Logistic Regression Using SAS: Theory and Application

Key differences between 2nd and 1st editions

• Time lapse between previous and current publications: 13 years (1999 vs 2012).
• The Second Ed. has been updated with the latest syntax and options in SAS 9.3
• Chapter 2 "Binary Logistic Regression with PROC LOGISTIC" inclues new sections:
• CLASS Variables
• Multiplicative Terms in the MODEL Statement
• Chapter 3 "Binary Logistic Regression: Details and Options 3.1 Introduction"
• Explanation of two major new features of LOGISTIC -- exact logistic regression and penalized likelihood
• New EFFECTPLOT statement for getting useful graphs of predicted values
• Graphs of odds ratios
• New R-squared measure
• New diagnostic graphs
• New section "ROC Curves"
• New section "Plotting Effects of Predictor Variables"
• In the Chapter 4 "Logit Analysis of Contingency Tables" PROC LOGISTIC is used to estimate the contingency table models rather than PROC GENMOD.
• In the Chapter 5 "Multinomial Logit Analysis" PROC LOGISTIC is used to estimate the multinomial logit model rather than PROC CATMOD.
• Chapter 6 " Logistic Regression for Ordered Categories" demonstrates how to use PROC QLIM to model heteroscedasticity in the cumulative logit model
• Chapter 7 " Logistic Regression for Ordered Categories" includes new section "More Advanced Models with PROC MDC"
• Chapter 8 "Logit Analysis of Longitudinal and Other Clustered Data" includes new topics on:
• New section "Robust Standard Errors".
• New section "Mixed Models with GLIMMIX".
• Discussion of using the alternating logistic regression to get an odds ratio parameterization of the correlation structure for PROC GENMOD.
• PROC LOGISTIC is used instead of PROC PHREG.
• Chapter 9 "Regression for Count Data" includes new section "Zero-Inflated Models".