The null hypothesis of the White’s test states that the residual error terms are
homoskedastic (no violation of the OLS assumption – this is good!). WinORS reports both the White statistic and
its associated P-value. A rejection of
the null hypothesis at the 95% confidence level would require a P-value of 0.05
or smaller (this is a violation of the assumption – bad!). That is, if the computed p-value is greater
than 0.05 (or some acceptable cutoff level), then do not reject the null
hypothesis (this is good!). When the
homoskedasticity assumption is violated, White’s test does not suggest what to
do next. However, we suggest trying weighted ordinary least squares.