Journal Issue: Opportunity in America Volume 16 Number 2 Fall 2006
Can Differences in School Quality Explain the Patterns?
Finally, we consider whether differences in school quality help explain why more privileged students complete more schooling than their less privileged counterparts. We begin by noting that the conventional measure of an individual's education—years of completed schooling—is rather limited. In particular, it ignores whether students with the same level of completed education may have received an education of different quality. By the conventional measure, completing one year of education should increase an individual's human capital by the same amount regardless of the school attended. But because one year at a poor school may increase human capital less than does one year at an excellent school, school quality could affect the value of education. It could also arguably affect the cost of education. A low-quality school, for example, may leave a student unprepared to master the skills of the next grade level, thus raising the costs in psychological terms (and also in time) of getting more education.
Does Family Background Affect the Quality of a Child's School?
In the United States, the school a child attends is largely determined by the neighborhood where he or she lives. To the extent that parental socioeconomic status affects the neighborhood where a child lives, it may thus also affect school quality. For example, less privileged parents certainly have fewer financial resources than more privileged ones. While many forms of financial aid are available to low-income students who want to attend college, no such credit is available to low-income parents who want to live in a high-quality school district. These borrowing constraints likely cause school quality to vary by family background. If poor school quality leads to lower educational attainment, then children of less privileged parents will have lower educational attainment than children of more privileged parents.
At the school or school district level, some potential indicators of school quality are clearly related to family background or income (which, in turn, is correlated with family socioeconomic status). An obvious first question is whether overall school spending differs from one district to the next by the average socioeconomic status of the residents of the district. Higher-income school districts, after all, have more money to spend on education, and in theory more money should buy higher school quality. Using data from the 2003 Common Core of Data, we calculate average per pupil spending in school districts with at least 70 percent of students eligible for free or reduced- price school lunch and districts with less than 20 percent of pupils eligible.
Not surprisingly, we find that average spending per pupil is rather similar. Districts with the larger share of disadvantaged children spend an average of $10,414 per pupil, as against $9,647 for districts with a smaller share of such children. The similarity in spending in part reflects school finance reforms since the 1970s that have tried to equalize school funding across poor and rich districts. But similar total spending per pupil does not necessarily reflect similar school quality, because different school districts may face different costs. Older school districts with aging buildings, for example, may have to spend more to maintain their facilities than newer suburban districts do. Some districts may have more special education students, who need smaller classes, which means hiring more teachers. And urban districts may face higher-wage labor markets than rural districts. Indeed, the recognition that some groups of students may need extra money to compensate for family disadvantage underlies the goal of closing achievement gaps between high- and low-performing children in Title I of the Elementary and Secondary Education Act of 1965 (of which the No Child Left Behind Act of 2001 is the most recent reauthorization.)
Given that instructional salaries and benefits make up more than 50 percent of schools' total current spending, class size could be another way in which school quality could vary by family background.32 Because data on class size are not readily available, we look at pupil-teacher ratios instead. We have also calculated the average pupil-teacher ratios for schools by family socioeconomic background. As with total school spending, the pupil-teacher ratios are quite similar: 16.9 for schools attended by children of disadvantaged family background, as against 17.4 for schools attended by more privileged children. 33 Does the lower ratio in schools serving poor children mean that the quality of schooling is better in those schools? Such an interpretation is not likely to be correct because those schools may have a larger share of special education or English-languagelearner students than schools serving more privileged children, which have fewer special education classrooms.34
One aspect of school quality that is less prone to distortion by compensatory education policies is teacher quality. Although a district may be able to raise salaries as an incentive to high-quality teachers, it cannot force such teachers to accept its job offers. One measure of teacher quality is teaching experience, and it is telling that schools serving poorer students are likely to have fewer experienced teachers. In this case, schools' socioeconomic status is defined by the percentage of students who are eligible for free or reducedprice school lunch. Eighty percent of teachers in low socioeconomic status schools (those in the top quartile by share eligible for free or reduced-price lunch) have more than three years of experience, compared with 89 percent of teachers in high socioeconomic status schools (those in the bottom quartile by share eligible).35
Hamilton Lankford, Susanna Loeb, and James Wyckoff look in more detail at differences in teacher quality by student characteristics for the state of New York. They find that poor students are more likely than nonpoor students to have a teacher who is not certified in any subject that he or she is teaching (21 percent versus 16 percent), who failed a certification exam on the first attempt (28 percent versus 20 percent), or who attended a college ranked “least competitive” by Barron's College Guide (25 percent versus 24 percent).36
Schools also vary in facility and peer quality. As figure 3 shows, low socioeconomic status schools (those with 70 percent or more children eligible for free or reduced-price school lunch) have worse facilities than high socioeconomic status schools (those with fewer than 20 percent of students eligible for free or reduced-price school lunch). Fifty-seven percent of low socioeconomic status schools have no temporary buildings, as against 65 percent of schools serving high socioeconomic status students. Similarly, 37 percent of schools serving poor children (low socioeconomic status schools) have fully adequate building features, compared with 55 percent of schools serving nonpoor children (high socioeconomic status schools).37
Peer quality as measured by college enrollment rates and Advanced Placement courses is also lower for less privileged children. Data from the NELS show that low socioeconomic students (those with parents in the bottom quartile by socioeconomic status) attend schools in which only 56 percent of students go on to some college, as against 75 percent of students in schools serving high socioeconomic status students (those in the top quartile by socioeconomic status). The share of students taking Advanced Placement courses is 16.9 percent in schools attended by students with low socioeconomic status, compared with 26.2 percent for schools attended by high socioeconomic status students. In short, the peers of less privileged students are not as academically oriented as the peers of wealthier students.
