Hey there,

I am using multiple regression analysis to analyse the data of an empirical study in the behavioural science. One of the scales I'm using consists of multiple items that I am combining into a single score to test against the dependent variables. The problem: The scale consists essentially of two sub scales, one of them measured with reflective and the other with formative items. I know, from Structural Equation Modeling, that those multi-dimensional constructs have to be interpreted differently. However, I am stuck with regression on this one and cannot use SEM.

Does anyone know, wether I can combine those items into a single score and use it in a regression analysis. If so, or if not, do you have any paper or source that can explain why (or why not) I am allowed to do that.

I am not sure, if the reflective vs. formative issue even plays a role in regression analysis. If not, please advice. And if you have a paper or textbook explaining, why it is not of interest, I would be so happy, if you could refer it here.

Thank you so much for any helpful guidance!