One controversial issue is whether a large-scale or universal pre-k program is best delivered through the public schools, or through private preschools.
My assessment of the evidence is that either approach can work. Oklahoma’s near-universal pre-k program seems to be successful. Oklahoma’s program is largely delivered through the public schools. But Georgia’s large-scale pre-k program also seems to be successful. Georgia’s program is delivered through a charter school approach, under which a variety of providers are authorized to receive a state payment per preschool child enrolled.
On the other hand, Florida’s approach to large scale preschool is generally considered to be less successful. Florida relies on vouchers to preschool parents. But the vouchers are modest enough that it is difficult for many Florida preschool programs to be high-quality. It is quite possible that with greater funding and additional regulation, a voucher approach to pre-k funding might also be successful.
As a practical matter, if a local area already has many high-quality preschool providers, it would seem wise to somehow make them eligible for funding in an expanded pre-k system. This seems wise for several reasons.
First, it seems politically sensible. The threat to private preschools seems to have motivated some of the opposition to the 2006 California ballot initiative to expand state funding for preschool. Opposition from some private preschool providers also seems to be motivating some opposition to expanded pre-k in New York State.
Second, it seems more sensible to build on what you have than to recreate it. If a local area already has some high-quality private preschool providers, there seems a lot of waste in driving them out of business and replacing them with public school provision.
Third, there is the ongoing argument that competition among different school providers may lead to more innovation, more ability to address diverse needs, and higher quality. Whether this is so in the preschool case obviously depends upon how that competition is structured. For example, it is not obvious that parental choice by itself, without any regulation, will be sufficient to ensure pre-k quality. But with some regulation that provides minimum quality standards, and promotes better measures of quality, a choice among different providers may promote quality.
Your claims about the three states you mention above are debatable. Florida, in fact, has increased 4th grade reading scores. Since 1998 the 4th grade reading scores have consistently gone up. In 98, the score was well below the national average at 207. In 2009 the 4th grade reading score was 226 which is above the national average.
The opposite can be said for Oklahoma and Georgia. There were no significant increases in 4th scores in Georgia and Oklahoma scores actually decreased. This information is taken from the Nation’s Report Card.
Oklahoma – Before implementing UPK in 1998, Oklahoma scored 220 on the 4th grade reading scores. It was above the national average of 215. Since implementing UPK, 4th grade reading scores have dropped below 220 and have been below the national average every single year.
Georgia – Since they implemented UPK in 1995, the 4th grade reading scores have slowly increased over the years. However, the increase is not significant and is still below the national average.
To sum up, Florida spends the least amount of money and is achieving the most significant results yet NIEER says they are low quality. Perhaps NIEER should reassess their benchmarks for quality and focus more on results.
Cindy: Thank you for your comments.
A brief reply for now. I may have time for a more extensive reply later.
One important point is that with multiple causes of state trends in 4th grade test scores, it is impossible with just aggregate data for three states to reach much in the way of conclusions about the effects of different pre-K models.
So, for example, surely 4th grade reading test scores are affected by more than the availability and quality of pre-K. At a minimum, these scores would be affected by the composition of different student characteristics in the state, the quality of state education K-3, and other social trends in early childhood. With all these factors playing potentially major roles, it would be difficult indeed to tell whether any aggregate trends are due to what is going on with pre-K.
In statistical terms, with only 3 aggregate observations (changes in average state reading test scores for 3 states), it is really impossible to reach a firm conclusion about the effect of any state-level variable.
Finally, one of the main point of my post was that a variety of models might work for universal pre-K, including a charter model and a voucher model. Both of these models include private delivery.
A second main point of my post was that for political and substantive reasons, a universal pre-k model would be wise to include private preschools. Do you disagree with this conclusion?
After perusing NAEP data and various analyses of NAEP data a bit more, I have some more comments.
First, since universal pre-k was started in Florida in the 2005-2006 school year (see NIEER 2009 State Yearbook section for Florida for preschool enrollment trends by years), and the 4th grade NAEP data is only available now through the 2008-09 school year, there are no Florida 4th graders for whom we have NAEP test results after universal pre-k. (Note that because NAEP is administered in January through March, the 2009 NAEP is for the 2008-09 school year.) Whatever the 4th grade NAEP test score trends are in Florida, they have nothing to do with universal pre-k in Florida. We will have to wait until the results are available for the NAEP for 2011 (which is this current school year) to see if we can discern any effect of Florida’s universal pre-k program on 4th grade NAEP scores.
Second, I note that the observed upward trend in Florida’s 4th grade NAEP scores, prior to universal pre-k possibly affecting these scores, illustrates my larger point that state NAEP trends are due to a variety of factors, not just universal pre-k.
