THE PROFESSIONAL JOURNAL FOR EDUCATION
PHI DELTA
THE PROFESSIONAL JOURNAL FOR EDUCATION
ONLINE ARTICLE | ||
 |
|
Using SIMS, TIMSS, and NAEP data sets, Mr. Biddle reveals that levels of school funding and rates of child poverty in the U.S. are strongly associated with differences in eighth-grade mathematics achievement among both districts and states. If our schools are to be improved, he suggests, Americans must ignore nonsensical arguments about higher standards and face up to the real problems of education.
ONE OF THE stranger episodes in the history of American debates about education is now unfolding in Washington. As a result of continuing claims about a "crisis" in our schools1 and assertions that our country is not now "first in the world" in achievement, Americans are being urged to adopt all sorts of controversial "reforms" in education. For his part, President Clinton is asking that each state adopt tough standards for achievement in mathematics, science, and other subjects - arguing that if such standards are adopted (and enforced), achievement in those states will rise.
In making this argument, the President has been supported by a host of powerful voices ranging from Louis Gerstner, Jr. (CEO at IBM) to Ronald Wolk (editor of Education Week) to the late Albert Shanker (formerly president of the American Federation of Teachers) and others. This argument assumes that "one of the [major] weaknesses of American education is that we lack clear goals for what students should learn and how well they should learn it."2 If this premise is correct, then if some states set (and enforce) high standards, it is argued, parents in those states will demand more from their schools, and those schools will revise their curricula, adopt tougher textbooks, retrain their teachers, and otherwise reform their formerly lax procedures.
This line of thought presupposes that educators are either confused or feckless but (perhaps like employees in private industry) will mend their ways if only they are given guidance or a swift kick. For the moment, let us ignore the unreality of this assumption. The major problem with the argument is that it is not supported by evidence. It is easy, of course, to find examples of individual school districts or states in which standards and achievement are both high, or where both are low, but this does not mean that these events are linked causally. To support such an argument, one would need evidence indicating that, when districts or states that represent the full panoply of American students and schools set high standards, they also generate greater levels of achievement - by comparison with other, comparable jurisdictions where high standards are not set. But such evidence has not yet appeared, nor is it likely that we will ever see it.
This doesn't mean that research has failed to explore what actually generates differences in achievement among school districts and states in America. On the contrary, a good deal of research has now appeared concerning the real causes of achievement deficits, and evidence from these efforts suggests strategies for improving American education that are quite different from those being advocated in Washington. For the purposes of this article, I focus on two such causes, both associated with social problems that are particularly severe in our country: poor school funding and poverty among children. As some readers may know, controversies have arisen over the educational effects of each of these factors.
Poor School Funding
I begin with the social problem posed by American differences in school funding. As it happens, the United States differs from most industrialized nations in that much of its funding for public schools comes from local sources. As a result, funding varies sharply from affluent to impoverished school districts throughout our country.
How large are these differences? Actually, they are enormous. In 1995, for example, New Jersey funded public schools at an average annual rate of $8,118 per student (adjusted for differences in cost of living), whereas Arkansas, the state with the worst spending figures, spent $3,599 per student.3 Large funding disparities also appear within many states. In 1992, for instance, the disparity in per-student support for schools in Alaska at the fifth and 95th percentiles was an enormous $12,737 - whereas in Montana the difference was "merely" $9,171, in New York it ran $7,571, and in Wyoming it was $7,400.4 What this means is that a few American students (who just happen to live in rich communities within generous states) are now attending "public" schools that are funded at $15,000 or more per student per year, whereas other American students (who are stuck in poor communities within niggardly states) must make do with $3,000 or less per year in public school funding. Differences of this magnitude generate tremendous disparities in the quality of school buildings, facilities, and curricula, equipment for instruction, teacher salaries, and student/teacher ratios. Such disparities do not appear in other industrialized nations.
It seems obvious that high levels of funding will generate greater student achievement. Indeed, that is one reason why rich suburbs support superb schools. Moreover, legal battles have been fought in various states seeking greater equity in school funding. And yet evidence confirming the presumed tie between funding and student achievement has proved elusive.
