[Gretl-users] Re: gretl vs Stata: different WLS estimates for weights=0

Le mer. 20 janv. 2021 à 14:35, Riccardo (Jack) Lucchetti < p002264(a)staff.univpm.it&gt; a écrit : Dear Riccardo, Not quite sure if this is what you asked for but here I am. <hansl> open greene12_1.gdt list lx = const income ownrent selfemp # some arbitrary weights set seed 100 w = abs(normal(0,1)) wls w expend lx # some arbitrary 0 weights w0=w w0[5]=0 w0[10]=0 w0[20]=0 wls w0 expend lx series uhat=$uhat series yhat=$yhat smpl w0==0 --restrict print expend w0 yhat uhat --byobs </hansl> Then, I exported the same data file to stata and executed the respective Stata commands which provides the output below (to compare bay analogy, with gretl's. Hope it helps, Best, Artur . use "/Users/user/Desktop/greene12_1.dta", clear (Written by gretl 2021a) . reg expend income ownrent selfemp [aw=w] (sum of wgt is 8.2284e+01) Source | SS df MS Number of obs = 100 -------------+------------------------------ F( 3, 96) = 5.65 Model | 949877.544 3 316625.848 Prob > F = 0.0013 Residual | 5382083.03 96 56063.3649 R-squared = 0.1500 -------------+------------------------------ Adj R-squared = 0.1235 Total | 6331960.58 99 63959.1977 Root MSE = 236.78 ------------------------------------------------------------------------------ expend | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- income | 35.13955 14.01452 2.51 0.014 7.320954 62.95814 ownrent | 114.1856 52.52346 2.17 0.032 9.927328 218.4438 selfemp | 11.53469 110.0354 0.10 0.917 -206.8838 229.9532 _cons | 8.651678 52.43104 0.17 0.869 -95.42311 112.7265 ------------------------------------------------------------------------------ . reg expend income ownrent selfemp [aw=w0] (sum of wgt is 7.9537e+01) Source | SS df MS Number of obs = 97 -------------+------------------------------ F( 3, 93) = 4.25 Model | 734603.026 3 244867.675 Prob > F = 0.0074 Residual | 5361805.11 93 57653.8184 R-squared = 0.1205 -------------+------------------------------ Adj R-squared = 0.0921 Total | 6096408.14 96 63504.2515 Root MSE = 240.11 ------------------------------------------------------------------------------ expend | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- income | 32.59137 15.52483 2.10 0.038 1.762137 63.42059 ownrent | 109.6996 53.48384 2.05 0.043 3.491276 215.9079 selfemp | 16.04231 112.8278 0.14 0.887 -208.0113 240.0959 _cons | 20.00571 56.5598 0.35 0.724 -92.31083 132.3223 ------------------------------------------------------------------------------ . quietly predict double yhat . quietly predict double uhat, resid . list expend w0 yhat uhat if w0==0 +--------------------------------------+ | expend w0 yhat uhat | |--------------------------------------| 5. | 546.5 0 448.77476 97.725243 | 10. | 52.58 0 124.29808 -71.71808 | 20. | 0 0 130.81635 -130.81635 | +--------------------------------------+