Third, there have been at least two analyses of how universal pre-k in Georgia and Oklahoma is correlated with 4th grade NAEP scores, one of Georgia by Fitzpatrick, and one of Georgia and Oklahoma by Barnett. Again, as these studies note, it is hard to tell whether state trends in 4th grade test scores are due to universal pre-k, or due to unobserved state factors. Both of these studies incorporate the appropriate time lag between universal pre-k and 4th grade test scores. Because Georgia started universal pre-k in the 1995-96 school year, we wouldn’t expect to see any universal pre-k effects on 4th grade test scores until 5 school years later, in the 2000-2001 school year. Because Oklahoma started universal pre-k in the 1998-99 school year, we wouldn’t expect to see any universal pre-k effects on 4th grade test scores until 5 years later, in the 2003-2004 school year.
Fitzpatrick finds some evidence of improvements in Georgia’s NAEP scores due to universal pre-k. These are particularly noticeable for white students, especially in non-urban areas. Test score improvements are also more noticeable for math than for reading. She finds some evidence that universal pre-k increases the proportion of blacks who appear to be in the right grade for their age, rather than older than expected for that grade, and this trend is particularly noticeable in urban areas. This suggests that universal pre-k is reducing grade retention among blacks. This is important for two reasons: (1) this is a non-test related behavioral benefit of universal pre-k, which may eventually affect high school graduation rates; (2) if more marginal students are promoted rather than retained, this may reduce the observed average test scores in 4th grade, even if most individual students are doing better.
Barnett finds some evidence that Georgia’s universal pre-k program increases NAEP scores for both whites and blacks, with some tendency for greater “effects” for math scores. For Oklahoma, fewer test score differences over time are significant for specific groups, due to limited sample sizes, but there do appear to be some significant increases in math scores.
It is possible that the greater increases in math scores could be due to math scores being more dependent on what the education system does than is true for reading. This suggests that any analysis of the correlations between state preschool policies and state NAEP trends should look at both reading and math scores. Fitzpatrick’s results also suggest that we should look at the number of students who are at the expected age for their grade.
I don’t think that these NAEP analyses should lead to doubts about the evidence from other research that pre-k and universal pre-k can make a difference to student success. This other evidence looks at individual differences in enrollment in preschool, not state differences in preschool policy. But this other evidence focuses on individual differences in enrollment in preschool that are due to random assignment, neighborhood assignment, or age, and which therefore are unlikely to bias the estimates of preschool’s effects. Exploiting these individual differences, rather than state differences, enormously increases statistical precision, and allows us to better separate the effects of preschool from unobserved factors.
Pingback: What can NAEP trends tell us about effects of universal pre-k? | investinginkids
None of the studies or statistics can account for all of the variables. There are so many to consider. My point is that it is not necessary to dump billions of public dollars into unproven programs. Florida is a perfect example of my argument. They have been successful with dramatically improving reading scores without spending inordinate amounts of money on universal pre-k programs, yet you say they are less successful. I’d have to disagree with you.
You refer to a study by Fitzpatrick. I read this study and this was Fitzpatrick’s conclusion. “The results of the study and it’s cost benefit analysis indicate scarce public funds may be used more efficiently by implementing target strategies in the designing of pre-k programs.” This sums it up. I agree with this approach and all the research supports this conclusion.
Thanks for commenting.
With respect to Florida, my point was that its 4th grade test score data cannot be used as evidence either for or against its approach to universal pre-k, as the program has started too recently for the 4th grade test scores to be affected by Florida’s universal pre-k program.
As you mention, Fitzpatrick takes a position in favor of targeted pre-k over universal pre-k. However, her empirical results find some evidence for positive effects of Georgia’s pre-k programs for a wide variety of students.
In her Table 6, she considers effects for 12 groups of students: 2 racial groups (white vs. black) times 2 income groups (eligible or ineligible for free or reduced price lunch) times 3 geographic areas (urban areas vs. urban fringe vs. rural areas). She considers for each group whether universal pre-k is correlated with math score increases, reading score increases, or being “on grade” for age.
For 10 of the 12 groups, the empirical results in Table 6 suggest statistically significant positive effects of Georgia’s universal pre-k program on at least one of the three dependent variables (math scores, reading scores, on grade for age). Of the 36 coefficients (12 groups times 3 dependent variables) estimated for the “effect” of universal pre-k, 17 are statistically significantly positive. None of the 36 estimated coefficients are statistically significantly negative.
The pattern of results does not suggest any simple targeting. There are more statistically significant positive effects in rural areas than in other areas. But there are no consistent patterns to upper income or lower income students having more positive effects.
Fitzpatrick’s benefit-cost analysis, is, as she says, a “back of the envelope calculation”, that is “a very simple cost benefit analysis”. This analysis is focusing on benefits from increases in tax revenues, not increases in earnings. A full social benefit-cost analysis considers all increases in earnings, not just the increase in earnings that goes to the government in tax revenues.
Fitzpatrick’s analysis also assumes that the “goal is to increase wages [and hence tax revenues] through test scores”. However, the pre-k literature frequently finds effects of pre-k on adult earnings that exceed those that would be predicted simply from test score increases. Pre-k may increases adult earnings due to increases in soft skills. Fitzpatrick’s results suggesting that universal pre-k increases the proportion of students who are “on grade” for their age may reflect such increases in soft skills.
