Overhauling Online Ratings Systems Leads November Tas

摘要

The November issue of The American Statistician (TAS) leads with a paper in the Interdisciplinary section (subcategory: Information Sciences) about improving product rating systems used by Amazon, Epinions, and YouTube, among others. Using Bayesian hierarchical modeling, Daniel Ho and Kevin Quinn propose replacing the current "one to five stars" system with one that automatically incorporates sample sizes, thereby avoiding spuriously 'good' or spuriously 'bad' ratings. The following paper in the Interdisciplinary section (subcategory: Political Science), "Regression to the Mean, Murder Rates, and Shall-Issue Laws" by Patricia Grambsch, is bound to stir controversy, as strong feelings exist on both sides of the gun control issue.