New results on preschool from a Tennessee RCT

This is the most extensive and careful study of preschool (pdf) I have seen to date, conducted by Lipsey, Farran, and Hofer of Vanderbilt.  The core result is this:

The third question we addressed involved the sustainability of effects on achievement and behavior beyond kindergarten entry. Children in both groups were followed and reassessed in the spring every year with over 90% of the initial sample located tested on each wave. By the end of kindergarten, the control children had caught up to the TN‐VPK [preschool] children and there were no longer significant differences between them on any achievement measures. The same result was obtained at the end of first grade using both composite achievement measures.

In second grade, however, the groups began to diverge with the TN‐VPK children scoring lower than the control children on most of the measures. The differences were significant on both achievement composite measures and on the math subtests.

In other words, after some period of time the children who had preschool actually did worse.  I found this interesting too:

First grade teachers rated the TN‐VPK children as less well prepared for school, having poorer work skills in the classrooms, and feeling more negative about school. It is notable that these ratings preceded the downward achievement trend we found for VPK children in second and third grade.

So does preschool make kids more grumpy?  Immigrant children by the way did well:

…whether or not ESL children experienced TN‐VPK, by the end of third grade, their achievement was greater than either of the native English speaking groups of children.

Arnold Kling offers comment, and for the pointer I thank Peter Metrinko.


Reality still has a liberal bias right?

Well, some sort of bias can be found pretty easily, in the section that Prof. Cowen cites as interesting - 'It is notable that these ratings preceded the downward achievement trend we found for VPK children in second and third grade.'

Seems like a bit of a causal explanation, actually, since according to the study, the two groups were comparable by measures not involving a teacher's opinion.

Since preschool actually costs money, a result that the two groups were comparable is a very strong result against preschool. Preschool, or at least spending public funds on preschool, is really only justified if one can demonstrate unambiguously strong evidence in favor of it. If a profit-seeking Big Pharma company were introducing a new drug for preschool-age children that purported to make them smarter and/or better behaved, would the FDA approve this drug given these RCT results? Why should the standard of efficacy be lower for a "treatment" of putting kids in a room all day with a teacher than for giving them a miracle drug? Why would the standard be lower rather than higher for spending public funds on something compared to simply approving something that people could choose to spend or not spend their own money on?

The preschool situation is more like the case where a specific drug is found ineffective for a condition, and some argue that we should stop trying to find drugs that work and ignore drugs that already appear to work.

I can't tell whether or not you agree that preschool programs should undergo RCT testing. When Big Pharma makes drugs for the Koch brothers and others wealthy enough to afford them, taxpayers pay for the FDA to ensure that those drugs are effective. Yet, some people want to force the poor into preschool programs --- and they are forced in the sense that they can't afford private preschool --- without providing access to the same types of effectiveness guarantees. The same applies to another RCT test that showed that the poor in Oregon were being forced into a Medicaid program that didn't produce better health outcomes. We should be providing free RCT testing to all of the programs that the poor are forced into so that the poor have equal access to the effectiveness guarantees that the rich already get from the FDA.

In fairness, it doesn't seem like a comprehensive financial evaluation.

For example, if preschool provides no long-term beneficial education effect for the child, it nonetheless still provides babysitting for the parent.

Though around here, "preschool" is where SAHMs send their kids a couple of mornings a week for some sort of enrichment experience, while "daycare" is where working moms park their kids between 8:30 and 5:30. I haven't read the study, so I'm not sure what they're referring to.

We don't worry about the cost. Some businessman will be taxed more to pay for this. So who cares?

There's a big push in CA for "Universal Pre-K" and they now want another "Millionaires Tax" to pay for it. (Right now there are two extra 1% tax brackets for people making more than $500,000/year, for a max rate of 13.9%)

From talking to the do-gooders at my left-leaning Synagogue who have a committee to advocate for Universal Pre-K, it's not that they care about universal pre-K. They just want to take more money away from people who "make too much." If there were an initiative on the ballot to confiscate another few % of income from people making more money than they do, with the proceeds to be burned or shredded, these people would support it.

Well, at least they're getting child care. Presumably parents don't put kids in pre-kindergarden unless they're working.

Or napping to get ready to go out clubbing tonight to make more children to put in taxpayer-supported preschool.

That's only certain parents, you know what I mean? /wink

Or studying for the bar exam. See? I can make up bullshit too.

That's probably worse from a social benefit standpoint than the clubbing parents, tbh.

Pre-k is pretty common among families with SAHMs.

Quite frowned upon if you do not enroll your child regardless of work status.

