Computing measures of explained variation for logistic regression models

Authors
Mittlbock, M Schemper, M
Citation
M. Mittlbock et M. Schemper, Computing measures of explained variation for logistic regression models, COMPUT M PR, 58(1), 1999, pp. 17-24
Citations number
8
Categorie Soggetti
Multidisciplinary
Journal title
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
ISSN journal
01692607 → ACNP
Volume
58
Issue
1
Year of publication
1999
Pages
17 - 24
Database
ISI
SICI code
0169-2607(199901)58:1<17:CMOEVF>2.0.ZU;2-G
Abstract
The proportion of explained variation (R-2) is frequently used in the gener al linear model but in logistic regression no standard definition of R-2 ex ists. We present a SAS macro which calculates two R-2-measures based on Pea rson and on deviance residuals for logistic regression. Also, adjusted vers ions for both measures are given, which should prevent the inflation of R-2 in small samples. (C) 1999 Elsevier Science Ireland Ltd. All rights reserv ed.