Changes in School Enrollment Patterns After the First-Time Release of School-Level Test Scores in Milwaukee’s School Voucher Program

Abstract

Since 1990, the Milwaukee Parental Choice Program (MPCP) has enabled low and middle-income parents to attend private Milwaukee schools at state expense. In 2010, the Wisconsin Department of Public Instruction released, for the first time, results of newly required state standardized tests for students using the MPCP. This article uses an original data set from Milwaukee’s two-decade-old private school voucher program to test several hypotheses on the impact of the first public release of test scores on school enrollment patterns. Specifically, the article uses quantitative methods to test (a) whether enrollment decreased at lower performing schools, (b) whether low-performing schools were more likely to close compared with higher performing schools, and (c) whether historical school-level growth patterns were significantly different after the release of test scores. The article finds that enrollment increases at higher performing schools were larger than at low-performing schools. However, the link between test scores and enrollment patterns disappears after controlling for available school-level characteristics. The analysis provides a first peek at how a transparency intervention affects enrollment patterns in an education marketplace.

Keywords

accountability educational policy choice school choice

Introduction and Research Question

The privatization of K-12 education in the City of Milwaukee has evolved from a small experiment allowing a limited number of low-income pupils to attend a limited number of nonsectarian private schools to a program serving 21% of the city’s student population. No longer is the Milwaukee Parental Choice Program (MPCP) limited to low-income pupils or nonsectarian schools. Today, almost 25,000 Milwaukee students from families earning up to 300% of the federal poverty level are attending more than 100 mostly religious private schools at public expense.

The steady growth of the MPCP, in terms of raw numbers and market share, begs the simple question, “Have private school vouchers improved the academic performance of Milwaukee pupils?” The true answer to this question is the subject of ongoing debate. Data from the Trial Urban Assessment of the National Assessment of Education Progress (Urban NAEP) show that students in the City of Milwaukee, 20-plus years after the introduction of vouchers, continue to fare poorly compared with other large cities (National Center for Education Statistics, 2011a). Though the Urban NAEP excludes private school students, the inclusion of the scores of MPCP pupils likely would not improve Milwaukee’s performance. Aggregate test scores in math and reading on the Wisconsin Knowledge and Concepts Exam (WKCE) are generally lower for MPCP pupils than students using the Milwaukee Public Schools (MPS).

However, the performance of Milwaukee students on the Urban NAEP and the WKCE alone does not determine the success or failure of voucher policy in Milwaukee. The answer to the more specific question of whether schools in the MPCP are having a greater impact on student achievement than schools in MPS requires a valid comparison group. A 5-year state-authorized longitudinal study of education in Milwaukee assembled such a group and found that schools in the MPCP had a slightly positive impact on reading scores, but no impact on math scores for similar students over the course of 5 years (Witte, Carlson, Cowen, Fleming, & Wolf, 2012). Similarly, graduation rates were slightly higher for MPCP students, though some scholars have questioned the substantive significance of that finding due to the 56% attrition rate of voucher users in the study (Cowen, Fleming, Witte, Wolf, & Kisida, 2013). Overall, there is no negative impact on voucher users in Milwaukee, but academic gains for voucher users in Milwaukee are substantively small.

Yet the program remains popular, and is growing (Howell, 2012). Why? First, there is wide variation in performance within both the traditional public and private voucher school sectors in Milwaukee (Jacob & Wolf, 2012). It is possible that parents are not making choices between different systems but rather are making choices between schools. Research from Carlson, Cowen, and Fleming (2013) revealing differences between the quality of school selections of low- and high-performing students who leave the MPCP for MPS supports this possibility. If this is the case, the dominant manner in which the efficacy of the MPCP has been tested, aggregate level program performance, is flawed. Second, scholars have bemoaned imperfections in the Milwaukee school marketplace, including, a lack of adequate performance information about participating schools (Van Dunk & Dickman, 2003).

