Quandt-Andrews test

[Alex]

I have performed Quandt-Andrews test using Eviews 6 (after downloading the patch available from your website).
My question relates to EXP Wald F statistics. I would expect that higher test statistic values are associated with lower Prob. values. In the tests I have done it is not always the case. For example, one regression gives me the test statistic value of 38.55 and Prob. of 1.0; a different regression (different dependant variable & regressors) gives me the test statistic value of 28.40 and Prob. of 0.01.
It is possible that some of the series I have “misbehave” or that EViews does not calculate the Prob values correctly. I would like to know whether anyone else reported this issue and, if this is an EViews issue, whether it has been fixed in version 7.

[EViews Gareth]

Could you post your workfile?