Yes, pediatricians recommend some sort of preschool beginning at age 3... structured play and social interaction. These are the kids that often just do half-days.

Still, based on this research, they should probably drop that recommendation.

I mean, it doesn't show that big a negative effect, but it sure doesn't show any positive effect.

So what? The kids _aren't_ _doing_ _better_. The "well at least they're getting child care" argument has little merit.

Agreed, it'd be like dropping them off at the hospital during the day and when they're even less healthy coming out of it saying "well at least they had daycare". Pre-k is a massively inefficient way to provide daycare and as a result of this study, should probably be skeptical of those wanted taxpayer funding of the universal kind.

At least he said "presumably."

The implication that pre-school is somehow actually how harmful for these kids tells me one of two things must be going on: the study randomization was flawed (as Heckman pointed out) or this particular program is really crappy. I've never seen any evidence for a mechanism by which decent pre-schools actually make kids less prepared for school than if they were with a babysitter.

People on this board, ask yourselves, are you going to not send your kid to pre-school based on this new information? My guess is that this has changed nobody's mind.

"the study randomization was flawed": that seems pretty likely to me. And necessarily it's not double blind.

If the study had been analyzed at kindergarten entry, as is the case with some other preschool studies, Heckman and others would all be bragging about the "positive" results seen at that time.

If the randomization was flawed, how does it explain the fact that the treatment group pulled ahead for a while, but then fell behind by 3rd grade? I can understand biased allocation as creating a bias against preschool, obviously, but the bias should have been there throughout the study, not just starting in 2nd grade.

The mechanism suggested would be that the group selected for the pre-K program was overall a poorer performing group than the control. Then the pre-K program gave them a leg up so they were doing well when they started kindergarten, only to have the effects of the pre-K program dissipate over time with everyone regressing towards their "default" (no intervention) conditions, wherein the control group was already better off.

Now, that's a long way to go to say that this study is invalid. That said, the only certainty I take away from this is that I'd rather not send any kids I know to TN‐VPK.

As a well-adjusted preschool graduate myself, I must say that I found my program to be stimulating and challenging with ample nap time included.

Okay, reading the results reported in the study rather than the summaries, it appears much simpler:

TN‐VPK makes kids a little better off (statistically significant but practically irrelevant) when they enter kindergarten.

By 3rd grade, everyone's in the same spot. The non-TN‐VPK kids have a slightly lower point estimate (effect size roughly 1/3 of the effect observed entering pre-K but in the opposite direction), and it's marginally significant (0.05 < p-value < 0.1). That's a big nothingburger.

The headline is that Tennessee has an ineffective pre-K program, and that's about it.

But don't worry, when we roll this out nationally, the top people in DC will make sure its "effective."

... maybe the post pre-school education was ineffective - or even destructive? Any assessment of the quality of K-3?

Pure speculation follows:

Maybe it's germs. Preschoolers get sick a lot more often than kids who just stay home. Daycare, meaning full-time care for working parents, is much worse as there are strong incentives to send kids even when they're really sick because mom can't easily skip the big meeting today.

Where does Heckman point this out?

The randomization here was actually quite good. You have to be free or reduced-price lunch to get into Tennessee Voluntary Prekindergarten and a parent or guardian has to apply. There were more applicants than spaces, which were given out by a lottery. The control group was the kids who applied and lost the lottery. The researchers compared the two groups and found them very similar.

The study does not deny that the pre-school kids were better prepared for school. It even says they do better for K and grade 1. However, it also says that in grades 2 and 3, they do worse. For many people (certainly not readers here but maybe low-income 6-year-olds) school is not a pleasant experience: sitting, following instructions, being expected to learn all sorts of things which you may not be interested in. I wondered, at first as a joke but later seriously, whether this was "burnout." Turns out the researchers actually use the term, though they diplomatically focus on the quality of schooling, rather than school in general:

"State programs that are not careful to protect the instructional environment for 4-year-olds may find the children burning out in the early grades from too much repetition of the same content and instructional format. Rather than building enthusiasm for learning, confidence in their abilities and a foundational understanding of literacy and math, the programs may only be teaching children how to behave in school, an enthusiasm that fades with repeated exposure.”

But some kids who were randomized to the pre-school group didn't end up going to the pre-school progam.

It confirms what a waste of money preschool is. It's merely a status good.

Yes, in other words, your theory of the world is unfalsifiable. All evidence against your theory is flawed/crappy. All evidence for your theory is wonderful and methodological sound.

So how does your ideology differ from religion again?