Indeed, for the first two decades of the MPCP, schools were not required to administer the Wisconsin state standardized assessment. Parents were obviously still making schooling decisions during this time but were presumably more reliant on emotional responses to marketing materials in the framework detailed by Lubienski (2007) than objective performance data. Lubienski (2007) concludes in his article on school advertising efforts that “competition alone does not always generate the incentives necessary for a rational consumer model to operate effectively” (p. 135). The response to the dearth of incentives is often government intervention. Such an intervention occurred in the MPCP in 2010 when the Wisconsin legislature, in a budget bill, required that students in private schools using vouchers take, and schools release the results of, the WKCE. For the first time in history, parents could compare the test scores of individual private schools in the MPCP with individual public schools. Though these data are inherently limited snapshot data that cannot show the impact individual schools have on student performance, their release still represents an increase in consumer information accessible to parents navigating Milwaukee’s education marketplace.

Under basic market theory, the release of publicly available performance information should lead to increased enrollment at high-performing schools as consumers flock to quality (Friedman, 1955). However, there are significant limitations that make the following analysis less than a pure test of market theory, including

Insufficient firsthand information regarding how Milwaukee parents make schooling decisions and

A limited amount of test results due to the newness of testing requirements.

A perfect test of the role of parental school decision making would include substantial longitudinal data from parental surveys matched with test score growth data for families with similar characteristics. But despite the limitations, a first look using available school-level data can provide useful information that will hopefully reveal the need for closer study of the link between information and parental schooling choices. This article uses data describing certain characteristics of schools participating in the MPCP to quantitatively test several hypotheses related to the research question, “Did enrollment patterns in MPCP schools significantly change after the first-time release of school-level test scores?”

Research Overview

The theory behind school vouchers as a policy strategy for improving education was first proposed by economist Milton Friedman in 1955 (Friedman, 1955, 1962). Friedman argued that education can function as a market commodity and that a system of school vouchers would enable parents to flock to high-performing schools, forcing low-performing schools to either improve or close. Chubb and Moe (1988) expanded on Friedman’s theory in an article that served as the intellectual foundation behind the country’s first urban school voucher program. That program, the MPCP, began in Milwaukee in 1990 at the behest of Wisconsin Republican Governor Tommy Thompson and a Milwaukee coalition of voucher supporters including Democratic State Representative Annette “Polly” Williams, civil rights activist Dr. Howard Fuller, and education consultants Susan and George Mitchell (Dougherty, 2004; Witte, 2000).

The MPCP was the subject of scholarly interest from its inception. Included in the legislation creating the original MPCP, which was capped at 1,000 students and limited to nonsectarian schools, was a provision calling for an official state evaluation. That evaluation, led by University of Wisconsin Political Scientist John Witte, found no clear difference in test scores between voucher users and comparable public school pupils (Witte, Sterr, & Thorn, 1995). However, subsequent research by Greene, Peterson, and Du. (1999) and Rouse (1998), found math and reading and math gains, respectively, for students using the MPCP. More recently, a 5-year longitudinal study conducted by the School Choice Demonstration Project at the University of Arkansas, and authorized in Wisconsin state statute, found that similar students using the MPCP and MPS tended to have similar changes in test scores over a 5-year period (Witte et al., 2012). While MPCP users did have a slight advantage in reading achievement, the advantage was substantively small, and possibly attributable to factors outside the MPCP (Witte et al., 2012).

Several studies have identified positive competitive effects from the MPCP. Both Hoxby (2003) and Chakrabarti (2008) found that public schools most exposed to competition from private schools accepting vouchers experienced achievement gains. However, Carnoy, Adamson, Chudgar, Luschei, & Witte, 2007), using a similar methodology as Hoxby, concluded that public school test score gains were caused by the one-time threat of competition from the voucher program’s religious school expansion in 1998 and that the competitive effects plateaued over time. Most recently, Greene and Marsh (2009) found a citywide positive, though modest, effect on public school test scores with the opening of each new private school accepting vouchers.

After two decades of existence, it is evident that participation in the Milwaukee voucher program does not automatically result in academic gains for program users. It is also clear that the competitive effects of vouchers in Milwaukee are real, but small. Perhaps the real lesson from the body of MPCP research is that the role of the program on education is far more nuanced and complicated than Friedman’s market theory suggested. Thus, understanding the actual interaction between this large-scale exercise in privatization and Milwaukee parents is crucial to evaluating its success as an education delivery strategy.

