Reading a novel regarding a legal drama, the author used some statistics...but not blatantly. One character was surprised when three defendants in a row had reneged on a plea bargain, demanding a trial. In the US, it's true that nearly all cases are resolved through plea bargains. Very few go to trial. 97% of federal cases are bargained while 94% of state cases end similarly.
The chances of one person going to trial in a state's courtroom is 6%. The chances of two consecutive people is 0.4%, already approaching a level exceeding any process control limit. This was the first hint that trouble was brewing for the prosecutor. The chances of three defendants is 0.02%. Approaching the highly improbable, certainly.
Later on the prosecutor learns that a particular judge enjoys a 90% conviction rate. However, in her courtroom, one defense lawyer wins acquittal for 70% of his cases; in other courtrooms, he manages to win 30%--still above the average--but not close to his win record in this particular judge's presence. Hmm, hints of suspicious activity.
Let's assume there are 8 judges in this court house. With a win rate of 30%, we would expect a range of 0% to 78% with any one particular judge. Having a win rate of 70% with one judge would not be statistically significant. However, if we didn't know this defense lawyer's overall record, and we compared it to all others, then we would expect a win range of 0% to 42% with any particular judge. With this comparison, the 70% win rate is significant and startling.
No comments:
Post a Comment