➤ How do you interpret a binary variable in a log-linear house price regression?
In a model with log house value and a dummy variable, the coefficient on the dummy gives the percentage difference in value between groups. The exact percent change equals 100*(exp(beta)-1). For small beta, you can approximate percent change as 100*beta. This adjusts for semilog transformation and yields an intuitive effect size.
Leave a Reply