Today, almost 25,000 students are using the MPCP to attend more than 100 private schools in Milwaukee and its suburbs (Kava, 2013). The program is means-tested; only students from households with incomes at or below 300% of the federal poverty level may participate (Kava, 2013). Participating schools may be located anywhere in the state and must meet a variety of regulations, including demonstrating fiscal viability prior to opening, conducting an annual independent audit, employing teachers with college degrees, making an array of schools’ policies available to parents and the public, taking the same standardized tests as public schools, and obtaining accreditation from an approved agency (Kava, 2013).

Parents seeking to use the program apply directly to the school they wish their child to attend during one of the school-specified open enrollment periods from the 1st to 20th of each month. The earliest that schools may hold an open enrollment period for the following school year is from February 1st to 20th, meaning active parents are likely evaluating schools for the next academic year up to 8 months prior to the start of classes. Schools may not screen students for academic aptitude, past behavior, special needs status, or any other characteristics aside from residency in the City of Milwaukee and income. If schools receive more applicants at any grade level than they have seats, a random lottery is conducted. Parents are notified of acceptance at the end of each enrollment period. As designed, parents have substantial power and responsibility in placing their children at participating schools, leading to the important question, “How do parents make their schooling choices?”

Research both in Milwaukee and elsewhere provides insight into how parents, particularly low-income parents, say they go about making schooling choices for their children. According to survey research conducted by the Center on Reinventing Public Education, low-income Milwaukee parents most often cite academic quality as the reason for choosing their child’s school (Teske, Fitzpatrick, & Kaplan, 2007). However, significant numbers of parents also say they choose on safety and religion (Teske et al., 2007). Teske et al. (2007) also find that test scores are not the primary way by which parents gauge the academic quality of a school. Witte, Wolf, Cowen, Fleming, and Lucas-McLean (2008) survey parents of public and voucher students in Milwaukee, and find academic and teacher quality and safety the most valued school characteristics. In addition, parents on whole indicate they feel well informed to make school decisions (Witte et al., 2008). Somewhat surprisingly, given the overall low level of achievement in Milwaukee schools, Witte et al. (2008) find Milwaukee parents generally feel satisfied with their school choices.

A more recent study in Detroit, the Detroit School Shoppers Survey authored by Patrick Wolf and Stewart (2012), explores how Detroit families select schools for their children. Wolf and Stewart conclude that mothers are more likely to shop for schools than fathers, that word of mouth is the most popular way to find information about a school, and that parents prize academic quality, school safety, and location convenience above other factors when choosing a school. Wolf and Stewart also identified great variation in the ways in which parents shop for schools.

Hypotheses and Theoretical Justifications

This article tests four hypotheses all informed by the market logic of school voucher theory. In other words, the hypotheses are rooted in the belief that when given the ability to choose, parents will flock to high-quality schools and flee low-quality schools. Notably absent in this analysis is firsthand information from parents. Because of this, the school, rather than the parent, is the relevant unit of analysis. Although this makes definitive conclusions on the direct relationship between the release of test score data and parental decision making in the Milwaukee voucher program impossible, it does enable a first look for evidence of the existence and nature of the test score–parental decision making relationship.

Hypothesis 1: Enrollment in lower performing schools decreased after test scores became public.

Under the theory of school vouchers originally articulated by economist Milton Friedman, parents will act as rational consumers and avoid lower performing schools. The release of test scores substantially increased the availability and uniformity of performance data for students in MPCP schools, and if school consumers did in fact react to the release of information, a shift away from the lowest performing schools should have occurred. This hypothesis is tested by measuring the relationship between changes in test scores and year-to-year enrollment changes in Milwaukee’s voucher schools.

Hypothesis 2: Historic school-level growth patterns differed significantly after the release of test scores.

Another way of measuring whether parents reacted to the release of test score information is through a comparison of historic enrollment growth patterns at individual schools and their enrollment trends after the first release of test scores. The logic of Hypothesis 2 is similar to Hypothesis 1, only, with the advantage of considering historical enrollment trends. Considering enrollment trends over time addresses a potential weakness of Hypothesis 1: Mistakenly connecting a 1-year enrollment change caused by something other than the release of school performance data with the release of school performance data. If both Hypotheses 1 and 2 are found to have merit, the case for a relationship between the test score release and enrollment changes will be strengthened.

Hypothesis 3: Low-performing schools were more likely to close than higher performing schools.

