Is it possible through some form of linear regression to compare the results of two different surveys (same questions) that used two different likert scale lengths (i.e. 6-point = Strongly Disagree, Disagree, Slightly Disagree, Slightly Agree, Agree, Strongly Agree ... and a 4 point scale = Strongly Disagree, Tend to Disagree, Trend to Agree, Strongly Agree)

Please know I am not a stat's wiz so I'll need a simple instruction if possible.

  • $\begingroup$ I think it's reasonable to think that the participants in the first survey that answered Disagree or Slightly Disagree would answer Tend to Disagree if they were given the second survey. Any reason you don't just want to collapse the 6-scale into the 4-scale like that? Is there any particular reason you want a linear regression solution? $\endgroup$ Sep 17, 2014 at 6:15
  • $\begingroup$ I like it! Sounds like a simple solution. I had done some research and found this article indicating that regression would be preferable over a simple mathematical equalization. But your suggestion sounds like a reasonable approach. Here's the link: rangevoting.org/comprati.pdf $\endgroup$
    – Susan
    Sep 17, 2014 at 15:35


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.