Hi all, a user brought to my attention the following slight glitch: assuming you have the "Wooldridge" dataset collection installed, the following script <hansl> open fertil3.gdt setobs 1 1913 ols gfr time tsq gfr_1 </hansl> produces an ols printout where gretl displays the DW statistic instead of Durbin's h, as it should (since the lagged dependent variable is among the regressors). This is no gretl bug: it's a consequence of the fact that the "fertil3.gdt" dataset does not contain the proper calendar information and, more importantly, the lagged variables are stored in the dataset but are not marked as such. A slight modification of the above script that deals with this problem could be <hansl> open fertil3.gdt setobs 1 1913 delete gfr_* ols gfr const time gfr(-1) </hansl> where the lags of "gfr" are cleaned first and then regenerated on the fly as needed. Nevertheless, it would be nice if gretl shipped a version of this dataset in which this issue is fixed. Is there anyone who uses regularly the Wooldridge example files who could spend an hour compiling a list of the data files that need fixing? ------------------------------------------------------- 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 -------------------------------------------------------