This hypothesis too is rooted in Friedman’s original school choice theory. If public payments to schools are directed by parental decision making, and parents are rational actors seeking out high-performing schools, poor performing schools will lose customers, funding, and eventually be forced to close their doors.

Data

As previously mentioned, the relevant unit of analysis for this article, given its focus on school enrollment patterns, is the school. Using the school as the unit of analysis has inherent strengths and weaknesses. The school is the most localized unit for which there is test score performance data available to parents. Though data disaggregated by classroom and teacher may be more useful in a perfect school marketplace due to the possibility of variation in employee performance within schools, such data are not publicly reported and cannot be expected to be part of the parental decision-making process in Milwaukee. Schools are also the organization that will either thrive or whither within a marketplace. A classroom does not go out of business if performance is lagging, but a school in a marketplace can. Similarly, parents likely do not choose a school because of a single teacher or classroom when the student will inevitably be exposed to many different teachers during his or her academic career. Most importantly, this article is not an evaluation of the efficacy of school voucher policy in general; rather, it is a first look at enrollment changes. Student-level data may be useful in future research, if this analysis reveals evidence of changing enrollment patterns.

Each observation in the data set is for a private school that participated in the Milwaukee voucher program in at least 2 years: The year before the first release of test score information (2010-2011), and the year after the release of school-level test score information (2011-2012). The data set for this article was created by the author between September 2012 and February 2013 using publicly available information. All of the statistics in Table 1 are from the Wisconsin Department of Public Instruction, while the religious affiliation data in Figure 1 originate from an analysis of public information compiled by the author for a previous study (Ford, 2011).

Table 1.

Summary Statistics.

Variable	n	M	SD	Minimum	Maximum
Enrollment change after 1 year of testing	90	22.59	37.12	−94	216
Percentage enrollment change after 1 year of testing	90	0.16	0.31	−.38	1.93
Number of tested students	97	109.87	118.65	1	829
Years in program	95	8.74	3.92	1	18
Average percent voucher enrollment	95	0.67	0.29	0.03	1
Percentage of students scoring advanced/proficient on reading WKCE	93	0.53	0.24	0	0.93
Percentage of students scoring advanced/proficient on math WKCE	93	0.34	0.21	0	0.91

Note. WKCE = Wisconsin Knowledge and Concepts Exam.

Figure 1.

Milwaukee Parental Choice Program schools in sample by religious affiliation.

The enrollment change after 1 year of testing variable is the difference between school enrollment in the 2010-2011 and 2011-2012 school years. The percentage enrollment change after 1 year of testing variable is the percentage difference between school enrollment in the 2010-2011 and 2011-2012 school years. Number of tested students is the number of students who took the WKCE in 2010-2011. Years in program is a measure of school stability; it is the number of years a private school has participated in the voucher program.

Average percent voucher enrollment is a school-level socioeconomic variable. During the years of this analysis, the MPCP was limited to students from families with incomes at or below 175% of the federal poverty level. Because voucher schools are private schools, they may also enroll private-pay students who presumably do not qualify for the voucher. Accordingly, the percentage of voucher users gives a general idea of the percentage of low-income pupils in a school.

The final two variables are measures of school-level performance on the WKCE. In Wisconsin, the combined percentage of students in a school scoring advanced and proficient on the WKCE is often used as an indicator of school quality. The combined percentage of students scoring advanced and proficient on the WKCE has also been used in high-profile comparisons of vouchers and public schools in Milwaukee. Though no standardized test is perfect, WKCE results serve as an intuitive and accepted indicator of school quality in Wisconsin (Richards, 2011).

Hypotheses Testing

The previously described hypotheses are tested in this section through a combination of means—comparison tests and several ordinary least squares (OLS) regression models. Although these methods are somewhat crude out of necessity due to the limited availability of school-level MPCP data, they can still yield meaningful information. More sophisticated techniques could be used if data on student performance history, special needs status, English language learner status, racial makeup, and teacher training information were available by school. However, the MPCP is made up of private schools, which, despite their acceptance of public money, are not required to release this information. If and when such information is made available, more advanced methods can be used.

Hypothesis 1: Enrollment in lower performing schools decreased after test scores became public.

