I have a multiple questionnaires with responses on Likert scales. They differ in that some are rated on a 7p scale, and others on a 5p scale. I would like to look at the reliability of each scale as well as the correlation between scales (and sub-scales).

By standardization I mean computing the z-score (subtract mean, divide by standard deviation), or should I merely normalize the range (scale every item to $[0,1]$), for better comparison?


You would definitely want to put the 5 and 7 point scale items onto some sort of a common range if you're trying to combine them into a single scale and assess reliability. The SPSS RELIABILITY procedure will actually produce both "raw" alpha based on variances and covariances, as well as a standardized item alpha based on correlations with standard unit variances if you do anything to invoke the construction of a covariance matrix. This can be done directly in command syntax by including


or in the dialogs by asking for any of a variety of statistics, including a covariance or correlation matrix.


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.