STATISTIK-FORUM.de

[gunman110]

Ich verstehe folgende Konsequenz von Heteroskedastizität nicht ganz.
Kann mir das einer Erklären?

Variance of the OLS estimator is inflated:
OLS may attribute systematic variation in the dependent variable to the independent variables, even though this variation is really due to the error term. Because positive “mistakes” of this kind are as likely as negative ones, the estimator remains unbiased. But because big “mistakes” are more likely, the sampling distribution of the OLS estimator is more spread out.

[DHA3000]

Hmm, das scheint mir der Versuche einer inhaltichen Erklärung zu sein. Allerdings eher unglücklich. Wo hast du das denn her?