Improving Judicial Ideal Point Estimates with a More Realistic Model of Opinion Content∗

We extend the dynamic item response theory model of Martin and Quinn (2002) by including information about the court of origin and the opinion writer using a dynamic item response theory (IRT) model with a hierarchical prior. The model is estimated using Markov chain Monte Carlo methods. This approach not only provides better estimates of quantities of interest, such as judicial ideal points, it also allows us to estimate both the status quo and policy alternative locations for each case, as well as the location of each circuit court. This application shows the value of incorporating institutional detail and other available information into statistical measurement models. ∗The authors are listed in reverse alphabetical order as determined by lot. We gratefully acknowledge financial support from the National Science Foundation (Grants SES 03-50646, SES 03-50613, SES 01-35855, and SES 01-36679). We also thank Steve Van Winkle for help forming reasonable circuit court priors. Data for replication, posterior density samples, R code to replicate the figures, and documented C++ code to fit the model will be made available at the authors’ websites. 1