I am analyzing the responses of a survey using spss statistics. I understand that I will need to do a Cronbach alpha first. But here is my problem:

The first part of my survey is "Awareness" and second part is "Attitudes". Under "Attitudes",there are 6 likert scale questions and 4 yes/no questions. So i was told to do data normalization before calculating Cronbach alpha for "Attitudes".

How should I do data normalization in spss?

Is data normalization same as transforming data in SPSS (arithmetric eg. Log10 or square root)? If so, how should I transform the yes/no questions? (my understanding is we can only transform likert-scale question)

I found very limited information in data normalization using spss.

Your answer is deeply appreciated. Thank you so much!!

  • 2
    $\begingroup$ "[I] was told to do data normalization": what did your boss say exactly. It's hard to see how Likert-scale questions can be "normalized". $\endgroup$
    – Nick Cox
    Nov 2 '15 at 11:46

To normalize means to rescale so that the range becomes [0,1]. See this answer. Normalizing to bring your [1,5] responses into a [0,1] range will not help you assess reliability. (They're not Likert responses--they're ordinal ratings. See here.)

If you are willing to run the Cronbach's alpha procedure on the 1-5 ratings, you arguably can also do so on the larger group of items that includes the binary items. It may not have been Cronbach's original intent to use non-continuous data, but think of it like computing a correlation using two binary variables. Instead of calling it r you would call it phi; it serves the same purpose and in many situations is an adequate indicator.


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.