Cuando ejecuto este código:
require(nlme)
a <- matrix(c(1,3,5,7,4,5,6,4,7,8,9))
b <- matrix(c(3,5,6,2,4,6,7,8,7,8,9))
res <- lm(a ~ b)
print(summary(res))
res_gls <- gls(a ~ b)
print(summary(res_gls))
Obtengo los mismos coeficientes y la misma significación estadística en los coeficientes:
Loading required package: nlme
Call:
lm(formula = a ~ b)
Residuals:
Min 1Q Median 3Q Max
-2.7361 -1.1348 -0.2955 1.2463 3.8234
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 2.0576 1.8732 1.098 0.3005
b 0.5595 0.2986 1.874 0.0937 .
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 2.088 on 9 degrees of freedom
Multiple R-squared: 0.2807, Adjusted R-squared: 0.2007
F-statistic: 3.512 on 1 and 9 DF, p-value: 0.09371
Generalized least squares fit by REML
Model: a ~ b
Data: NULL
AIC BIC logLik
51.0801 51.67177 -22.54005
Coefficients:
Value Std.Error t-value p-value
(Intercept) 2.0576208 1.8731573 1.098477 0.3005
b 0.5594796 0.2985566 1.873948 0.0937
Correlation:
(Intr)
b -0.942
Standardized residuals:
Min Q1 Med Q3 Max
-1.3104006 -0.5434780 -0.1415446 0.5968911 1.8311781
Residual standard error: 2.087956
Degrees of freedom: 11 total; 9 residual
¿Por qué ocurre esto? ¿En qué casos las estimaciones OLS son iguales a las GLS?