Since level of school funding is correlated with family advantage, and family advantage is also presumably tied to achievement, one must control for the effects of advantage when estimating the (net) impact of funding - and prior studies using such controls have generated controversy. Nearly 100 studies have now been published that appear to bear on this topic. Many have not found significant effects, and some reviewers have decided that this means that funding has "no strong or systematic" net effect on achievement in America5 - a conclusion welcomed by some economists and ideologues hostile to the public schools, teacher unions, and the public sector generally. In contrast, other reviewers have looked at effect sizes reported in these studies and concluded that the net impact of school funding is actually "substantial."6 Their conclusion has been welcomed by educators, of course, but has been attacked by those who want to believe that level of funding has no impact on education; thus the issue has remained unresolved.
There is but one small problem with this controversy. Both of these conclusions presuppose that the typical study from this literature provides competent evidence - that its methods and designs were appropriate for estimating the net effects of school funding. Unfortunately, this is far from the case. Most of these studies were based on small and nonrepresentative samples and did not examine school funding directly but rather funding-associated school characteristics (such as teacher salaries) that may not have simple ties with achievement. Many also employed questionable operations, nonvalidated scales, poor regression models associated with multicolinearity problems, and inappropriate analysis techniques, all of which generate falsely depressed estimates for the impact of funding. In addition, the studies have used a surprising array of measures for family advantage that have different implications for student achievement.
This does not mean that all prior work on the issue is valueless. On the contrary, a few studies have employed large samples, good operations focused on funding, and appropriate models and procedures - and a number of these have reported significant net effects for funding.7 But until recently it has been hard to find good studies of the topic that have made use of national samples, and the latter are necessary if one is to estimate achievement effects for the huge range of school-funding differences that now exist in America.
Thus, because of widespread and serious flaws in studies from this research tradition, one cannot use the bulk of its "findings" to reach valid conclusions about the effects of school funding in America. This situation has not stopped confused reviewers and conservative sources hostile to public education from stating such conclusions, of course, but the evidence backing their claims has not been persuasive.8
Poverty Among Children
The United States also differs from other advanced nations in a second way that bears on educational achievement: the rate of poverty among children is far higher in America.
Again, how bad is the problem? According to official government estimates, the child poverty rate now exceeds 20% in our country. This means that at least one-fifth of the children in America are likely to live in substandard housing, have an inadequate diet, wear only cast-off or torn clothes, lack health insurance, suffer from chronic dental or health problems, and be members of a family headed either by a single mother or by two overburdened parents who subsist on welfare or work long hours at miserably paid jobs.
Such a poverty rate far exceeds that of other industrialized nations. Evidence confirming this unpleasant truth comes from a group of scholars who have been studying income distributions in Western nations for more than a decade, the Luxembourg Income Study (LIS). The latest LIS papers on child poverty report data from roughly 1990 and compare 18 industrialized countries.9 Their figures show that the rate of child poverty in our country is more than 50% higher than for all other nations studied and five to eight times greater than the rates for some nations with which we are often compared with regard to educational achievement.
Why should child poverty in America be so extensive? For one thing, recent shifts in the industrial culture, political climate, and tax laws of our nation have generated a massive upward redistribution of income and wealth - away from poor and middle-class Americans and into the hands of the super rich.10 For another, child poverty always tends to be high within families that are headed by single mothers, and America has a huge number of single-mother families. Above all, the U.S. does not provide various tax-supported services, common in other industrialized nations, that serve the needs of poor children: a national health care system, tax-supported preschools, paid leaves for child and prenatal care, universal child allowances, child-support programs for single parents, safety nets for families with special needs, and secure unemployment or low-wage supplemental income programs.
Child poverty is both a huge social problem in our country and an obvious generator of educational difficulties. |
But poor children are also the clients of schools, and they are uniquely handicapped for education because of their poverty. The homes of poor children provide little access to the books, writing materials, computers, and other supports for education that are normally present in middle-class or affluent homes in America. Impoverished students are also distracted by chronic pain and disease; have poorer nourishment; tend to live in communities that are afflicted by physical decay, serious crime, gangs, and drugs; and must face problems in their personal lives because their parents or older siblings have left home, died, been incarcerated, or lead seriously disturbed lives. All of this means that poor children have a much harder time in school than their more affluent peers.
