Print

[lenderguy]

I do not believe stratification (using the "by" option in SAS proc logistic) counts as controlling for a variable (I will read up on this in the near future - I don't see a problem with it, its just I've never seen anyone do it that way, and there's usually a reason for that).

Maybe I used the wrong word?  But if I just want to examine things in different buckets, isn't that good enough?

All variables that are considered to be controls are entered as independent variables in the model in step 1. In step 2 the variable of interest is enter. If the Type III Sums of Squares is significant, then you've got something.

How do I get the Type III sum of squares in SAS?  I saw in my output once, but then again, SAS prints out a lot of output...

[Staneslav]

a

[nonattender]

I guess my point is that "everybody knows" an AA won't take a loan at 29%, and is waiting to get it bid down.  But the one time you're wrong, you won't know until it's too late.

Partially why we (anciently) requested an upper rate boundary on SO's (you know, back in ancient days when we thought SO's were viable).

In the most general terms:  I might bid on AA's at a 10% minimum rate, but only on AA's who the "market" judges to be non-ridiculous and bids down to at least 16%.  This could have been set up as a "trigger rate" or an "upper bound" or any number of other ways, but... nope.

-t