What's up with that?

Last night I played tennis with a friend. It was after dark but the courts were well lit. Adjacent to the tennis courts was a basketball court. No lights. This isn't the first time I've noticed such a design. Why not light those as well?

Is the bouncing of a basketball so loud that it's a nuisance? I'm inclined to think this isn't the answer since the park was a good distance from any homes. Is there greater demand for tennis at night than basketball? Are they trying to keep away the types of people who would play basketball after dark? Is there something else I'm missing?

Another puzzle?

There are far more children born every year with fetal alcohol syndrome than are born with autism or Down syndrome. Fetal alcohol damage is the leading cause of mental retardation, with an estimated forty thousand babies born with the syndrome annually. And yet, the condition receives far less attention and support than other disabilities.

I wonder why. (Kidding. I'm convinced that it's because it's perceived as a poor folks problem.)

Once again, the quote is from the excellent All God's Children by Rene Denfeld. I'm looking forward to reading the second half on my trip to Chicago today but I think it's safe to say that it's a great book. I admire and appreciate the author's ability to remain purely descriptive. The book is written as a series of interconnected vignettes--each street youth's story is well researched and fascinating. (If you were wondering about the context of the quote, one of the girls had FAS.)

In which competition fails to lower prices?

"In Portland, the first major street family came into being in 1998 with the Nihilistic Gutter Punks....The NGP family soon claimed between seventy and a hundred members. The family grew too large to squat together, and smaller families began to splinter off the main group....As the youths split into smaller factions, the families developed increasingly severe initiation rites. New members of the Sick Boys were beaten into the family and, at times, were told to inject eighty units (a large syringe) of methamphetamine into the neck, a shot that is now known as a Sick Boy. The female members of the Sick Boys were called Sick Bitches, and they had to commit assaults and robberies in order to earn the right to wear "Bitch Bangs," or the short fringe of hair that female street youths sometimes keep when they shave the rest of their hair off. If they wanted to tie objects in their Bitch Bangs, the Sick Bitches had to commit more assaults. The harsher the initiations, the more the street kids clamored to join...."

This excerpt (with emphasis added) is from the very interesting and well-written All God's Children (which is currently 80% off at amazon.)

The first bolded sentence is interesting. Thinking about the street families as suppliers in a service industry trying to attract customers (kids) we would expect increased competition to make prices (initiation) fall. But the opposite happened. Now, it's obvious that the suppliers would want to increase their price but it's harder to explain how they could get away with it. Why wouldn't a no cost family spring up and attract all the kids who have yet to join a family?

I can think of two explanations that I think are plausible. One, street families are an example of a mythical (?) giffen good in which demand increases in price. This is the position the author takes in the second bolded sentence. I don't really buy it. I think correlation is being mistaken for causation. Demand for street families was already booming. That kids continued to join after the initiations became more severe does not mean that the harsher initiations caused demand to increase. That said, it's plausible since street families are clearly about status. Nevertheless, I think another explanation is more likely. That is, despite the increase in the number of firms (the splintering of the street families), market power of the suppliers was actually increasing due to increasing demand.

Both of these hypothesis seem testable with the right data. Perhaps the world needs an economics of street families to accompany the economics of drug-running gangs? Interestingly enough, one of the experts on street families, Bill McCarthy, is here at Davis in the sociology department. Perhaps I'll have an opportunity to ask his opinion at some point.

This is not the first time this month I've thought sociologists may hold the answers.

EDIT: A Rambling Grad Student has thoughts on this puzzle.

The determinants of fanatical product support

A rambling grad student offers a few reasons why such things as trucks, operating systems, and video game systems develop fanatical product support while other things such as couches do not. He lists 5 candidate explanations:

1. Price: The big ticket items are more likely to develop fanaticism - works for trucks and electronics, but furniture costs a lot, too . . . why does no one even claim LaZBoy roks and Pottery Barn sux?
2. Social visibility: Maybe people are more likely to be fanatical with high profile visible goods - but that kind of fails when it comes to computers and video game systems.
3. The people that buy them: Maybe people that consume these particular goods are just the kind of folk that feel the need to identify themselves with one side or the other (or at least the type that are willing to broadcast their product choice and publicly berate all other choices)
4. Longevity: Maybe the longer timeframe investments warrant more fanaticism to justify purchase . . . but the cycle time on electronics is far too short for that rational.
5. Amount of user interaction: Electronics and trucks both have high user interaction time. Maybe items with a lot of product/user interaction require greater mental justification, and thus support higher amounts of fanaticism.

