On Thu, 31 Dec 2020, Ivan Falconi wrote: Ivan, you (or somebody using your login to the gretl package upload site) has been busy uploading lots of packages written by other people and already published on the server, as if they were new submissions. This is a nuisance and waste of time for the gretl team. If you have been doing this, please stop. If it wasn't you, please to try find out who's using your login. An autocorrelation test is not about homogeneity of variance, it's about serial correlation of the error term, which is generally not meaningful in the context of cross-sectional data. If you want to test for homogeneity of variance you have a number of heteroskedasticity tests available, for both cross-section and time series data. Allin Cottrell

On Fri, 1 Jan 2021, Ivan Falconi wrote: From the help: --------------------------------------------------------------------------- Function packages Gretl's functionality can be extended by the use of function packages. These come in two sorts: official “Addons” and contributed packages. Jointly, they cover many estimators and utilities not available as built-in commands or functions. [...] You can browse the contributed packages installed on your computer via the menu item On local machine, and if you are online you can access a listing of available packages via the item On server. Both items are found under /File/Function packages. [...] For full details on installing and working with function packages, see Function package guide (under the Help menu). This guide also contains details on writing function packages. A nice, short take on package writing can be found at http://gretl.sf.net/gfnguide/gfn_for_dummies.html. --------------------------------------------------------------------------- So uploading is something you'll want to do if you have written a new function package. If you just want to use an existing one, no upload is necessary. We're glad you find gretl more intuitive than Stata. Of course, intuitiveness is closely related to a person's own taste, so... As for Stata being "more powerful": in my personal opinion, it depends on your definition of "powerful". Certainly Stata has a larger user base and arguably a larger array of econometric techniques that come "pre-cooked" and ready for the end user (my feeling is that this is especially true in microeconometrics, though perhaps not so true for time-series analysis or macroeconometrics). However, I believe that the programming facilities (matrix manipulation, optimisation techniques, random number generation...) that gretl offers for coding new estimators, tests etc. are not inferior to Stata's. If you ask me, the area where Stata has an edge on all other packages (not just gretl) is the quality of the documentation, that I find simply stellar. ------------------------------------------------------- Riccardo (Jack) Lucchetti Dipartimento di Scienze Economiche e Sociali (DiSES) Università Politecnica delle Marche (formerly known as Università di Ancona) r.lucchetti(a)univpm.it http://www2.econ.univpm.it/servizi/hpp/lucchetti -------------------------------------------------------