The simplest way to test this hypothesis is by comparing changes in enrollment between the 2010-2011 and 2011-2012 school years (the years before and after the release of WKCE scores) in MPCP schools with below- and above-average test scores. Average is defined as the average percentage of pupils scoring advanced or proficient on the WKCE in all MPCP schools. As can be seen in Table 2, enrollment in schools with above-average levels of reading and math proficiency grew by a significantly larger amount than schools with below-average levels of reading and math proficiency. At the very least, this suggests more parents are enrolling their children at higher performing MPCP schools compared with the previous year. However, the bottom half of schools are still experiencing some enrollment growth.

Table 2.

Enrollment Growth (2011-2012) in Schools With Below- and Above-Average Achievement.

	Mean percentage enrollment growth
Schools with below-average WKCE proficiency in reading	13.6
Schools with above-average WKCE proficiency reading	29^a
Schools with below-average WKCE proficiency in math	14.72
Schools with above-average WKCE proficiency math	33.22^a

Note. WKCE = Wisconsin Knowledge and Concepts Exam.

Significantly larger at the 95% confidence level.

What about the schools that are actually losing enrollment? Between 2010-2011 and 2011-2012, 15 schools participating in the voucher program experienced an enrollment loss and 76% experienced growth. Table 3 shows that the schools gaining students had significantly higher percentages of students scoring at least proficient on state math and reading tests. In other words, the schools losing market share are on average lower performing. Although this limited evidence supports Friedman’s original school choice theory, key questions remain unanswered. Are the enrollment shifts identified actually driven by the release of test score information? Or, are other factors such as the religious affiliation of individual schools responsible for changing enrollment patterns?

Table 3.

Mean of Students Scoring Advanced/Proficient on the WKCE in Schools With Enrollment Growth and Decline.

	Schools with enrollment decline (%)	Schools with enrollment growth (%)
Reading	40	55^a
Math	22	36^a

Note. WKCE = Wisconsin Knowledge and Concepts Exam.

Significantly larger at the 95% confidence level.

Table 4 presents the results of four OLS regression models using school-level enrollment changes (Models 1 and 2) and school-level percentage enrollment changes (Models 3 and 4) between the 2010-2011 and 2011-2012 school years as the dependent variables, and the percentage of students in a school scoring advanced and proficient on state math and reading tests as the respective independent variables. The models are not particularly strong, Models 1 and 2 have an R² statistic of about .123, and Models 3 and 4 of about .140, meaning much of the variation in the dependent variable is unexplained by the model. Likely much of the unexplained variation in the model is caused by the absence of school-level student demographic information and, given the importance placed on teacher quality and safety by Milwaukee parents, school-level teaching quality and safety data. However, control variables for religious affiliation (using nonsectarian as the reference variable), the number of tested students, the number of years a school was in the voucher program, and the average percentage of voucher users in the schools are included in the models. Neither independent variable is statistically significant in any of the models, suggesting that even if parents are fleeing low-performing schools in favor of higher performers, the available data after 1 year of testing does not indicate that it is school-level math and reading proficiency levels driving the migration.

Table 4.

OLS Regression Results: Models 1 to 4.

	Model 1	Model 2	Model 3	Model 4
Variables	Enrollment change after 1 year of testing	Enrollment change after 1 year of testing	Percentage enrollment change after 1 year of testing	Percentage enrollment change after 1 year of testing
Advanced/proficient reading	21.70		0.0478
	(19.31)		(0.121)
Advanced/proficient math		24.66		−0.00272
		(21.19)		(0.133)
Tested students	0.0768*	0.0783*	−0.000207	−0.000197
Tested students	(0.0385)	(0.0384)	(0.000241)	(0.000241)
Years in program	−0.950	−0.891	−0.0182*	−0.0177*
Years in program	(1.190)	(1.182)	(0.00744)	(0.00741)
Average percent voucher	2.794	3.970	−0.128	−0.135
Average percent voucher	(18.75)	(18.89)	(0.117)	(0.118)
Catholic	14.27	15.93	0.142	0.149
Catholic	(15.99)	(15.80)	(0.100)	(0.0990)
Lutheran	12.36	13.71	0.0788	0.0868
Lutheran	(15.17)	(14.95)	(0.0949)	(0.0936)
Other Christian	0.160	1.967	0.0741	0.0756
Other Christian	(14.63)	(14.63)	(0.0915)	(0.0917)
Muslim or Jewish	8.135	8.395	0.132	0.140
Muslim or Jewish	(25.42)	(25.36)	(0.159)	(0.159)
Constant	0.633	0.917	0.308*	0.329*
Constant	(22.82)	(22.63)	(0.143)	(0.142)
Observations	87	87	87	87
R ²	.123	.124	.140	.139

Note. Standard errors in parentheses. OLS = ordinary least squares.

p < .05.