Finally, we have found some evidence that school districts that are low in socioeconomic status may not spend resources as efficiently as districts with higher socioeconomic status, suggesting that they may be more poorly managed.38 This finding, in combination with the descriptive data above (in figure 3), leads us to conclude that school quality varies according to parental socioeconomic status.
Does School Quality Affect Children's Educational Attainment?
The next question is whether these differences in school quality translate into worse outcomes for less privileged children. By the early 1990s, many people were convinced that once one took account of differences in family background, school resources—including money—did not matter for student achievement. In a 1996 article economist Eric Hanushek wrote, “Three decades of intensive research leave a clear picture that school resource variations are not closely related to variations in student outcomes and, by implication, that aggressive spending programs are unlikely to be good investment programs unless coupled with other fundamental reforms.”39
Although Hanushek's analyses of the effects of school resources on student achievement have been very influential, other researchers have criticized his findings on methodological grounds.40 For example, one independent analysis of one of Hanushek's studies concludes that the effect of per pupil spending on student achievement is large and educationally significant.41 More recent studies that make explicit attempts to account for the compensatory nature of much educational expenditure also provide evidence that money matters. One of our own studies finds that the market values school spending in terms of property values. And Jonathan Guryan finds that a $1,000 increase in per pupil spending in Massachusetts increases average test scores for fourth- and eighthgrade students by one-third to one-half of a standard deviation. Importantly, in summarizing the findings of seventeen federal studies, Geoffrey Borman and Jerome D'Agostino conclude that Title I of the Elementary and Secondary Education Act, which aims to provide additional funding to schools and districts serving disadvantaged students, has indeed improved the educational outcomes of children it has served. Further, in studying the effect of state efforts to equalize funding between wealthier and poorer school districts, David Card and A. Abigail Payne find that such reforms have narrowed gaps in spending as well as in educational outcomes.42
Whether money matters must depend in part on how the money is spent. Probably the best evidence to date on the effect of class size comes from the Tennessee Student-Teacher Achievement Ratio experiment (known as Project STAR), the nation's largest randomized experiment aimed at understanding how smaller class sizes affect student achievement. 43 In the 1985–86 school year some 6,000 kindergarten students in Tennessee were randomly assigned to one of three groups: small classes (13–17 students per teacher), regular-sized classes (22–25 students), and regular-sized classes with a teacher's aide. The experiment, ultimately involving some 11,600 students, lasted four years. After the third grade, all students returned to regular-sized classes.44 The data have been analyzed by a variety of researchers, with a remarkably consistent finding: smaller classes result in higher student achievement.45 One study finds that the class-size effects are larger for students eligible for free or reduced-price school lunch than for more well-to-do students. Another reports that the students who were (randomly) placed in smaller classes in grades K–3 performed better on standardized tests when they reached the eighth grade. They were also more likely to take a college entrance exam (such as the ACT or SAT)—a signal that they may have been more likely to attend college as well.46
Yet another study, by David Card and Alan Krueger, relating the quality of schooling received by people born between 1920 and 1949 to their earnings in 1979 found that a reduction in the pupil-teacher ratio of 10 students increased average earnings by 4.2 percent. 47 Other studies reviewed by these same authors in a later study find that reductions in pupil-teacher ratios are associated with increased average earnings, although several of the estimates are not statistically significant.48
Economic studies also broadly agree that teacher quality matters, though they agree much less about what makes a high-quality teacher.49 Developing credible studies of the effects of particular teacher characteristics on student achievement is extremely difficult. Because teachers are not randomly assigned to schools, studies find ostensibly “better” teachers at schools attended by more advantaged students. Thus, as in other areas, the researchers can develop links between certain teacher characteristics and student outcomes but cannot be assured that the teacher characteristics caused the change in student outcomes. In addition, such studies typically rely on administrative data that do not contain many of the characteristics that likely make a good teacher, such as classroom management, motivation, professionalism, and a thorough understanding of how to communicate new concepts to students. That said, some studies have found that teachers improve greatly after one or two years of experience. If that finding is accurate, the fact that schools serving poorer students have more teachers with very little experience suggests that these students will have lower achievement as a result.
