Dear Chiara,

I believe it's OK to conduct correlations analysis between the two sets of rating data, as correlation relates to the issue of similar rank-ordering between two variables.

However, it may not appropriate to run other inferential statistical analyses such as analysis of variance. I would suggest transforming both sets of the rating data to standardized Z scores before statistical analysis. An alternative approach is to transform the ordinal rating data to interval data (in logit) via Rasch modelling usually implemented through Winsteps or Conquest.

You may also consult the literature that discusses the reliability issue rising from the use of scales of different lengths.

Good luck to your research.

Cheers,

Chao

Dear Chiara,

You can run correlation and regression analysis, if your data has been tested and proved for normality.

And yes, you can use dummy variable for standardization .

Best,

Praheli

There is a difference between a Likert-scored item that uses a given number of reference points and a scale that combines a number of such items to create an overall measure. Assuming that you are using true scales (and not individual items), then you can correlate them as would any other continuous variable, regardless of the number of underlying "points."