I have a data set with some demographics of consumers who bought a product that can be used to imply their preference (beta) using Cobb-Douglas (see comments of original question). I’d like to check if any of the other features (specifically household size) have a significant impact on the derived beta of the consumer. The eventual goal is to incorporate the results into an agent based model when generating the preference variable of the agents (which would then be used to generate the Cobb-Douglas function unique to that agent).

The income (budget) of the consumer plays a role in the derivation of the beta so perhaps it’s better to measure the effect of these features on income instead of the preference?

I’m familiar with the standard regression techniques, but wanted to make sure there wasn’t any issue with running a regression against a derived variable.

  • $\begingroup$ Without coming off as too nosy, may I ask what general type of project you're using this for? I only ask because depending on the type of project and intended audience, different modeling assumptions may be necessary, which could impact how you set up the regression. For example, if this is an empirical IO project, I think you may have to spend a lot of time defending the CBUF choice. If this is a theory/field experiment paper for decision theory, then assumptions about decision making under uncertainty should be clarified. (Etc.) Thanks! $\endgroup$
    – AndrewC
    Jul 20, 2018 at 21:54


Your Answer

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