Monday, September 28, 2015

Annals of Derp: Why Conservatives Can't Read

Basically, because they don't want to. You could get all stuffed up with information, and it might contradict your assumptions. Have some, if you must, but responsibly, and always leave a few paragraphs on your plate.

Photo by Mark Bonifacio/New York Daily News.

What's up in the world of universal pre-K, Callie Gable for National Review?
It helped get progressive stalwart Bill de Blasio elected mayor of New York, and it resurfaces again and again in speeches by both Bernie Sanders and Hillary Clinton, but the Left’s dream of publicly funded child care is suddenly getting pushback from an unlikely source: Jonathan Gruber, the liberal economist who helped design Obamacare.
Is that right?

Um, no, it's not. In the sense that Jon Gruber did not help design the Affordable Care Act (he did play a noteworthy role in the 2006 Massachusetts health care reform law, from which the ACA borrowed a number of important elements, but was only an outside consultant on limited aspects of the latter), and that the study we're talking about today, "Non-Cognitive Deficits and Young Adult Outcomes: The Long-Run Impacts of a Universal Child Care Program" (NBER Working Paper 21571, September 2015) had three authors, Michael Baker of U. Toronto, Jonathan Gruber of MIT, and Kevin Milligan of UBC, working for the National Bureau of Economic Research, and far from unlikely or sudden, it is a followup study to one the three published seven years ago, “Universal Child Care, Maternal Labor Supply, and Family Well-Being,” Journal of Political Economy 116 (4), 2008.
In a newly released study, Gruber examined the universal child-care program run by the Canadian province of Quebec. In 1997, Quebec began phasing in subsidized child-care, regardless of family income, for kids four-and-under.
Not quite. For one thing, they did not examine the program. In the 2008 paper Baker, Gruber and Milligan (or BGM) examined Québec children born in 1999, who would have been eligible for the program from ages zero to four, and those born between 1993 and 1998, who would have been eligible for the program over some of the time; this would include all the children who actually participated in the program from 1997 to 2003, and thus provide some evidence as to the program's effect, not on the individual kids, but the province as a whole; and compared them on various parameters to children of the same age cohorts in the rest of Canada, where subsidized preschool childcare was much less developed (reaching 20% of children by 2008, whereas it was reaching some 60% of Québec children at the same time).  In the new study, they look at that same age cohort through a somewhat different data set, and then examine their further development up through 2013, when they were aged 15 to 20.
It turns out that exposure to childcare in Quebec negatively impacts children in both the short and long terms. Compared with their peers in other provinces, young Québécois children had higher rates of anxiety, aggression, and hyperactivity. As teenagers, these children committed more crimes and were generally less satisfied with their lives. The program had no long-term positive effect on children’s standardized test scores or cognitive abilities, and may even have had a negative effect.
A bit of an inference too far. It turned out in the earlier study that Québec children born between 1993 and 1998 had higher rates of anxiety, aggression, and hyperactivity than other Canadian children, and this kind of "noncognitive deficit" was corroborated in a number of other studies; since far more of the Québec children were enrolled in cheap preschool than the other children were, it seemed likely that the cheap preschool had something to do with this, and there weren't any other big factors (such as child poverty, income inequality, or minority and immigrant populations, in all of which Québec is roughly comparable to Ontario and Manitoba and British Columbia, certainly not worse) to ascribe the difference to. I guess nobody thinks speaking French makes you anxious or hyperactive.

And then in the new study it was found that these same kids were less happy, healthy, and law-abiding—that the noncognitive deficits seemed to have persisted into adolescence and beginning adulthood. Not that I'm going to dispute the methodology or anything, or that it looks like anything other than bad news: it really seems as if "universal" early childhood education may not only have failed to do these kids any good, it may really have done them some harm.
Why is this program failing children in Quebec so miserably? Nobody knows for sure. There is no evidence that the program was underfunded or understaffed, and its negative effects persisted even after new regulations designed to improve the quality of care and increase wages. So it seems unlikely that substandard services are the culprit.
Whoosh! I didn't realize we were at the top of that roller coaster already!

There's a good deal of evidence, matter of fact, that the program was pretty poorly managed during the period under study, with many inadequately trained staffers. In 2001, 40% of the nonprofit Centres de la Petite Enfance did not meet even minimum standards, and only a tiny minority met the good to very good standard. And the for-profit garderies set up as a temporary expedient were worse; overall, 45% of family day care providers lacked any post–high school qualification, and 84% lacked early childhood education training. So there's that. The "new regulations" Gable read about in Baker, Gruber, and Milligan were those implemented in 1997-99 as the program was beginning, and they hadn't been properly implemented.