As I said before, I find estimates using state variation in pre-k to be at best suggestive, not definitive. There are too many unobserved state trends that could be affecting the results.
I intend to address this issue of targeted vs. universal pre-k more fully in future posts.
Researches and economists advocating for Universal Pre-K are able to talk the jargon to make people believe a universal approach (or as you say, large scale approach) would benefit most children regardless of the conclusive evidence that not all children do.
Yes, children that would otherwise not have an opportunity to attend preschool, such as those living in rural areas will benefit. Yes, those with English as a 2nd language will benefit. Yes, those would come from disadvantaged backgrounds will benefit. Cherry picking the benefits and applying them to the larger population is not reason enough to implement government funded universal programs, whether they be provided by public or private institutions. We need to be much more responsible with our resources.
I keep going back to targeted vs. universal issue because the entire premise of your blog is to encourage the development of government funded preschool for all children. Obviously, I am just as passionate about this issue as the advocates and feel all sides should be presented to our representatives not just those the advocates choose to publicize.
Thanks for your comment. I will get into the targeted vs. universal issue in more detail in future posts. For the moment, let me just say that I disagree with you about what the research evidence says about how benefits vary across different groups of children.
My estimates of the benefits of universal pre-k do not assume that all children benefit equally. In fact, I assume that benefits are reduced for middle income and upper income children compared to lower income children. Obviously I don’t believe they are reduced as much as you believe, but I don’t assume that the effects estimated for lower income children apply to all children equally. Even under these assumptions, I find that universal pre-k has considerable benefits.
I should also add that the premise of my blog and my book is considerably broader than “universal pre-k”. For example, the book extensively discusses how targeted pre-k may have large benefits. The book also discusses the benefits of high-quality child care for the disadvantaged, and the benefits of nurse home visitation programs targeted at the disadvantaged. Finally, the book discusses issues in the design of business incentives.
As I will discuss in future blog posts, there are arguments (and evidence) pro and con on universal vs. targeted pre-k. On the whole, I think the arguments and evidence are stronger for universal pre-k, but I respect the opinions of people who differ on this issue.
I want to add one further point about the purpose of my book. The main point of my book is to argue that the research evidence shows that investment in high-quality early childhood programs can pay off in improved state and local economic development. This is a point that applies equally well to targeted as well as universal pre-k.
The issue of targeted vs. universal pre-k is important, and is extensively discussed in my book in one chapter, chapter 8. In that chapter, I present estimates for the rates of return for universal vs. targeted pre-k for different income groups, under different assumptions. I also discuss the arguments pro and con.
However, most of the book is not focused on that issue, but rather what early childhood investments can do for state and local economies, how early childhood programs should be designed to be high-quality and learn from experience, how their rates of return vary in different local economies, what are the merits of federal vs. state and local funding of early childhood programs, the ethical issues raised by government intervention in early childhood programs, etc. Almost all of this discussion applies to expansion of targeted pre-k programs as well as universal pre-k.
Some of the people who have given favorable advance reviews of my book include Nobel prize-winning economist James Heckman, and Art Rolnick, former director of research at the Minneapolis Fed, both of whom are advocates for targeted pre-k rather than universal pre-k. I will come back to the difficult issues about targeted vs. universal pre-k in future blog posts. However, I want to emphasize that this is only one of many important issues considered in my book.
Thanks. I look forward to your discussion regarding targeted vs. universal. I know how poorly the “universal” concept is working in New York. Again, I appreciate the dialogue.
When you do your research regarding economic benefits, do you take into consideration the negative effects these programs have on the local community? Are you considering the fact that local businesses are being put out causing people to lose their jobs? Unemployment rates go up, tax revenue goes down and those that used to pay for their own preschool are put on the tax rolls.
That is what is happening in my area. Many, quality preschools that have been serving the community for decades have been forced out of business. They aren’t reaching more children, they are just shifting the preschool population from the privately paid to publicly funded.
My brief reply is that my model takes into account that an expanded pre-k program, whether it is a “universal” program or more “targeted”, may displace some current activity taking place. Obviously if the program is adding fewer net new slots or spending, this reduces economic development impacts, both because there is less true new spending and there will be fewer additional child participants in high-quality preschool.
The goal should be reaching more children with high-quality programs, in order to achieve higher impacts on economic development and other goals. There are a variety of ways to do this. As I have mentioned before, some models rely more heavily on public school delivery, but that is not the only option, nor necessarily the most desirable option. Other models involve funding high-quality private preschools to expand services. What you are reporting for your area is not the way in which expanded preschool access is inevitably implemented.
Not to mention ripple effects. Buildings are empty because preschool owners that rented space aren’t there.
This link will take you to an excellent article that sums up the arguments I have been making.