#5 is by far the most interesting and clever hypothesis.

Nevertheless, I think the answer is a twist on #1. That is, fanatical product support is driven by price relative to income. I imagine that those who are fanatical about trucks first purchased their truck when its cost was a large share of their income. More generally, it's probably stronger for people with low income. The same for videogame systems and computers--it develops for young folks with little disposable income. This also explains why no one cares about couch brand names. When it comes time to buy a couch, it tends to be a drop in the bucket.

I'm sure you have many personal counterexamples to this explanation. I also know plenty of folks who are well-off yet fanatical product supporters. This is easily explained by a gravity-type model. Once a bunch of people have sorted into their respective camps, there is pressure for others to join rather than to remain agnostic.

Should schools teach kids how to talk?

Lately I've been reading selected articles from The New Education Reader: Leading Educators Speak Out. One of the articles pointed something out that is completely obvious that I have never thought of. Phonics is less likely to be effective for children who don't speak "properly" since they may not associate the "right" sound with letters.

This may explain why some kids have trouble learning to read. Especially kids from disadvantaged backgrounds. Also, this suggests that kids must be taught how to speak "properly" to teach phonics. I wonder if people are comfortable with this.

(You might argue that, more generally, teaching reading necessitates teaching how to speak properly but I don't think this is the case. Not in the sense that reading is word recognition anyways.)

Why aren't prostitutes paid like stadium vendors?

Tonight I watched a documentary called American Pimp. It was fairly interesting although it was difficult to watch the pimps talk so easily about manipulating women. I think I was made most uncomfortable by the fact that I sometimes felt that they were kind of charming. (That was an awkward sentence. Unintentional symbolism? Yup, that's it.)

Anyways, the most interesting thing I learned was the percentage that prostitutes get of their revenue. Zilch. (I'm not sure how exactly they're paid but that much was clear.) This seems like a bad business plan since it gives little incentive to the women to generate additional revenue. That is, in theory the pimps could make more money by giving the prostitutes a cut so that they work harder. Stadium vendors, for example, get paid based on how much they sell. What is different about prostitution so that this business model doesn't apply?

Perhaps the answer is that it's not different and the industry is due for a shake-up from an economics savvy pimp. But street prostitution has been around so long that I doubt a flawed strategy has persisted. Especially when the alternative is so obvious. Thus, I do think it's correct to ask why the commision-based model doesn't apply.

I'm tempted to relate the situation to the one described by Levitt and Venkatesh of crack dealers working for peanuts in hopes that they can work their way up the organization. But there is no up for these women. Or perhaps "up" is an occasional night out with their pimp. (Okay, I'm getting more depressed now.)

Alternatively, pimps may not need to offer a carrot (percentage of revenue) if their sticks (psychological and physical punishment) are effective enough to get their prostitutes to work as hard as they can. I think this is more likely the case. The pimps can easily monitor their prostitutes effort and punish them if they don't seem to be working hard. Both the cost of monitoring and the cost of punishing are probably low. Monitoring is cheap since the pimps don't have much else to do and the cost of punishing is cheap since the women lack other options. (Well, they could go to another pimp but I wouldn't rule out oligopoly behavior amongst the pimps as far as this is concerned.)

Turning the question around, why aren't stadium vendors paid like prostitutes? I think it's likely that monitoring and punishment are relatively expensive. (And, obviously, punishment is limited.) So, anyone have the data to test this hypothesis? (Perhaps a better question is: what data would you need to test this hypothesis?)

Bad Bias: Handwriting and Student Evaluation

While I'm not as huge a propononent of overcoming bias as these folks, I think this case deserves wider recognition and some action....

There have been a ton of studies going way back (to at least 1927) that show student handwriting impacts teacher's evaluation of their work. To me, the most obvious explanation for this phenomenon is that good handwriting is a signal of intelligence (or high achievement) which teachers use as a shortcut to evaluate the work.

It is potentially very easy to rule out this explanation. If handwriting is not related to intelligence or school performance as measured by a computerized test, then we can rule out the above explanation. Of course, if it is related to these measures, it doesn't mean we can rule out competing hypotheses that may play a role in the story.

