Aegisdottir et al. (2006) used 173 effect sizes from 69 studies published over the past 56 years, and concluded that there was a somewhat greater accuracy for statistical rather than clinical judgment methods. Similarly, Martin, Quinn, Ruger, and Kim (2004) found that statistical models could predict the outcomes of United States Supreme Court decisions more effectively than a set of independent predictions by 83 legal experts.
Seems like a discussion for.the beginnning of the book
