Please remove me from this mailing list.
-Nathalie Chernoff
________________________________________
From: Research of postgraduate psychologists. [[log in to unmask]] on behalf of Dan Holman [[log in to unmask]]
Sent: 14 November 2011 1:33 PM
To: [log in to unmask]
Subject: R2 measures for binary logistic models
Is it acceptable to use these as rough estimates of explained variance for binary logistic models? I know there is no standard measure for such models and some disagreement but I only want an approximate measure.
What is the most appropriate measure (with mostly sociodemographics as predictors)? Or should I be looking at some other statistic besides the pseudo r2 values?
Is there an adjusted version of these as there is with OLS regression?
To what extent could I compare the measure the explained variance of binary models with linear models?
As I said, I have many, many models to compare and precision is not the goal.
Many thanks
Dan
|