NPR’s Morning Edition on February 18, 2013 had an interview by NPR host Linda Wertheimer with NPR science correspondent Shankar Vedantam that gave an unduly negative spin to what research shows about the effectiveness of universal preschool.
The program began by quoting President Obama from his State of the Union address:
“Every dollar we invest in high-quality early childhood education can save more than seven dollars later on — by boosting graduation rates, reducing teen pregnancy, even reducing violent crime. In states that make it a priority to educate our youngest children, like Georgia or Oklahoma, studies show students grow up more likely to read and do math at grade level, graduate high school, hold a job, form more stable families of their own. We know this works.”
Linda Wertheimer went on to ask Shankar Vedantam the following:
“President Obama says that spending money in preschool gives us seven times our investment. Where does that number come from?”
Shankar Vedantam responded:
“Those numbers come from a couple of studies called the Perry Preschool program and the Abecedarian program …that targeted very high-quality and fairly expensive interventions at very disadvantaged children.”
My comment: I think there is also very good and more relevant long-term evidence of success from the Chicago Child-Parent Center program. This program is much cheaper than Perry or the Abecedarian program. The Chicago CPC program cost about $5400 for a one-year preschool program at age 4 that was shown to have a high benefit-cost ratio. Perry cost over $17,000 for the two-year program that was tested; the Abecedarian program cost around $40,000 for a five year program that was tested. (See my paper on Tulsa with Gormley and Adelstein for sources for these cost figures.)
“ And what those programs found — they followed these children out not just for years but for decades – is that the programs didn’t have just cognitive benefits in other words improvements in performance in academic scores but they had life benefits, they … reduced the teen pregnancy rate, they reduced the crime rate, they had huge benefits later on. So the President is on very solid footing when he talks about the return on investment when it comes to those narrowly targeted programs. But when he rhetorically links those programs with larger programs, such as the experiences in states such as Oklahoma and Georgia, in some ways he’s venturing off the ledge of science. There have been studies looking at the experience of those states, and I have not seen any data that suggests that return on investment in those states is anywhere close to seven times our investment. “
My comment: My study with Gormley and Adelstein found large effects of the Tulsa, Oklahoma program on kindergarten entrance exams for both low-income and middle-income students. We used these short-term effects on test scores to project long-term effects that would increase the present value of earnings by three to four times the investment. This was true for all income groups of students. In many of the studies that do have long-term data, the anti-crime benefits are often greater than the earnings benefits when calculated in dollar terms. Therefore, it would not be surprising to me if the Tulsa program ended up having a benefit cost ratio, when all benefits were included, of seven to one or more. This is more likely for the low-income children in the Tulsa program, for which it seems reasonable to assume considerable anti-crime benefits. But even if middle-class children did not have any reduction at all in their crime rates, which is an extremely conservative assumption, we would have rates of return on investment of 3 to 1 for these children.
“I want to emphasize that studies in the states have found that the programs do have benefits, they just don’t have benefits of the same magnitude as the highly targeted programs. I spoke with Bill Gormley, he’s a researcher at Georgetown University. He said there were very clearly cognitive benefits among children in Tulsa, Oklahoma, who he studied in one of his studies.”
My comment: These cognitive benefits can be used to predict future “life benefits”. Studies such as those by Harvard Professor Raj Chetty and his colleagues show a connection between short-term cognitive benefits of educational interventions, and future life benefits such as increased adult earnings. Applying these findings to Tulsa preschool requires some extrapolation, but not unreasonable extrapolation.
“Remember that the programs that targeted the highly disadvantaged children were highly focused, they focused on the children who were most in need, and they gave them the very best resources. When you ramp up a program and you make it universal and you make it state-wide, what happens is some of the children who end up using the program are children who don’t really need the program that much, they would have turned out fine anyway.”
My comment: It is also possible that some highly disadvantaged children are so disadvantaged that the preschool intervention is insufficient to make a difference. On the other hand, some middle-class children may have more limited kindergarten readiness issues with social skills that can be effectively addressed by preschool. Therefore, theoretically it’s hard to tell who will benefit the most from high-quality preschool.
“And the second thing that happens [when you go from small targeted programs to universal, statewide programs], unless your budget expands astronomically, is that the quality of the programs tends to go down. “
My comment: Quality does cost money. But the cost is not astronomical. We can get results with a half-day school-year program for 4-year-olds that costs about $5,000 per child. (See the Institute for Women’s Policy Research study entitled “Meaningful Investments in Pre-K”.) If we project out these costs to the roughly 3 million additional 4-year olds we would need to enroll in high-quality preschool to get to “universal” preschool, the additional cost is about $15 billion per year. (See chapter 4 of my book Investing in Kids. The Center for American Progress has a more expansive program for three and four year olds that appears to have a full implementation cost of about $25 billion per year.) $15 billion is about $50 per capita. This is an amount that is affordable by either the federal government, or by state governments on their own. We can see that it is affordable in practice, because states such as Oklahoma already do fund high-quality preschool as an entitlement for all four-year-olds.
Vedantam then responds to a question about Head Start:
“I think Head Start has done wonderful things, but in terms of the return on investment and how effective it is at boosting scores over the long-term, there are significant questions that have been raised. There has been a Congressionally mandated study that just came out a couple months ago that found that even though the programs in general were pretty good, by the time children reached 3rd grade, there was really no difference between the children who had been through the Head Start program and those who had not.”
My comment: This misses that the research on the impacts of Head Start is quite mixed. There are good studies that show long-run effects of Head Start. I have reviewed this in detail in previous blog posts.
“In some ways I think that the issue of Head Start raises the larger question which is there’s a tension here… between what’s politically popular, which is parents love these programs, they love subsidized child care, they love preschool, and lots of people want to sign up for them. Scientifically, though, what seems to have the biggest benefit is when you target your limited dollars at the people most in need… How you square the popularity of a program that reaches all families versus the benefits of a program that [is] more narrowly targeted, I don’t know how you square that politically.”
My comment: I think this is an unfortunate wrap-up that may be attributable to this being an unscripted interview. Head Start is targeted. It is arguable that part of the problem with Head Start is because it is so targeted on the poor, there is insufficient political pressure to continuously improve the quality of the program. The notion that more targeted programs will result in more resources for low-income children may not be true, because such programs lack as much political support.
In addition, none of Mr. Vedantam’s comments address possible peer effects in preschool. Is it really optimal to put low-income kids in income-segregated preschool classrooms? This is another issue to consider in analyzing the pros and cons of targeted versus universal programs.
In sum, I don’t think that Mr. Vedantam’s report provides a balanced assessment of the research evidence for targeted versus universal preschool. It is fair to say that there is more long-term evidence from random assignment studies of preschool’s benefits for low-income students. We don’t have long-term random assignment studies of preschool’s effects on middle-income students. But as the cliché goes, absence of evidence is not evidence of absence. And we have good evidence that large-scale universal state preschool programs produce short-term benefits for both low-income and middle-class children that are consistent with sizable long-term benefits. The costs of implementing such universal programs are not astronomical, and there are some political as well as substantive advantages to having income-integrated programs.