Given the seriousness of the poverty problem, one might assume that a host of good studies investigating the effects of child poverty on achievement would have appeared. Actually, such studies are hard to find. Part of the difficulty seems to be that good data on the poverty of individual students or their families are not often gathered in America. Thus studies of poverty effects in schools must usually make do with indirect indicators, such as students' eligibility for free or reduced-price lunches. When such studies are conducted, they find, of course, that child poverty has a negative impact on school success.11 But many of these studies have not involved statistical controls, so it is often not clear whether their findings reflect poverty alone or also the effects of other "risk" factors likely to be correlated with poverty, such as race, ethnicity, or immigrant status. However, a few studies have employed such controls, and they report sizable net effects for student poverty.12
These findings are important, but they still leave a major question unanswered. Poor children in the U.S. are also likely to attend underfunded schools, so raw statistics indicating that their achievement is low may also reflect the inadequacies of those schools. Given this latter issue, one would assume that at least a few of the many school-funding studies I discussed above would also have examined the impact of child poverty. Alas, this has not been the case. Some school-funding studies have, indeed, included measures for child poverty in scales for estimating family socioeconomic characteristics, but it is difficult to find even one such study in which child poverty was measured by itself as an indicator of disadvantage. Thus one cannot tell from prior research whether child poverty has achievement effects that are independent from those of poor school funding.
In part, the paucity of good research on the effects of child poverty seems to reflect a general unwillingness to debate or even think about poverty and its impact within the current political climate of America. This is absurd when it comes to poverty and education. Child poverty is not only a huge social problem in our country but also an obvious generator of educational difficulties, and it deserves at least as much attention as any other component of (dis)advantage.
A New Research Strategy
Given the evident problems of prior studies, the time is surely ripe for better research on the achievement effects of poor school funding and child poverty in America. Ideally, such research should be based on sizable, random, national samples; should make use of good measures for school funding, child poverty, achievement, and various control variables; and should feature appropriate models and techniques for analysis. As might be imagined, to conduct such research from scratch would be very costly. However, good databases have recently appeared that provide information on the characteristics of school districts from across the nation, and it is now possible to combine data from these sources with information from other sources concerned with student achievement to construct less-expensive studies based on secondary analyses.
One of my students and I have recently conducted such a study in which we explored predictors of eighth-grade achievement scores for public schools using the American data from the Second International Mathematics Study (SIMS).13 Our project examined not only school funding and child poverty but also many control variables. Results from the study revealed statistically significant, net effects for both school funding (ß=+.296, p<.01) and child poverty (ß=-.358, p<.01). These effects persisted even when controls were entered for such potent variables as race and level of curriculum to which students had been exposed.14
Moreover, the results we found were substantial. District-level differences in school funding and child poverty explained more than 25% of the variance of differences in mathematics achievement. Since the American SIMS data were part of a larger international study, we were also able to compare our funding and poverty effects with those generated by cross-national comparisons. In doing so, we estimated that the mathematics achievement score for a typical "advantaged" public school district in America (i.e., one with high funding and low child poverty) would be above those reported for all European countries in the original SIMS data and second only to Japan - the top-scoring nation in that study - whereas the score for a typical "disadvantaged" American school district would be below all European nations and approaching those of Nigeria and Swaziland!
As it happens, the first of these estimates has now been stunningly supported. Data for IEA's Third International Mathematics and Science Study (TIMSS) have recently been released.15 Some months ago a separate consortium of "advantaged" school districts from the Chicago North Shore was also given permission to participate in the TIMSS program, and a report of their achievement scores has
just reached our desks.16 The mathematics achievement scores they report are indeed (as we had estimated) above many European countries participating in TIMSS and significantly lower only than Singapore, the top-scoring nation in that study!
State Differences in Achievement
In short, good evidence has now appeared indicating that differences between public school districts in levels of funding and rates of child poverty generate large disparities in eighth-grade mathematics achievement. But what about differences between the states? Do equivalent effects appear when we compare states that are high and low in school funding and child poverty?
