Logistic Regression Using SAS®: Theory and Application

7.6. Ranked Data

The discrete-choice approach can be extended even further. Instead of asking people to choose one or more items within an option set, we can ask them to rank the items on any criterion we specify. Here’s an example. In Chapter 2, we studied 147 murder cases that went to a penalty jury to decide on a life-or-death sentence. In an effort to get an unbiased measure of the seriousness of the crimes, an auxiliary study was conducted in which 50 trial judges evaluated the murder cases. Each judge was asked to read documents describing 14 or 15 cases in some detail. Then they ranked the cases from 1 to 14 (or 15) in order of assessed culpability with 1 indicating the most serious. Each case was evaluated by four to six judges. There ...