I am analysing a model which has 5 constructs. Each construct has 3 Likert scale questions. For instance, I want to analyse the relationship and effect of the construct "PERCEIVED USEFULNESS" which is the independent variable upon the construct "BEHAVIOURAL INTENTION" which is the dependent variable. Both constructs have 3 questions each.

How do I proceed? I wanted to use Pearson correlation and Multiple Regression analysis. Should I compute the MEAN of each construct and then RUN the tests?

  • $\begingroup$ There are many, many threads on this site about Likert-type data. In terms of your immediate question, whether means make sense for Likert-type is a contentious matter. But you are expected to search for similar questions before posting. $\endgroup$
    – Nick Cox
    Aug 4, 2013 at 7:08
  • $\begingroup$ The way you describe it, structural equation modeling would seem a natural choice. $\endgroup$
    – Gala
    Aug 4, 2013 at 8:07

1 Answer 1


When most people say multiple constructs they mean latent constructs. If that is the case, then you are probably after Structural Equation Modeling (SEM). If so, most packages pull that off well. Lavaan, a package you can download in R, will help you estimate SEM models. On the SEM estimation front, you want to take special care to deal with the fact that these are likert data, so use polychromic correlations as input data for SEM model, or a different estimation procedure that is specific to your data type.


Your Answer

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

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