It is actually quite easy to explore this question. The 1996 NAEP data for mathematics achievement have just been released, and they include a breakdown of average eighth-grade achievement scores for public schools in 40 of the 50 states.17 In order to explore the effects of school funding and child poverty on these achievement data, then, we need only locate state-level data for funding and poverty. The former may be found, of course, in Education Week's recently released supplement, Quality Counts,18 whereas the latter also appear in the NAEP mathematics report.19
If one compares information from these three data sets, one discovers that state differences in school funding are correlated with mathematics achievement at r=+.433 (p<.01), whereas the child poverty/achievement correlation is a mammoth r=-.700 (p<.001)! Moreover, when one examines funding and poverty as joint predictors of achievement in a regression analysis, one learns that the net effects of both factors remain statistically significant, with ß=+.262 (p<.03) for school funding and ß=-.629 (p<.001) for child poverty, and that these two factors predict an astounding 55% of the variance of state differences in average achievement. In short, not only do differences in school funding and child poverty matter at the state level, they are major predictors of state-level averages in mathematics achievement. Indeed, the impact of child poverty seems to be stronger at the state level than at the district level.
I confess that I was startled by the large size of these effects when I first computed them. Since then, the opportunity to check their validity has arisen with the publication of the second and last NAEP data set to be released this year, this one concerned with science achievement in 1996. The report of these latter data also provides information about the average eighth-grade achievement scores for public schools from 40 of the 50 states,20 and again it is easy to study how these scores are predicted by state differences in school funding and child poverty. The results from this comparison largely parallel those for mathematics. This time the correlation obtained for school funding is r=+.389 (p<.02), while that for child poverty is r=-.696 (p<.001). Results for the regression analysis differ slightly, however. This time, although the net effect of school funding remains substantial, it drops slightly below the conventional .05 level of statistical significance (ß=+.215, p<.075), and the proportion of state-level achievement variance predicted by funding and poverty is "only" 53%. However, the net effect for child poverty remains huge (ß=-.628, p<.001). Thus, once again, one learns that these two factors are major predictors of state differences in achievement - this time, science achievement - and that the impact of child poverty is very strong at the state level.
This finding raises an interesting question. Why should child poverty be a stronger predictor of achievement at the state than at the district level? The answer seems to be that more experiences related to both child poverty and achievement are tapped in larger units of analysis. Child poverty has but one type of effect at the level of the individual student: it interferes with that person's achievement. At the level of the school and district, however, it can generate effects in two ways: by affecting the individual student and by creating a collective "environment of poverty" within the school or community. At the level of the state, child poverty is likely to be associated with additional environmental problems and missing social services that further diminish educational possibilities for the student. Thus states with especially high levels of child poverty handicap their students for education in many ways.
Implications
The effects I report here help us understand why setting higher standards will have so little impact on achievement. If many, many schools in America are poorly funded and must contend with high levels of child poverty, then their problems stem not from confusion or lack of will on the part of educators but rather from lack of badly needed resources. In fact, setting higher standards for those disadvantaged schools can even make things worse. If they are told that they now must meet higher standards, or - worse - if they are chastised because they cannot do so, then they will have been punished for events beyond their control. Thus arguments about higher standards are not just nonsensical; if adopted, the programs they advocate can lead to lower morale and reduced effectiveness among the many educators in the U.S. who must cope with poor school funding and extensive child poverty.
Isn't it time to reconstitute the terms of the debate? Surely our schools can be improved, but if this is to happen Americans must be willing to face up to the real problems of education. I submit that one way - perhaps the best way - to begin is by debating and (at long last) taking action to solve the serious problems of inequities in school funding and massive child poverty in our country. Solving these problems will require that additional resources be devoted to the needs of poor and middle-class persons, of course. But surely it is time that those persons - - the majority of Americans -- spoke up, demanded equity, and explained why a fairer and more effective system of education would benefit all Americans.
1. See David C. Berliner and Bruce J. Biddle, The Manufactured Crisis: Myths, Fraud, and the Attack on America's Public Schools (New York: Addison-Wesley-Longman, 1995).
2. Quality Counts: A Report Card on the Condition of Public Education in the 50 States, special supplement, Education Week, 22 January 1997, p. 9.
3. Ibid., p. 56.
4. Ibid., p. 57.
5. See, for example, Eric A. Hanushek, "The Impact of Differential Expenditures on School Performance," Educational Researcher, May 1989, pp. 45-65; and idem, "A More Complete Picture of School Resource Policies," Review of Educational Research, vol. 66, 1996, pp. 397-409.