If you haven't been paying attention for all these years, the comments to this post demonstrate that liberalism (or progressivism or whatever) is faith-based and totally beyond the reach of reason. There is no study that can be done that will budge their unshakeable belief in the power of their good intentions, fueled by other peoples' money. They cling to their Heckman (the science has been settled for years!) and breezily dismiss any result not in accordance with their beliefs. "Arguing" with them is futile - Jonathan Swift was dead-on when he said, "It is useless to attempt to reason a man out of a thing he was never reasoned into." Liberalism is emotion divorced from reason; its main function is to make its adherents feel righteous.

Too funny. Did this make you feel better?

I like how you resort to ridicule without responding to a single point Mr. Berger made, precisely proving his point. Well played.

He made no points. It was a harangue written in his mind before he saw this or any other study.

Pretty funny, but it was not a harangue, but rather a summary of what I have observed over and over again. I invite you to provide a counter-example of a program (other than defense) where the libs/progs have concluded that it was not working and ended it.

In a related vein, Ted Cruz recently jousted with the head of the Sierra Club regarding their position on Global Warming (aka "climate change") and many comical pauses ensued -

The science is settled, heh.

It's fairly easy to replace "liberalism" with "conservatism" in that vomit of words by Rich. Being in the middle of things politically, when someone from a "side" erupts like this, I giggle, wondering what their life is like to think of them spending their time making such a useless comment on a blog.

And for the life of me, I can't see how me making fun of him makes his point - considering I didn't say anything about the subject being discussed. duh.


I proposed a challenge which you have not met.

You need to think at the margin. The most ardent liberals and conservatives will not be persuaded by any evidence that their opinion is wrong. They usually avoid engaging with opposing opinions by dismissing them out of hand and pointing out that it is pointless to debate the issue because the other side are morally corrupt or stupid. They justify the knave/fool model of their opposition by pointing to the unflinching ideologues and political opportunists. This way they don't really have to examine their own beliefs with any thoroughness. But there are plenty of "liberals" who think that pre-k works because they believe the evidence supports it. These liberal's opinions are shifted by evidence, the same goes for conservatives closer to the margin.

True, but as Jan asked above, is anyone going to now refrain from sending their kids to pre-school, based on this new 'evidence'?

I suppose that the issue is really about how much should the public subsidize it, on that topic I'm a bit torn. It may be a use of funds with a low ROI, but I'd rather fund this low ROI project than one of the many others we fund currently. Education and support for children seems like a #1 priority to me.

Yes, the children are our future, so we must pour money down the drain.

It's a negative ROI project, right? We're harming the kids, albeit not much.

Great comment.

As msgkings suggest, most people aren't going to change their opinion of sending their own kids to pre-school based on this one study.

Sifting and winnowing, right? This is but one study and the topic is incredibly important moving forward, if we believe that most work will be knowledge work of some sort. Personally, I'd rather have more people (at the margin) productively contributing to the economy and with that, investments in human capital matter, so let's make good ones.

Seems to me like there is a confounding variable - the parents.

If you look at the Vanderbilt study, the pre-K students were all pulled from families with lower socioeconomic status. I think that a better reading of this result is that state-provided pre-K substitutes for less-involved parents, but the effects of the substitution fade over three years relative to the children who receive ongoing parental support.

We need government parents for underprivileged children and government girlfriends for involuntarily celibate men.

Exactly, HL.

...we do have government parents...its called the foster care system...which takes away children from bad parents and gives them to better ones...

Where are the government gfs?????

How can parental characteristics be a confounding variable when this evaluation was a randomized controlled trial? If randomization was done correctly then observables and unobservables are balanced between the treatment and control groups.

You randomize for a group needing treatment. Giving radiation to a group without cancer does not increase well-being.

Thus I can see the argument that if you identify a group in need, that should be your test group.

The control group was ALSO "pulled from families with lower socioeconomic status." The control group was the kids who applied to but lost the lottery to get into Tennessee Voluntary Prekindergarten. The researchers also compared the groups on about twenty variables and found them very similar.

This is consistent with earlier literature, like the Quebec study. Obviously the ESL kids were mostly people like me or Asians, not typical Hispanics. I, too, caught up quickly w/ non-ESL USAians and surpassed 90+% of them.

Are there a lot of Asian sociopaths?

The results indicate some sort of burnout, or more likely a reversion to the mean. There is no gap at age 6 or 7, and the gaps at ages 8 and especially 9 seem negligible. If the study had school outcomes beyond age 9, you could perhaps separate whether this is a burnout effect or reversion to the mean. I assume it's the latter. It doesn't surprise me that pre-K investments wear off, but I am not convinced that they should produce negative long term benefits.

