THE BOOK--Playing The Percentages In Baseball

The first commenter on Tyler’s site actually makes the best point of all. That is, most incidents of domestic abuse and sexual abuse aren’t reported at all, so trying to reach conclusions based on such small numbers and over such a short period of time is pretty much impossible.

The non-reporting is not an issue, unless it’s biased.

For example, suppose that women report disproportionately more often following a sporting event.  Then, we see an uptick in reports after sporting events, and conclude that sporting events are somehow a causal agent to the rape.

In fact, the sporting event can be the causal agent to the reporting of the abuse.

If “bigger” events, like the Stanley Cup cause disproportionately more reporting of abuse, then the numbers may be showing that, rather than an increase in actual abuse.

It’s all a question of bias when it comes to sampling data.

Sure, that makes perfect sense. When she talks about domestic violence, for example, that seems to limit the chance for bias. (But we still need to admit that the report rates for domestic violence are so low that even a the appearance of a huge drop or rise in the rate could just be noise.) But the meaningfulness of the report-rate for sexual assault really does depend on unpacking the pre-existing relationship between the assailant and the victim. (That is, we would expect the rate when the assailant is a stranger or anonymous, or the assault is public or particularly violent, to be pretty consistent because the report rate for those assaults is much higher.)

My point, though, was simply that if the majority of sexual assaults are unreported, then it’s difficult to draw conclusions of any kind from the data. How, for instance, do you conclude that an increase from, say, 3 reports to 10 in a given timeframe is indicative of a variance in the actual number of occurrences or that it just fits the expected variance in the reporting rate? So even Tyler’s refutation seems a bit suspect, to me - I think they’re both claiming that the numbers say something that they can’t say. It feels, to me, like they’re making a claim about a player’s change in talent level based on their performance on Saturdays, alone.