A paper by Michael J. Kottelenberg and Stephen F. Lehrer, "New Evidence on the Impacts of Access to and Attending Universal Child-Care in Canada", Canadian Public Policy 39/2 (2013), visible through JSTOR, brought the BGM analysis up to date for those born in 2003 and entering kindergarten in 2008, corroborating the earlier findings, but also they did look at the program, and the children in it, directly; and they found a couple of new things: (1) that the children with negative effects tended to be those who were sent into day care only because the government program existed; i.e., who had one parent, normally the mother, not originally in the labor force, though she might have taken a job when the childcare program became available; and (2) that the average child in the program showed positive, though insignificant, effects on both developmental and behavioral outcomes (as well as a significant lessening of the chances of depression for mothers who were working), suggesting to me at least that a more granular analysis would find small significant positive effects for some relatively larger group of children and large negative effects for a relatively small group; if you could figure out which was which, you could have an essentially universal program in which certain parents could be advised that their children might not benefit. (We might find an experimental test of this in the ongoing New York City case, where a large number of particularly well-off parents will not consider using the subsidized program but stick with the high-priced all-private options they have had for decades.)

Various efforts have been introduced (if you check out the French-language literature, missed by Kottelenberg and Lehrer because their study cuts off too early) since the period under study, following the introduction of a broad improvement plan (without additional money) in 2004, a new program Accueillir la Petite Enfance in 2007, and the provision of inspectors in 2010-11, but there doesn't seem to be any quantitative measurement out yet of their effects, except for one important experimental program, Cap Qualité, which showed promise and was starting to be implemented on a wider scale in 2014.

BGM do not suggest that there is anything in principle wrong with the idea of universal preschool care. It's a pretty weird thesis, when you think about it; what would it be about the universality of the program, as opposed to any other aspect, that would make some small but significant number of kids anxious as preschoolers and criminal as teenagers? I beg your pardon? But perhaps the research of Kottelenberg and Lehrer points to some alternative.

BGM do wonder whether the universality could be the problem, since that's the big difference between the Canadian case and the familiar very successful programs in the US. But the thing they emphasize has to do with well-known research in the US around evaluating popular programs like Head Start and the Perry Preschool program, which have suggested that the cognitive benefits (improved learning ability in school disciplines like language and math skills) are relatively short-term, tending to disappear over longer periods, while the non-cognitive social and economic and moral benefits are extremely durable.

The other thing has to do with the fact that these benefits in the North American research tend to be concentrated among disadvantaged people; those who are well off to start with will be equally successful in these domains whether the programs exist or not. In Europe, with the social commitment to egalitarianism, universal systems like the famous French école maternelle are the norm (and nobody questions whether they're good for the children, because it seems so obvious), but it's not clear that that works in the US or Canada:
the external validity of the European evidence to other jurisdictions is not clear. European programs are run under different funding levels, which reflect the public’s greater acceptance of an active state and government’s assumption of a larger proportion of economic activity.
They weren't able to draw a conclusion as to whether early childhood education should be provided universally or not, but they were able say say confidently that the noncognitive component was the key, and whatever is wrong with the Québec program, in constrast to Perry Preschool, must have to do with its failure in noncognitive development:
The implications of these findings for early child care policy are profound. They provide strong support for the argument that non-cognitive development is a crucial determinant of the long-term success of child care programs. This suggests that measuring the impact of child care programs on the non-cognitive development is important. When a child care program fails to improve non-cognitive development, it may have no long-standing positive effects on children.... As non-cognitive skills can be evaluated relatively easily in the short-run, this suggests noncognitive skill development can act as an important guide for policy in this area.
To which I'd add that the quality of the program, as with Head Start and the Perry project, is the key; it has to aim at noncognitive development and it has to be done well, which probably means not cheaply, and with less private initiative and more government than our overlords would prefer. The New York approach will be offering some great tests of that.

As to the National Review, to the conservative mindset every question is a question about conservatism, even when you're trying to talk sports, or movies. They start with opposition to universal pre-K as a default position because it costs tax money and pays off more for the disadavantaged classes than for the country club, and because they're always looking for ways to punish working women for trying to juggle motherhood and employment; and they read BGM's paper (with a little extra titillation over the presence of Jonathan Gruber's name) looking for ammunition to shoot the idea—"the Left's dream"—down. That's not how you do social science that works.

No comments:

Post a Comment