Wouldn't kids who start on a regular schedule get burn out at some point too? Do we see universal dips int he 3rd and 4th grades?

I think I started despising school around the 6th grade, and it didn't stop until college.

Well, if the theory is that kids placed in pre-K are susceptible to burnout because of repetitious lessons (i.e., familiar subject matter being covered in Kindergarten for the benefit of children who did not attend pre-K), then we might expect that result to be unique to pre-K students entering a mixed K-12 population, no?

I wonder if kindergarten does any good. BTW I loved kindergarten.

The kids who arrive not being able to count to 10 and name colours probably get a lot out of kindergarten, but I'm not sure of its academic benefit for the rest of us. I think the only thing I remember from kindergarten that would have been new is some kids songs which are traditional to the area I studied in and some fables.

Other countries seem to start school much later and do fine:

I really think the babysitting angle of early education is under-emphasized, perhaps because people are a little ashamed of it.

Pretty interesting, and apparent from it that it is possible to design a too academic preschool.

Lots of program are designed this way:

1. See what the smart kids do.

2. Make everyone else do that.

If you are blank-slater, this is really appealing, because we could make everyone smart. However, what it really does is push everyone through a grind that most people -- except for the insufferable nerds, like me -- utterly despise.

It's really cruel to keep on trying to force people who will never be engineers to do calculus, but for some reason we keep at it, for their own good.

Stop trying to make everyone into nerds. It doesn't even make the nerds' life better.

By then end of my kid's kindergarten the kids know how to read. It's more like starting first grade a year earlier than the kindergarten I went to.

Hasn't knowing how to read after kindergarten always been typical?

No, neither my wife or myself (Kindergarten 1980) or our siblings knew how to read coming out of Kindergarten. My wife and her sibling's kindergarten was only half day. Kindergarten was just getting used to being at school and similar to the preschool my kids went to.

insert epic #humblebrag here

I don't think Kling is right. Heckman's principal criticism does not seem to be that the Vanderbilt study uses noisy third grade data as its primary evidence of fade out. Here is what Heckman writes:
Vanderbilt Pre-K Study: You get what you pay for
by James J. Heckman | Oct 05, 2015

Vanderbilt University’s study of Tennessee’s Voluntary Preschool Program evaluates a low quality early childhood program using a flawed methodology. Randomization was corrupted by noncompliance with the intended experimental protocol. The press release accompanying the report exaggerates the importance of the findings and the quality of the evidence.

Many students assigned to treatment refused to cooperate and the investigators were forced to use the very same non-experimental methods that they fault other studies for using. Their claim of program effect fadeout is a consequence of the control group catching up, not a decline in performance in the treated.

The real lesson from the program is that you get what you pay for, and Tennessee did not put much into its program. Low quality programs produce weak and even sometimes harmful results.

This study does not refute the strong evidence that high quality preschool programs—criticized by the authors as “too expensive”—more than pay for themselves in terms of ROI.

Randomization was severely compromised in the studies that Heckman prefers, too, but it doesn't seem to bother him there. For example, "in the Abecedarian program... seven families assigned to the experimental group and one family assigned to the control group [dropped] out of the program after learning about their random assignment" (Wikipedia).

The authors of the Vanderbilt study also show that the program is not lower quality than other programs, including ones praised by Pre-K advocates.

We can still evaluate the results by including those kids with the group they were originally assigned to for the analysis, see if it changes the outcomes.

If only one could design a pre school arrangement so unlike the k-12 and every other human endeavor the results would be amazing. So amazing!

Give it up. Handing over your children to government bureaucrats is only a job creation program for the otherwise unproductive who would be better off breeding and having their own children to raise.

Would you support a pre-school voucher program?

A tax decrease is the most fungible of vouchers.

Well then poor people wouldn't be able to participate, would they? Now, I know that is what you're hoping for, but be transparent.

Why would poor people not be able to participate?

Wouldn't progressives step up and personally donate some funds to create the necessary scholarships?

No, the idea is that kids should be able to go to pre-school, no matter if they were lucky enough to be born to parents who can afford it.

Leftists never donate their own money! That's the first rule of leftism!

I donate my money to subsidize tons of rich people's mortgage deductions. Do we count that?

By the way, this is the same argument as liberals should just send uncle sam extra money if they think taxes should be higher. It is probably the stupidest policy argument there is. When you use it, you've lost.

"The real lesson from the program is that you get what you pay for." As always, the excuse for every failed government social program is insufficient funding, and the solution is more funding, accompanied by higher taxation.

It is really pathetic that Heckman is still pushing the Perry and Abecedarian studies. They were implemented 40-50 years ago in a completely different society, and knowing what we know about the very low reproducibility of small, non-replicated experiments, it's insane to give these studies the weight he does.