6. See, for example, Larry V. Hedges, Richard D. Laine, and Rob Greenwald, "Does Money Matter? A Meta-Analysis of Studies of the Effects of Differential School Inputs on Student Outcomes," Educational Researcher, April 1994, pp. 5-14; and Rob Greenwald, Larry V. Hedges, and Richard D. Laine, "The Effect of School Resources on School Achievement," Review of Educational Research, vol. 66, 1996, pp. 361-96.
7. See, for example, Frederick D. Sebold, "School Funding and Student Achievement: An Empirical Analysis," Public Finance Quarterly, vol. 9, 1981, pp. 91-105; Robert C. Dolan and Robert M. Schmidt, "Assessing the Impact of Expenditure on Achievement: Some Methodological and Policy Considerations," Economics of Education Review, vol. 6, 1987, pp. 285-99; and Ronald F. Ferguson, "Paying for Public Education: New Evidence on How and Why Money Matters," Harvard Journal on Legislation, vol. 28, 1991, pp. 465-98.
8. See Gary Burtless, ed., Does Money Matter? The Effect of School Resources on School Achievement and Adult Success (Washington, D.C.: Brookings Institution Press, 1996); and Heritage Foundation, Education Update, vol. 12, no. 4, 1989.
9. Lee Rainwater and Timothy M. Smeeding, "Doing Poorly: The Real Income of American Children in a Comparative Perspective," Working Paper No. 127, Luxembourg Income Study, Maxwell School of Citizenship and Public Affairs, Syracuse University, Syracuse, N.Y., 1995.
10. Denny Braun, The Rich Get Richer: The Rise of Income Inequality in the United States and the World (Chicago: Nelson-Hall Publishers, 1997); and Edward N. Wolff, Top Heavy: The Increasing Inequality of Wealth in America and What Can Be Done About It (New York: New Press, 1995).
11. See Gary Natriello, Edward L. McDill, and Aaron M. Pallas, Schooling Disadvantaged Children: Racing Against Catastrophe (New York: Teachers College Press, 1990).
12. See, for example, Mary M. Kennedy, Robert K. Jung, and Martin E. Orland, Poverty, Achievement, and the Distribution of Compensatory Education Services (Washington, D.C.: Office of Educational Research and Improvement, U.S. Department of Education, 1986); and Sanders Korenman, Jane E. Miller, and J. E. Sjaastad, "Long-Term Poverty and Child Development in the United States: Results from the NLSY," Child and Youth Services Review, vol. 17, 1995, pp. 127-55.
13. See Kenneth J. Travers and Ian Westbury, The IEA Study of Mathematics I: Analysis of Mathematics Curricula (New York: Pergamon, 1989).
14. Kevin J. Payne and Bruce J. Biddle, "Poor School Funding, Child Poverty, and Mathematics Achievement," unpublished paper, 1997. (This paper has been submitted for publication. Requests for reprints should be directed to the second author.)
15. Albert E. Beaton et al., Science Achievement in the Middle School Years: IEA's Third International Mathematics and Science Study (Chestnut Hill, Mass.: TIMSS International Study Center, Boston College, 1996); and idem, Mathematics Achievement in the Middle School Years: IEA's Third International Mathematics and Science Study (Chestnut Hill, Mass.: TIMSS International Study Center, Boston College, 1996).
16. David J. Kroeze and Daniel P. Johnson, Achieving Excellence: A Report of Initial Findings (from the First in the World Consortium) for Eighth-Grade Performance from the Third International Mathematics and Science Study (Oak Brook, Ill.: North Central Regional Educational Laboratory, n.d.).
17. Clyde M. Reese et al., NAEP 1996 Mathematics Report Card for the Nation and the States (Washington, D.C.: National Center for Education Statistics, 1997), p. 129. Ten of the states apparently did not participate in the 1996 NAEP effort. This makes it slightly more difficult to sort out predictors for state differences in achievement. Data that those 10 states might have provided are missing, and only 40 cases are now left in the analysis, which means that even substantial effects may not generate statistical significance.
18. Quality Counts, p. 56.
19. Reese et al., p. 139.
20. Christine Y. O'Sullivan, Clyde M. Reese, and John Mazzeo, NAEP 1996 Science Report Card for the Nation and the States (Washington, D.C.: National Center for Education Statistics, 1997), p. 25.