense to me--and this is also what one of my committee members suggest that I do. Thanks! $\endgroup$ – Eaners Jan 27 '15 at 22:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.