Am 07.07.2022 um 07:49 schrieb Riccardo (Jack) Lucchetti: Fred, it would be helpful to be more specific about what is actually needed, like what kind and size of dataset needs to be analyzed with what regressors/predictors. I kind of doubt that your students really require all the features and capabilities of that package. Then maybe we can come up with a practical suggestion. To avoid false impressions, we do agree that fixed-effects estimation in linear models is not in principle a problem, don't we? From what I understand, the problem appears only when the datasets get huge, right? I have no idea how that Stata package actually handles things, but I kind of suspect that Stata's memory-saving way of storing dummy variables may play a role. As long as gretl stores every dummy as a double (FP64), I'm skeptical that really large datasets can be handled, with several fixed-effects layers. (Don't get me wrong, I'm not pushing for gretl to change its underlying architecture there, which might be error-prone and divert resources from other activities.) cheers sven