THE BOOK--Playing The Percentages In Baseball

For those who don’t click through to the link, she is recommending that you watch tonight’s episode of Law and Order: Special Victims’ Unit.  ‘Personal Fouls’ is the name of the episode.

One in six? Why do I feel like most “statistics” like those are completely made up. If you added up everything that you had a 1 in XXXX chance of having, statistically, we should all be walking around with a dozen ailments and conditions.

Captcha: truth66

I knew I was right!

"If you added up everything that you had a 1 in XXXX chance of having, statistically, we should all be walking around with a dozen ailments and conditions.”

Funny, I’ve always thought the same thing!  Not that this particular number might not be somewhat correct.  At best, it is a rough estimate of course though.

What exactly constitutes sexual abuse these days? Does verbal harassment of a sexual nature count as abuse, or does there have to be actual physical, sexual contact? If the former, then IMHO, the definition has become way too broad. If the latter, there’s no way that I believe it. I think back to when I was growing up. I never knew any boys who claimed to be abused. Of course, it’s not the sort of thing that one talks about, but there were never any rumors or gossip, which is hard to believe if one in six were abused.

This must be a tough issue to measure accurately, given the highly personal nature of the issue and the consequent reluctance some might have to answer honestly.

One interesting technique I recall hearing about involves telling the subjects involved to flip a coin (privately) when they answer the potentially embarrassing/incriminating question - if it’s heads, give a certain answer no matter what, if it’s tails answer honestly.  For example, if you’re surveying teens and asking if they’ve ever used illegal drugs, tell them to flip the coin, then write down “yes” if it’s heads (whether or not they have used), and answer honestly if they flip tails.  That way if the results somehow get released (gee, does that ever happen?), a subject can say “sure I said yes, but I flipped heads”.  Once you have the results, just subtract 50% and there’s your answer.

I wonder if they came up with the 1 in 6 this way…

Greg: genius!

They should have done that for the initial testing back in 2003 or so.  (That was the one where they were trying to figure out the % of users, without attaching names.)

They should have automatically and randomly contaminated half of the tests, and treated them as positive.  Basically, add in a random PED drug to fail the test.

Where were you ten years ago?

The key, here, is “unwanted or abusive” - “unwanted” could constitute a game of doctor that you played when you were 7, when you took off your shirt even though you didn’t want to. They’re erasing the difference between something that simply merely made you uncomfortable and something that caused harm - since both are unwanted, they’re grouped together.

It makes a certain amount of sense, because otherwise you’re making wholly subjective and somewhat arbitrary judgements about which unwanted acts constitute “abuse” and which are just bothersome - and it *is* a very ambiguous line that separates some of those acts. On the other hand, of course, it risks sensationalizing something that’s trivial and delegitimizing the whole exercise.

ChuckO/4: The thing is, sexual abuse is probably not evenly distributed in the population. Sure, it definitely affects people of all income levels, races, geographical areas, etc., but being poor and non-white are probably significant risk factors. So even if you were right (and you probably are not) that you never knew anyone who was subjected to sexual abuse, extrapolating from your own experience would be unwise.

"1 in 6” is based on a very widely cited study by the CDC (linked off my name).  The data from that study are about 15 years old, but I help to do this type of research, and I can tell you that the numbers have sadly not changed. 

Men often refuse to talk about sexual abuse, particularly if it was perpetrated by a man, because it’s considered “gay.” In other words, a young boy is taken advantage of through no fault of his own and he’s made to feel that he did something bad by society.  That’s why there’s a lot of guys out there who suffer in silence.  That has to end.  Now.

If you’re the one in six, and it still bothers you, but you’ve never told anyone about it, treatment is out there, it works, and there is nothing shameful at all about seeking help.

Greg, that is fantastic!  I never heard of that either.  The benefit is not only if the results get released, but it also saves the person from embarrassment since the researcher does not know the real answers from the false ones.  Even when someone takes an anonymous test on paper, they will still lie for the same reasons (embarrassment, pride, shame, etc.) they would if it were given to them by another person. How many of you have felt embarrassed asking weird questions to Siri on your iPhone 4?

Anyway, the problem is not whether the 1 in 6 statistic is true or not.  The problem, as usual, is a semantic one. That kind of statement should never be used in a serious setting without defining the words. It DOES matter whether you are 5 and your 8 year old cousin “made” you play doctor or you were 5 and your 30 year old uncle raped you. Lumping them together is ludicrous. It’s like using batting average to describe a player’s value to his team. And, the things that are least damaging to a child are the ones that may be the most frequent.  So, 1 in 6 children may have experienced something relatively innocuous of a sexual nature, and 1 in 20 may have experienced something serious (I am making numbers up). If both are put in the same category of sexual abuse, guess what you get? 1 in 6 have been sexually abused! And what if they made the definition so broad as to include just about anything of a sexual nature?  That would probably lead to a result such as 1 in 2 children have been sexually abused! I am (obviously) not trying to downplay the serious nature of sexual abuse. Just expressing caution about people who communicate “numbers” without being careful about the words and definitions they use when describing or using those numbers…

Greg #5, I like the idea to preserve confidentiality, but you have to get your math right for correct analysis of the data.  Say you get 70 positives out of 100; you subtract 50 and have 20 left.  But that is out of 50, or 40%.  You have to double what is left after subtracting 50%.

The method Greg refers to is known as randomized response (http://en.wikipedia.org/wiki/Randomized_response), and I believe is a fairly common tactic in survey sampling. At least it is common in academic circles, I have no idea how often it is used in practice.

A slight alternative is the unmatched/item count technique, where you are asked several questions, and simply respond with the number of times you answered yes, but don’t specify which questions they are.
http://en.wikipedia.org/wiki/Unmatched_count

kds #11 - yeah, you’re right, I forgot about doubling the number.

Even with the randomized response technique, subjects may feel some desire to give the safe/clean/good response, just in case the results do somehow come out, thereby fouling up the survey.  You can isolate the subjects from identification completely by having them answer with some sort of token, instead of writing or speaking an answer.  Put two boxes inside a voting booth, marked “Answer” and “Discard”, give each participant a “yes” marble and a “no” marble, have them enter the voting booth and put the marbles in the appropriate bins.  That way the results literally cannot be linked to the participants (short of fingerprinting the marbles, which you could also preclude if you really felt you had to).