Heckman also writes:

The decision to judge programs based on third grade test scores dismisses the full range of skills and capacities developed through early childhood education that strongly contribute to future achievement and life outcomes.

The Tennessee program measured a range of non-cognitive outcomes, and by the end of the third grade, there were no significant differences in these outcomes between the treatment and control groups (Table 12 in the report).

Heckman also claims that non-cognitive "skills facilitate better performance on achievement tests despite treated children performing no better on IQ tests." This was refuted in a recent reanalysis of one of Heckman's studies, described in detail here. The statistical and theoretical ineptness of that refuted Heckman study raises questions about the veracity of his other work on pre-K, non-cognitive skills, and so on.

The Heckman comment I was quoting from is here:

Do you have alternate approaches that follow subjects well into their adult lives? By definition they will be at least 2 decades old.

The fact that we don't have good studies with follow-ups to adulthood does not mean that we should accept the results of bad studies and spend billions of dollars on scaling them up. Anyway, if Heckman had begun a well-powered replication effort of one of his favorite studies when he first started enthusing about them, we would soon have results on adults from it.

Now we know that life doesn't end at birth, it ends at third grade. I suppose that's an improvement of sorts.

In sum, the pre-school was crappy and of low quality, only used for poor people.

And the randomization didn't work, because many kids' parents refused to put them in the pre-school. Therefore, the results are flawed. They should have used an intent-to-treat analysis and also presented those results.

It's funny how this is becoming the progressive conventional wisdom: The only solution is to send send all the black and Hispanic children in the country to "good schools" -- i.e., ones where the overwhelming majority of students are white.

It's a little hard to reconcile the arithmetic of that with the arithmetic of the other pillar of progressive conventional wisdom: the only solution is to import vast numbers of Hispanics so that whites will no longer be the majority.

But who cares about logic when it comes to race?

I think the logic is more to have a good mix of folks in every school. At least that's how it plays out in Canada. In my personal experience, I found that it added a lot to the value of education, having the perspectives of people from diverse backgrounds in every classroom.

Please define "a good mix".

Gee, that's not subjective.

How about a handful of kids in every class from some different religions (e.g., Jewish, Muslim, Buddhist) and parts of the world (Asian, Middle East, African), ideally with good representation across the socioeconomic spectrum, drawing kids from both poor and rich areas?

It might not be very relevant in math class, but in science you might find someone pointing out that in fact something was invented elsewhere, in history class you get a totally different perspective in class discussions, in English (literature) there can be some wacky creative cultural mix to many projects and discussions, etc.

But if you disagree you're wrong. Because it's not subjective :)

This is already the case in America.

You must be imagining what a person from NYC imagines when they hear Tennessee.

A school full of Canadians is a pretty good mix.

I bet those future Turkish-Kurdish, Shia-Sunni classroom discussions will be enriching.

That's why PISA test scores in Finland and Taiwan are so high: diversity!

Maybe they have good education systems and students study hard? I imagine an extra two hours a day of doing homework over 10 years positively contributes to being able to perform well on a test that evaluates things including reading skills, math skills, logic, etc., in addition to the ability to buckle down and focus on a test.

You imagine this, but what if it's not true, and you are just torturing kids for 2 hours a night for 10 years for nothing?

Yeah, Steve, that is exactly what everyone is saying.

I really don't think that you are correctly understanding Jan's argument, and I don't believe for a second that it's not technically possible to make a good pre-school composed of entirely poor kids. The behavioral problems that we might think of as endemic to poor secondary schools either do not exist at the pre-school level (no gangs) or exist at all pre-schools fairly equally (three year olds go crazy no matter what the income level). It really probably is an investment level and administration thing. I think the real issue is that scale matters, so vouchers or whatever is probably better than a state-run program.

So, perhaps pre-kindergaden has no effect on child development. However, programs like these are the reaction to the Joneses grandparents educational status consumption race that started many years ago. Since every child is gifted and needs extra educational consumption......these are the results.

Yes. At least in regards to private pre-school. In my locality (Los Angeles) pre-school is more of a status good, than a child enhancement. No way you could convince my wife not to send our kid to a top pre-school because all of her friends send their kids to top pre-schools. Also, for many people, going to pre-school is about getting a great private elementary school since many pre-schools are feeder schools into private schools.

Also, this has nothing to do with whether or not the state should spend more money on pre-K. I'm pretty sure it shouldn't.

