Download presentation
Presentation is loading. Please wait.
1
Linear statistical models 2008 Binary and binomial responses The response probabilities are modelled as functions of the predictors Link functions: the probit link: the logit link: the log-log link:
2
Linear statistical models 2008 Binary and binomial responses the probit link: the logit link: the log-log link:
3
Linear statistical models 2008 Link functions
4
Linear statistical models 2008 Probit analysis
5
Linear statistical models 2008 Probit analysis proc GENMOD data=linear.rotenone; model Affected/insects=logconc /link=probit dist=bin; run;
6
Linear statistical models 2008 Logit (logistic) regression proc GENMOD data=linear.rotenone; model Affected/insects=logconc /link=logit dist=bin; run;
7
Linear statistical models 2008 Multiple logistic regression Model building principles Significance testing Forward selection Backward selection Stepwise selection Best subset selection Cross-validation
8
Linear statistical models 2008 Odds and odds ratios
9
Linear statistical models 2008 Odds ratios proc GENMOD data=linear.oddsratio; class smoker; freq frequency; model survived=smoker/link=logit dist=bin; run;
10
Linear statistical models 2008 Odds ratios where x is one for smokers and 0 for non-smokers
11
Linear statistical models 2008 Count data and contingency tables and log-linear models Expected frequency: Log-linear models are models of the log expected frequency (log is used as link function)
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.