I'm working with a random effects model for a panel dataset for a number of retailers and their daily sales. Since I am trying to infer the effect of an interaction between two independent variables for the different retailers, running a random slope for this interaction became too complex and so I switched to the approach of modelling each retailer separately.

Since the data for each individual retailer won't have any of the random effects that I was modelling (they were all based on 1|Retailer), I was trying to run a plm(model="within") fixed effects model for each one. However, since each of these subsets is a time series dataset rather than a panel one, the formula keeps telling me that the returned model is empty. Is my approach wrong? Should I aim to approximate each one simply using OLS with a factor() control variable for months and years?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.