I feel r moving in right direction.

Hi, you could try to use factor analysis to all the 5-scale likert items in your questionnaire and afterwards use the new variables (factors) in the regression model or hypotheses tests, as long as they are accurate and validated by the theory.

1. To combine 3 or 4 questions, you can use weighted average of the answers on Likert scale to these questions with weights depending on the importance of the question in the new variable.

2. For the 9 variables generated from 32 questions, you can use ordinal logistic regression model

Dear Ajit kumar Roy thank 4 ur response

Dear Shashi Ajit Chiplonkar thanks. does average of Likert scale have correct meaning? for example 4 means agree and 5 means completely agree but what does 4.3 or 4.8 mean?

Dear Stamatis Ntanos thank you. Could u pls introduce me a good reference?

Dear Pegah, a similar methodology to the one I proposed earlier (first factor and then regression), is presented in my article entitled :

Article Public Perceptions and Willingness to Pay for Renewable Ener...

You can use this article as reference in your research.

Dear Pegah,

If you have designed a series of questions that when combined measure a particular trait or attitude or beliefs, you can use means and standard deviations to describe the scale. By calculating a composite score (sum or mean) from four or more Likert-type questions, the mean would be interpreted in terms of the low or high scores. The composite score for Likert scales should be analyzed at the interval measurement scale. Additional data analysis procedures appropriate for interval scale items would include the Pearson's r, t-test, ANOVA, and regression procedures.

PLS-SEM is one of the best choices in your case.