Looks like the Kindergarten teacher focused on the problem students, leading to the VPK students to fall behind; VPK students are better prepared for Kindergarten, but their teachers are not teaching them. Then, the label of underperformance in first grade has a stronger stigma than overperforming in Kindergarten. In other wods, the teachers were unfair.

Excellent display of liberal logic: The ivory tower theory on which we spent a bunch of taxpayer's money didn't actually work. In fact it actually caused more harm than good. BUT THAT'S ONLY BECAUSE WE DIDN'T SPEND ENOUGH TAXPAYER MONEY!!!!

This isn't about the children. This is a typical cover your butt manoeuvre to buttress the failed industrial schooling model. The kids are failing because we don't have them early enough or long enough. The whole endeavor coincides with so many social goals; proper grooming of young minds, more teachers for the union meaning more dues for political hardball, another sinkhole where taxpayers can be extorted 'for the children'.

The fact that it doesn't work is a benefit. Studies show that it will keep them poor and stupid. Only haters would have a problem with that.

Nowhere does the conservative echo chamber hurt your more than in the politics of scandal and conspiracy theory.

This type of rhetoric is why you now have the northern version of a weak-minded Kennedy spinoff running your country. Keep it up.

The pharmaceuticals, meat industry and oil industry buy off the regulators (via campaign donations to politicians on BOTH sides), and this is defended in the name of free speech.

But when a public service or support is offered, the only explanations that come to mind are the dumbing down of minds, intentionally trapping people in poverty so they will vote for more freebies (forget that it costs money to get out of poverty), and public sector workers salivating at the opportunity to brainwash impressionable minds to support the leftist agenda.

(Republican folklore).

Weak minded, yes, but Obama is a Kennedy spin-off? That's a new one.

The likely case is that these types of studies are impossible to do, like many dietary studies, or paleo-climatology. Therefore the politics creeps in.

If you can't do studies, look to the market. The market for teachers tells you this: that teaching is increasingly important the older a student gets. This squares with the obvious fact that a human being is capable of absorbing more complex and profound information as he or she ages.

It's possible the market is right, that teaching young children is largely a baby-sitting function and should be compensated accordingly. There may be a harm reduction value in removing poor children from bad families, but that's not really teaching per se.

This is not an attempt to knock or demonize teachers, just a reflection of the obvious: immature humans don't need sophisticated supervision until they are able to absorb sophisticated information and concepts.

Or teaching the younger students may be as/or more important than teaching the older, but the supply of those who can teach the older students is smaller and thus they cost more.

Your "obvious" is not obvious.

"The market for teachers tells you this: that teaching is increasingly important the older a student gets."

I presume this is a reference to the higher compensation rates as one moves further up the educational food chain. Which tells us very little about how important a teacher is at any given level of development, and much more about how difficult it might be to teach at that level of development.

As you rightfully observe, younger children are not capable of handling sophisticated concepts for the most part; they're in school to learn basic concepts such as the major mathematical operations and common English grammar conventions. So it stands to reason that the supply of teachers will be highest at the lowest level, because the simplicity of the material makes for the widest possible pool of prospective qualified teachers, while teaching more sophisticated concepts requires greater intelligence, which in turn lowers the supply, thus increasing the price.

That said, learning common English grammar conventions is really, really important, because (1) a poor communicator is going to have a great deal of difficulty finding a job, and (2) a poor communicator is not going to be able to interact with teachers presenting more sophisticated ideas. So there's an argument to be made that early education is, at least at certain stages, more important. The questions are figuring out (1) where along the line that point falls, and (2) how to cost-effectively deliver the education.

Me touring pre-schools:

Administrator: "And here, we are doing some really innovative educational techniques. You see, we let children grab toys of their choosing off of the shelf, and then they can play with the toys by themselves or choose to play with other children with it. This is called free-play. Our educational methodology is called Play-based."

Me: "Um..."

I see your study and rise you a much more comprehensive set of surveys that show the opposite:

I also note the recent study proving most study's are wrong:

Richard Feynman offered that from a distance it seemed as if a great deal of research in education was 'cargo cult science'. All the trappings are there, but no planes come to land. E.D. Hirsch's interpretation is that the output of teachers' colleges suffers systemic problems: misspecified models, research methods inappropriate to task, &c. which produces one project after another which isn't replicable or which has inconclusive results.

Everyone will interpret the study as having confirmed their previous biases, and, of course, the moderator gets the ball rolling by remarking the way in which the study supports OPEN BORDERS.

This would only be supporting evidence if most immigrants were high performing Asians or similar.

I was just reading some 1915 vintage quotes about how we can't let the coolie vermin in.

Trends in racism.

