Outlier elimination when comparing models in nonlinear regression.

Prism can identify and remove outliers when fitting a model with nonlinear regression. But what happens when you ask Prism to compare two models? 

What we intended is for Prism to:

1. Fit the first model, and identify outliers from that fit.
2. Fit the second model, and identify outliers from that fit.
3. Remove both sets of outliers.
4. Fit both models and compare the fits.

In fact, that is not what Prism 5 does, nor Prism 6 (including 6.01 and 6.0b). For these versions, Prism does not do step 1. It only identifies and removes outliers from the second model -- the one chosen on the Compare tab. That means that simply swapping the two models (the one chosen on the Fit tab, and the one chosen on the Compare tab) can (in rare cases) affect the results. 

Prsim 6.02 and 6.0c will fix this issue, and properly remove outliers from either fit.