- Shortly after this in the class we stated with Difference-inDifference (DID) with panel data we can inference the [ ] by creating the interaction terrn. We used the same term when we said "with the DD] with panel data you can touch the [ ]". (c) Why do we square experience term and not the other independent variables in the wage equation? (d) Aren't we always going tn have omitted variables when trying to explain something with statistical models? Is it ever possible to capture every relevant independent variable? If so, could this mean that lDLS is never really a valid method of explaining something? (e) If the estimator achieved by ZSLS has a higher SE and doesn't appear to be as signicantly signicant as does the estimator achieved by 0L5, why do we prefer it