This can't be a coincidence, but the NYT just published an Upshot column by David Leonhardt about a report released today by the Urban Institute in which schools are graded on a curve, propelling traditionally low performing states (such as Florida and Texas) up near the top while dropping traditionally high performing states (such as Nebraska and Iowa) well down the list. Something for everybody.

I've been pointing out for years that Texas does really well on the federal NAEP test considering its demographic makeup.

I'm becoming less confident, however, that proves Texas does a good job educating the students it has. The NAEP is a low stakes test, and Texas may be doing things to goose NAEP scores that other states aren't doing. I'd like to see a study of SAT / ACT scores by race in Texas vs., say, California, to see if Texas' advantage holds up on high stakes tests.

Advanced sports statistics enthusiasts have known for years that you have to adjust a team's win-loss record for strength of schedule to get a true evaluation of the quality of the team.

I think the main point of "pre-school" is not to prepare children for school; it is taking care of the children when the parents are working.

Then the proper comparision is "probability of having an accident in pre-school versus alone at home or playing in the street".

If the point is merely to provide daycare, make the case for that and reap the political benefit of a cheaper proposal to boot.

School is awful. Kids hate school. Starting kids in school earlier brings that out sooner. I hope this doesn't surprise anyone who actually remembers what being in school was like.

Perhaps the kids who learned their abcs and numbers in preschool found kindergarden and 1st grade boring. You can't expect the pre-schoolers to stay ahead if you make them re-learn the same material.

It has nothing to do with how well children are doing in third grade. It has superficially everything to do with how moms feel at 3 o'clock, a strange thing on which to base policy, unless there are other motives.
And were follow-up studies done with mom, they might find in many cases that the happy feeling of freedom the first time you drop your child off for full-day kindergarten, fades much like the reputed achievements of early-childhood education.

It sounds like further study should identify selection criteria for pre-k assistance.

Surely the set of children which can benefit cannot be null.

Neither, is it apparently everyone.

Good point. I'd like to know what the "control" kids did during that "pre-K" year.

It appears that most of the pre-school kids were in fact out of control.

I'll show myself out.

A small suggestion to those who want pre-K and how to gain more support:

1) Vouchers

2) Only allow private programs to avoid public Pre-K with unions, and its associated laundering of taxpayer money to a certain party.

I proposed vouchers above. You said no.

If I'm lucky enough to have children, they will be sent to Pre-K as soon as possible. Not for any cognitive benefits, but to get them out of the house.

But you must not think it would be harmful, then. That is, you also don't believe the results of this study.

The effect is pretty small, and it's possible it's specific to the low-SES kids subjected to the program.

So maybe the babysitting out-weighs this result.

That's what I would be betting on. Also, that the possibility that the differences in performance are small enough not to have real-world consequences.

FWIW, preschool (3 mornings a week for a few hours) is usually fun for little kids in a way that daycare (5 days a week for 9 or 10 hours) is not.

It might be worth sacrificing a little bit of outcome for a better childhood.

Right, that's my recollection: 5 or 10 hours of pre-K nursery school per week was a blast. But 40 hours per week sounds like it would have been a complete drag.

Exactly. The pre-kindergarten year was a chance for my only child to play with other kids, 3 mornings a week, 9 to twelve, in the care of a couple of women with far more experience of children than I. The benefits accrued precisely from our being able to choose a limited amount of it.
Even full-day kindergarten is exhausting for children. You can only take so much enrichment. And the ones who head to after-school care in the cafeteria when the others leave at three, wear the expressions of condemned men.

Seems like our school system in Ontario is unique. Kids start full-day kindergarten at age four, and it lasts two years. The "junior kindergarten" year is English-only. Full-day French immersion (for those families who want it) starts in the "senior kindergarten" year (age five) or Grade 1.

Pretty obvious design flaw in the study if the teachers were aware that their students had been in a program to lift their level of attainment. Being in the program predisposes teachers expectations of the quality of the student they believe they are dealing with.

Have just been reading Leonard Mlodinow's Subliminal: How Your Unconscious Mind Rules Your Behavior

Citing research by Rosenthal, teachers expectations shape student academic performance. Teachers were told that random students were lower IQ and others higher, and that some had higher potential than others. Teachers did not know the actual IQ scores, but false scores. Teachers reported later that kids labeled not gifted (but who actually had higher IQs) rated them as less curious, and their subsequent grades reflected that.

More interestingly, though, were the results of IQ tests given 8 months later: 80 percent of the kids falsely labeled as gifted increased their IQ by 10 points. ""What's more, 20 percent of the "[assigned} gifted group gained 30 or more IQ points, while only 5 percent of the other children gained that many [IQ points]....branding a child as a poor learner will to contribute to making the child exactly that."

