Re: [Gretl-users] issue defining a panel with sub-sample data

On Tue, 1 Oct 2013, Riccardo (Jack) Lucchetti wrote: [...] Yes, here's the point. If you subsample a panel dataset so that it becomes unbalanced, then insist on turning it into a panel again, we have to rebalance it by inserting "padding" rows of NAs. Then if you later do "smpl full" we have to remove the padding and stick back in the original values, while carrying back to the full dataset, on the right rows, any data added or revised while the subsampling + padding was in effect. This is on the very edge of what we can do reliably (moreover, it's not 100% clear how to define "doing it right"), so after Artur's report of a segmentation fault (referenced in the posting that Jack cites above -- Artur, you should remember this ;-) I decided to go for "safety first" and ban that sort of rebalancing via padding. In fact, however, there's a "back door" which allows you to do this: when you impose the unbalancing restriction, add the --balanced option to the smpl command! This whole thing needs some more investigation. I guess for consistency we should either scrap the --balanced option or reactivate the balancing mode that's now disabled. Whenever I try to think this issue through it makes my head hurt, but I'll see what I can do... Meanwhile, of course, the safe option is to "store" the subsampled dataset, reopen it, and apply --panel-vars at will. Allin