As far as competing hypotheses are concerned, it's possible that good handwriting signals something else (or something more) than intelligence or high achievement. Based on my experience grading plenty of university-level homework and exams, the most likely candidate is gender and, to a lesser degree, race. That is, female and Asian students tend to have the nicest handwriting. So one possible explanation is that teacher's behavior is biased by gender and/or racial attitudes.

Regardless of what exactly is driving the bias, the costs are potentially severe. In addition to the possibility that this bias can affect a child every time he/she turns in a piece of handwritten work is the fact that the single most important test that most students ever take is also subject to the bias. The SAT is supposed to be a great measure of ability because it's standardized and because the scores are objective. But how can these tests (in which cursive essays score higher than printed essays) be truly objective while maintaining a handwritten writing section?

What does all this mean for you? You should put effort into improving your child's handwriting. This is unfortunate because it really seems like a waste of time. (That is, it's unlikely to improve their real productivity as adults.)

What does all this mean for us? Given that this problem has been well known for so long, it's a wonder that no direct efforts have been made to deal with it (that I know of.) Perhaps remedies weren't so easy when a lot of this research was done in the 70s and 80s. In any case, today there are a number of low cost measures that can be taken immediately to mitigate the problem. First, we can spread the word about the bias and commit to not succumbing to it ourselves. Second, student evaluation can be more focused on multiple choice type questions. (Cheer on fellow TAs!) Of course, the extent to which multiple choice questions can test understanding is limited. For this reason, term papers (in which the identity of the students are masked prior to evaluation) are a good option. Fully computerized work by students seems to be the ultimate solution--perhaps this is coming anyways.

As far as research goes, as I said before, much work has been done on this topic. At the same time, based on the limited amount of reading I've done, the quality leaves much to be desired. Here's the experiment I'd run:

1. Have a collection of students write a page of random words.
2. Give students an answer to a question and have them write it in their own handwriting.
3. Randomly assign samples from #2 to teachers to evaluate.
4. Randomly assign samples from #1 to teachers. Have them evaluate the quality of handwriting on a 1-10 scale. Also have them evaluate how confident they are that the sample is from a boy/girl on a 1-10 scale.
5. Make sure to collect demographic data on students and teachers.

There are two main innovations of this experimental design. First, the evaluation of the handwriting quality is separated from the evaluation of the "work." Second, and more importantly, is obtaining data on the teachers impression of the child's sex. With this data we could to separate the sex signal from the handwriting signal. And we could also test the initial hypothesis I posed above (by collecting data on the kids prior academic achievement.)

Too bad this would probably be a lot of work. Anyone want want to do the heavy lifting?

Grandma's generosity is a result of evolution?

I recently learned of "The Grandmother Hypothesis" (thanks Will). Basically, the hypothesis is that women evolved menopause because, beyond a certain age, it's more effective to invest in their existing children (and grandchildren) than it is for them to continue having children to ensure that their genes persist into future generations. The driving force is the fact that childbirth becomes more dangerous as women age. Neat story.

I believe economist Esther Duflo found behavioral evidence for this theory when she studied the South African Old Age Pension Program although I don't think she realized it (or perhaps she thought it not worth mentioning). She finds that payments to grandmothers improve their granddaughter's health (in terms of weight and height) but payments to grandfathers had no effect. (Interestingly enough, neither has an effect on grandson's health.) Duflo argues "grandmothers are likely to have a stronger incentive than grandfathers to invest in children because they will benefit from them for a longer time" since they are expected to live longer. This is a reasonable explanation but the grandmother hypothesis seems a nice alternative.

War and Child Maltreatment

The stress of having an Army spouse in a combat zone leads to a 60% increase in the rate of moderate to severe maltreatment of children by the spouse left behind, researchers said Tuesday.

This is from an LA Times article describing a recent study in the Journal of the American Medical Association. I think too much attention is being placed on the spouses. The stress of having a parent in a combat zone is surely incredible as well. The stress the kids are under may affect their behavior in ways that make them more likely to be maltreated by their remaining parent. The original study seems to miss this point as well.

To be clear, I'm not "blaming the victim" but determining good policies to improve these sorts of things relies on understanding what leads to what.