The teachers knew. They should not have told the teachers.

There is another deficiency in the study as well, but I dont know how it is avoidable.

When they do a randomized trial in which some kids get treatment and their friends who are in the same social economic status do not then it is perfectly logical, given social influence, that the treated kids might gravitate towards the untreated kids performance, if they are friends and socially interact.

It might be better to do randomized trials with sets of kids who interact with each other during the treatment and measure treatment effects between sets.

If you believe in social network theory and analysis, randomized trials treating people differently within the same social network might lead to results that are different than if the complete social graph of a kid is treated with the same treatment.

From the paper itself: "However, by the end of the second grade there are no longer any significant differences between TN‐VPK participants and nonparticipants, and that pattern continues into third grade with the exception of a marginally significant positive effect for the TN‐VPK participants on the teachers’ ratings of peer relations."

Lots of questions:
- Even small improvements in socialization are hardly negligible. The question is, are they economically quantifiable?
- Don't we need to keep following these kids to determine if advantages "fade" or are merely latent or cyclical?
- Am I right that the "randomized" non-participants turned down a seat in the program? Meaning that their parents had
a more desirable option (which in itself can be interpreted in many ways)?
- Is it at all possible that the kids in the pre-k program were *pulling up their peers*...
- ...until the kids with more natural advantages at home eventually passed them?
- Does that equate to levelling the playing field, or raising the bar? It would seem to be one of the two.
- Or maybe even low-performing pre-school in Tennesee is better than its average second grade?
- Even if the test scores and behavioral analyses come out dead even or worse, early childhood
education and childcare yields additional social and economic benefits for both parents and their
employers, not to mention the broader economy, and the long-term welfare and usefulness of
individuals. You need to stay tuned for another twenty years and figure out how many of these
kids stay out of jail. Jail is daycare at public expense for the rest of your life.

The paper may have answered some of the above, but I didn't manage to tease out a satisfactory answer. In any case, anyone drawing hard conclusions on the value of pre-k from this kind of research is thinking rather small.

The control group was kids who lost the lottery to get into the program.

Even if the test scores and behavioral analyses come out dead even or worse, early childhood
education and childcare yields additional social and economic benefits for both parents and their
employers, not to mention the broader economy, and the long-term welfare and usefulness of

We don't know that. We know that pre-K provides daycare but we don't really know what ordinary pre-K (not a non-replicable boutique program like Perry Preschool) does for people long-term.

I was reading another blog entry with this same topic the other day.

It seems it is well established that early education is harmful.

Meh, that's academic training. Most educational methodologies in pre-school don't focus on academics: Montessori, Developmental, Play-based, Reggio Emilia, Waldorf, etc.

If it was a big positive effect we would likely have found it by now.


For example, that graduating from high school is a very big deal leaps out at you from all sorts of different data when looked at in lots of different ways. But the pre-K stuff, I don't know, it seems more like maybe, maybe not. It's certainly worth studying, but whether or not it's worth spending colossal amounts of money on seems up in the air. I'd say, rather than have the federal government jump in massively, let some states try spending a lot on pre-K and see how it works out for them: the Laboratory of Democracy and all that.

Jim Heckman has some comments on how the good news about early childhood education is getting lost over concern with third grade test scores, rather than more meaningful life outcomes..

We have massive longitudinal studies of young people, such as the NLSY79, which now has data on thousands of the children of the young women who entered the study in 1979. There's also the NLSY97 that now has 18 years of data on over 10,000 individuals. There's a huge ongoing study in, I believe, Otago, New Zealand of every child born in that city about 43 years ago.

We have a huge amount of data to use to adjust for other factors besides pre-K attendance, so what do these databases tell us about pre-K?

I realize everybody likes to call RCTs the gold standard, but as soon as they get published everybody starts to squabble over methodological details. So, what do we learn about pre-K from the silver standard of giant longitudinal tracking studies?

What is telling is that Tyler thinks this is the best study he has seen. Other studies have followed kids much longer. They have that the kids who had pre-school had lower rates of incarceration, better health and higher incomes. Tyler appears to believe that those are less important than third grade test scores, or he just hasn't seen those tests.


Those others were conducted 50 years ago with tiny samples.

Those others were boutique pre-schools. Never duplicated. And very different from what universal pre-K would be.

The point of the preschool movement is not to turn the kids into high achievers.
Precisely the opposite, to brainwash them early on to become dependent on (government managed) "help".
It seems to be working as planned.

Which is why affluent parents seek out pre-school for their children.

Comments for this post are closed