EViews.com

So Gareth, is there a fix to the bug in Eviews 6. Also, we have observed that even when the data looks quite stable using the CUSUM curves, the Quant-Andrews BreakPoint Test in Eviews 7 displays a probability that indicates a significant confidance that an addtional breakpoint exists. Eviews 6 would give a probablity of 91.3% while the same anlaysis in Eviews 7 yields a probablity of 0.77%. Can Eviews 6 be that far off?

Possibly. There is no fix for EViews 6 (since that product is now two versions out of date).