The proportion of explained variation (R²) is frequently used in the general linear model and it becomes also frequently used in Poisson regression models. We present a SAS macro (R2POI), which calculates unadjusted and adjusted R²-measures based on the log-likelihood and on sums-of-squares for Poisson regressions. The adjusted versions of both measures should prevent the inflation of R² in small samples and/or many prognostic factors.

 

References:
Mittlböck, M. (2002): "Calculating Adjusted R² Measures for Poisson Regression Models", Computer Methods and Programs in Biomedicine 68, 205 - 214
Mittlböck, M., Waldhör, T. (2000): "Adjustments for R²-Measures for Poisson regression models", Computational Statistics & Data Analysis 34, 461 - 472

Currently there are two SAS macros available: