September 28, 2013

Has grade inflation struck IQ tests, too? A look at kids of New York elite parents

Via Steve Sailer, here is an NYT article on the increasingly high-stakes contest among New York parents to get their kids into elite kindergartens. The focus is on the IQ test used to sort out the top 1% from the rest. Several sources express concern that something funny is going on because the fraction of all New York kids clearing the national top 1% mark has shot up in recent years. Just 6 years ago, only 18% cleared that mark, while last year 29% did.

The IQ / HBD crowd should show some more interest in corruption, test prep, and other forms of grade inflation among the elite -- not just in the obvious places like the Atlanta public school system (wink wink).

Intra-elite competition has been ramping up for decades now, so we can't assume that the most powerful and wealthy group of parents in the country aren't making good use of that power and wealth to get artificially better outcomes for their kids, and hence lower outcomes for everyone else, where those resources are limited.

Looking through the ERB report on changes over time, it's clear that something funny is going on. There's an increase in the mean combined score of 0.375 Standard Deviations -- in just 5 years (fall '07 to fall '12). If we phrased it in terms of height, it's as though these kids had gotten 1.1 inches taller on average in just 5 years. Little of that is due to changes in verbal scores, which are pretty flat. The performance scores, however, have increased by 0.6 SD, almost linearly over time. In height terms, that's 1.8 inches taller on average than 5 years ago.

Moreover, there are no changes in the national sample of children, whether for the verbal, performance, or combined scores, and whether you look at mean scores or 90th or 98th percentiles. Something is going on specifically in the New York bunch of test-takers.

Now, what would champions of the "cognitive elite" say about a group of Atlanta public school kids whose mean IQ scores had apparently risen in 5 short years by nearly 0.4 SD, and whose performance scores had risen by 0.6?

Demographics? Nope -- the ERB looks at age and sex, and there's no change. They don't mention race, but by fall 2007, New York super-parents were already fully white / Jewish / NE Asian. Environmental improvement? Nope -- those kids of Manhattan super-parents weren't starving, disease-stricken, etc., "back" in 2007, only to have recovered by now.

That leaves artificial causes. Since the verbal scores are flat, you might think that it's mostly test prep -- little human beings are already designed for verbal communication and reasoning, but not for weird new things like "what picture comes next in this sequence?" or "3 of the cells of this matrix are filled in. What goes in the missing cell?" There are larger marginal returns if you prep for the performance sub-test, and don't waste too much time on the verbal one.

That's some pretty damn good test prep, if that's all there is to the rise in scores. Remember -- 0.6 SD in just 5 years, on an IQ performance test, not some quiz of factoids. I don't doubt test prep is the new normal for Manhattanite children, and that smarter kids will get more out of it than duller ones. But, not enough to produce such a huge fast increase.

Rather, the likely causes are some mix of tacit grade inflation and outright corruption. IQ-focused nerds can join the 21st century and recognize how rampant such practices are -- including among the very top of the elite, like grade inflation and "No Child Left Without Latin Honors" at Harvard.

I have no experience with Manhattan super-parents or the local test-production and test-prep industries, so I have no intuition about what mix is the most likely. Grade inflation sounds more likely -- understood as necessary by the test-makers, or the super-parents will take their business elsewhere. That will hit the company financially and reputationally.

It's like during the last economic bubble, the ratings agencies inflated the worthiness of various financial packages because if one of them were more honest, the customer would shop it around to another ratings agency that would give it higher marks. The customer holding this batch of mortgages (or whatever) was not interested in an honest appraisal -- they wanted to pass it on to some sucker for as much money as possible, and a higher rating means more money.

Similarly, parents don't want an honest score for their kids -- they want the highest score possible, in order to pass them along to the school that will give them the greatest resource-earning potential into the future. They can't all get the maximum score, or the jig will be up. But that still leaves plenty of room for gradual, subtle rating inflation. The cumulative effect is more pronounced (like the '07 to '12 comparison), but by then your kid has already taken the test and gotten his inflated score, so what do you care if the bubble bursts in the future and wrecks the community? Gotta look out for Number One (and mini-One).

Corruption is not out of the question either, but probably less of a factor. You see that more when the test-takers are below average, and the proctors will just erase the wrong answers the students gave and write in the correct ones. Maybe with orders from above, lest the test make the public school system look bad. But not so much at the elite level -- parents in general aren't greasing the administrators' palms to get their kid into Harvard. Same with the housing bubble -- holders of the mortgages didn't plunk down however-many dollars on the conference table. They used the less detectable tactic of shopping their product around to find the highest rater, with competitive ratings agencies only too eager to accommodate the demand for inflated ratings.

I'm not sure how much revenue the ERB brings in from New York parents, in order for their shopping-around to drive possible decisions to inflate scores. I'm guessing it's more of a reputational concern -- if New York elite parents drop your test like a hot potato, you look like losers. That would hold even for an established IQ test like the SAT -- but would really slam a more unknown test that's just starting out (only existing since the mid-2000s), trying to distinguish itself from all the other tests that elite parents could choose from. Theirs is an endorsement that money can't buy. You have to make it worth their while in other ways.

Both of these causes still don't explain why the rise in scores is most dramatic for the performance and not the verbal scores. Perhaps the grade-inflaters know how to make it seem less noticeable. It is what they do for a living, after all. Like, if the goal is to inflate overall GPA, then (at least at first) inflate the grades for music class but not for math class, which would raise more red flags. And perhaps part of what those $200-per-hour tutoring sessions get you is some first-hand knowledge of the questions through someone who's socially connected and hard-up for money in uber-expensive New York City, though restricted to just the performance sub-test. Again, don't want to give away too much and have it look obvious. And to maintain a decent bargaining position, you want to still be holding onto something they want, not just give it all up on the first date.

All of these hypotheses and their predictions should be followed up on, to see where the culprit truly lies. Though good luck figuring it out. It's not simply a case of "getting your hands on the data" -- that's just test score data. The "data" that you're really after is the behavior and relationships among the various players -- and they're not likely to reveal that much to outsiders.

There's also likely an understanding that the test is designed to get the Children Who Matter into the Schools That Matter. If they don't, why, the global economy could unravel or blow up in 20-30 years. How could America run itself without the right children getting into the right schools? So, you cut them some slack in advance appreciation for the all the society-holding-together work their children will be contributing throughout their lives. Like driving home-ownership rates up to 100%, handing out mortgages to illiterate Mexican strawberry-pickers, opening the floodgates of immigration, and so on.

In summary -- time for the IQ crowd to stop pretending we have a meritocracy these days, to stop being such naive pawns in the elite's game of self-advancement. Intra-elite competition couldn't get any fiercer (although stay tuned for some scenes from next week's episode), and evidence of "rating inflation" of one kind or another has been obvious for decades now. As elite households fight more ruthlessly against one another in naked self-promotion, they'll result to whatever it takes. We don't live in a restrained, sportsmanslike meritocracy but in a no-holds-barred war of all against all.

4 comments:

  1. The customers of ratings agencies weren't buying inflated AAAs because they intended to dump them on a greater fool. It was regulatory arbitrage. Customers wanted highly rated securities even if they weren't that safe, because those met capital requirements.

    ReplyDelete
  2. The "test makers" are not to blame here. The test in question is a sub-set of the WPPSI (Wechsler) IQ test, which is meant to be a diagnostic test, not a school placement exam. Real placement exams have safeguards against cheating in place - multiple forms that change very year, etc. The WPPSI has none of that. It's the schools fault (now about to be corrected) for using the WPSSI for something it was not designed to do and as the stakes became higher it was inevitable that it's no existent safeguards would be breached.

    The other party at fault here are psychologists who administer the test. 4 year olds can't take a paper and pencil test. A human has to administer it. As with the rating agencies, someone who gives low scores loses business. Now the rich, white, high class WASP and Jewish toddlers who take this test are a smart bunch to begin with, as befits people whose parents are in the 1% of wealth (which is highly if not totally correlated with intelligence). But 4 year olds can be stubborn. Test administrator - "What do you call the animal in this picture" 4YO - I dunno. "Bad TA" - marks down 0 points, goes on to next question. "Good TA"- "I forgot the name of this animal. Can you please help me?" 4YO - "A duck". Good TA - 1 point.

    But the test makers themselves are faultless. As you say, the national norms have not moved, only these 3,000 kids in Manhattan have suddenly gotten smarter.

    ReplyDelete
  3. "Customers wanted highly rated securities even if they weren't that safe, because those met capital requirements."

    Same thing. It's a cover-your-ass policy. Not like they were just going to sit on those things. It made the owners look like big shots -- "I'm managing $X billion right now." Just like, "Our college houses a large student body with a 3.5 GPA."

    Eventually somebody is going to buy that product, though.

    ReplyDelete
  4. Is this not more reflective of competition among the lower levels of the elite than the whole of the elite? The Rockefeller heirs and Georgina Bloomberg are well-fixed no matter what. But the 4 year-old son of a run-of-the-mill investment banker might won't be if he doesn't end up at HYPS.

    ReplyDelete

You MUST enter a nickname with the "Name/URL" option if you're not signed in. We can't follow who is saying what if everyone is "Anonymous."