The dataset presented here illustrates to students the utility of logistic regression. Its analysis results in a fit that explains much of how senators vote on a particular bill, and allows for quantification of the effects of ideology and money on the vote. A number of interesting quantitative interpretations follow from a good fit. A successful analysis makes use of a number of ideas discussed in applied courses: descriptive statistics, inferential methods, transformation of variables, and the handling of outliers and special cases. All these issues arise in the context of data on variables that require of students no specialized knowledge. Students have strong qualitative preconceptions about the relationships among the variables. The final results quantify, and nicely confirm, many of those conceptions.
The CAUSE Research Group is supported in part by a member initiative grant from the American Statistical Association’s Section on Statistics and Data Science Education