Using ROC Analysis to plot ROC and PR curves

A principal investigator recruited 2955 participants including 1726 females and 1229 males. 762 females did not have any type of insurance while 445 males did not have any type of insurance. A binary index variable is used to denote whether a participant has insurance or not. The investigator collected the age information for all participants at the moment the survey was conducted, and desired to plot both ROC and PR curves separately for the females and males. Furthermore, she asked a clinical data analyst on her team to conduct an appropriate hypothesis test to make inference regarding the difference between the two ROC curves.

This information is collected in meps1.sav. See the topic Sample Files for more information.

Next