Does Improving School Accountability Improve Student Performance?
Given already high levels of educational spending, policymakers are looking for ways to provide incentives for schools to improve without large increases in revenues. “School accountability” programs come in two forms.50 Institutional school accountability programs, such as the No Child Left Behind Act of 2001, set up a system of rewards and sanctions determined by school performance— typically, student performance on standardized tests. Significantly, No Child Left Behind makes each school's performance public. These reforms are popular because they are relatively inexpensive and because they aim to make school systems more transparent, so that parents can more readily compare their child's school with others. Although research on the effects of school accountability on student achievement is growing, it is still fledgling.51 At best, such programs generate small improvements in student achievement. At the same time, researchers have documented several unintended consequences. For example, one study estimates that teachers cheat in 4–5 percent of elementary school classrooms each year in Chicago and suggests that cheating increases when teachers have an incentive to do so, as they have with high-stakes tests.52 Other researchers find that administrators reclassify low-achieving students as learning disabled so that the (presumably low) scores of these students will not be included in the school's average test score calculation.53 David Figlio reports that schools are more likely to suspend students during the testing cycle, apparently to alter the composition of the testing pool. Brian Jacob finds some evidence that teachers focused more on the high-stakes test material than on the lowstakes test material following the introduction of Chicago's school accountability system.54
Another potential form of accountability is through the market. Because students are assigned to schools based on their neighborhood, many observers have argued that local public schools are not required to be accountable to local citizens. Thus, if parents could “vote with their feet,” competitive pressure and the threat of losing students would force such schools to improve. Two often talked-about forms of competitive pressure are charter schools—public schools that are exempt from many of the regulations that apply to traditional public schools—and school vouchers for use at private schools. Both forms of competition would give parents alternatives to the local public school, thus presumably improving both the educational achievement of their children and the quality of the local public schools. Importantly, because the accountability is enforced by parental choices rather than the rules of a system, there is less scope for the unintended consequences noted above.
Although these arguments are theoretically persuasive, there is little empirical evidence that either charter schools or school vouchers improve student test scores (which should, in turn, improve educational attainment). For example, three sets of researchers, using statewide data from North Carolina, Florida, and Texas, respectively, have studied whether students who attend charter schools have higher test score gains than students in local public schools.55 Their findings are remarkably similar: there are no achievement gains for students who attend charter schools, even after controlling for a rich set of student characteristics. In fact, the students in charter schools appear to perform worse, perhaps because these are often new schools.
Evidence on school vouchers is also decidedly mixed. The best-designed study of school vouchers was conducted by William Howell and Paul Peterson in New York City, beginning in 1997.56 It randomly assigned 1,300 students to two groups. One group received a (privately funded) scholarship to attend a private school; the other, control, group did not. After three years, the study found that overall there were no test score gains among the students who were offered a voucher or among the students who actually took advantage of the voucher offer and attended private schools. Howell and Peterson reported educationally large and statistically meaningful gains among African American students, but their findings have been disputed in a reanalysis of the data by Krueger and Pei Zhu.57
Evidence from publicly funded voucher programs in Milwaukee and Cleveland does not help to clarify the issue. One study of Milwaukee's Parental Choice Program, the oldest publicly funded choice program in the United States, suggests that students gained in math but not in reading; another suggests no gains in either math or reading.58 The most recent evidence from the Cleveland Scholarship and Tutoring Program suggests that vouchers have not significantly benefited the recipient students.59 After five years the test scores of voucher students are generally quite similar to those of a group of students who applied for, but did not receive, a voucher.
Importantly, all these studies examine small-scale programs. None addresses the question of whether a large-scale program would generate enough competitive pressure on the public schools to induce them to improve. Evidence from Florida's school accountability system (which includes a school voucher for students attending persistently “failing” schools) suggests that even the threat of losing students through vouchers may not be a prime motivator for school improvement.60 Although schools faced with the possibility of becoming voucher-eligible appear to improve slightly, such improvement appears to spring from avoidance of the stigma of being labeled a failing school rather than the threat of vouchers per se.
Although these studies are not likely to be the last word on the effectiveness of institutional school accountability systems, charter schools, or school vouchers, together they indicate that the gains from improving school accountability are likely modest, at best.