Chow Test for Subset of Parameters

[Mila]

Hi,
I have what is probably a basic question, but it's really got me confused. I really hope an eviews expert could provide some clarification.

I'm running chow tests for a series of dynamic time series that have dummy variables in them, and I'm interested in ascertaining if instability arises in terms of particular parameters.

I understand that I can't perform the test by putting all regressors including the dummy variables in the "regressors to vary..." box, because there will possibly be multicollinearity with the constant in one subsample.

I choose to vary only certain parameters - the ones I'm interested in. I'm just wondering, how is it that I'm able to do the test, by only choosing selected regressors? What regressions is Eviews running in the background? Am I to assume that the restricted and unrestricted regressions do not contain any dummy variables? In which case, the restricted regression no longer corresponds to my initial restricted regression! And therefore may display instability because I haven't excluded the outlier.

Please help!
Many, many thanks.