Hypothesis 2: Historic school-level growth patterns differed significantly after the release of test scores.

Though understanding the enrollment changes in private schools accepting vouchers in the first year after the release of standardized test scores is instructive, it is more informative to compare the growth of private schools in the year after they released their WKCE scores for the first time with their historic annual growth rates. Is there evidence that enrollment patterns shifted from their norms after test results were released?

In Table 5, the schools in the Milwaukee voucher program are again divided among those with a below and above program average percentage of students scoring advanced or proficient in math and reading (the average percentage of students scoring advanced and proficient on the 2010-2011 WKCE was 34% in math and 53% in reading). Though enrollment declines in below-average schools did not significantly change from historic program trends, increases in enrollment at above-average schools were statistically larger than historic program trends. Hence, there is partial support for this hypothesis; enrollment increases accelerated from historic norms in above-average MPCP schools after the release of test scores.

Table 5.

School Enrollment Growth Comparison Between Schools With Below- and Above-Average Percentage of Students Scoring Proficient on the WKCE.

	Historic average annual growth	Growth the year after test score release
Below-average WKCE proficiency in reading	19.17	14.21
Above-average WKCE proficiency reading	15.76	29^a
Below-average WKCE proficiency in math	20.82	14.1
Above-average WKCE proficiency math	13.14	32.29^a

Note. WKCE = Wisconsin Knowledge and Concepts Exam.

Significantly larger at the 95% confidence level.

Hypothesis 3: Low-performing schools were more likely to close than higher performing schools.

Since the 2010-2011 school year, eight of the schools included in the original release of test score data have left the MPCP. As Table 6 shows, those eight schools had levels of reading proficiency below those schools that are still in the MPCP, and levels of math proficiency lower, but statistically similar, to schools still in the MPCP in 2013. These data suggest lower performing schools, at least in reading, were more likely to close than higher performing schools. However, the small number of schools closed demands that this information be taken with a grain of salt. It does, however support the findings of an earlier analysis of closed schools conducted by the School Choice Demonstration Project (Kisida, Jensen, & Wolf, 2011).

Table 6.

Mean Percentage of Students Scoring Advanced/Proficient on the WKCE Still in the MPCP and Schools That Left the MPCP.

	Left MPCP (%)	Still in MPCP (%)
Reading	36	54^a
Math	25	34

Note. WKCE = Wisconsin Knowledge and Concepts Exam; MPCP = Milwaukee Parental Choice Program.

Significantly larger at the 95% confidence level.

Results and Discussion

Overall, there is some evidence suggesting that Milwaukee parents are migrating in greater numbers to higher performing voucher schools, that low-performing voucher schools are attracting comparatively fewer new parents, and that low-performing voucher schools are closing at a higher rate than higher performing voucher schools in the year after the first-time release of MPCP performance data. However, low-performing schools are still attracting parents, and there is no evidence that the first-time release of WKCE tests scores is the reason more parents are enrolling their children in above-average schools, or in particular fleeing low-performing schools. It is entirely plausible that enrollment shifts are a continuation of some prior trend or other unidentified cause.

There are many points to consider to put these results in their proper context. First, the schools in the MPCP do not operate in a vacuum. In addition to each other, they compete for students with the Milwaukee Public Schools, independent charter schools, and suburban public schools that enroll more than 6,000 Milwaukee students through Wisconsin’s public school choice programs. It is possible that any enrollment shifts in specific private schools were influenced by changing conditions in the various other options available to parents. There is substantial school switching in Milwaukee across school types, making it likely that the effect of the first-time release in voucher test scores could have been influenced by changes in other parts of Milwaukee’s school marketplace. Though there is nothing unique in the enrollment patterns in other Milwaukee school types in 2010-2011 to suspect the influence of different school types to be any different in years past, their influence cannot be discounted.

Second, it is possible that some Milwaukee parents evaluate schools on a strictly case-by-case basis, rather than through comparisons between schools. Certainly research from Teske et al. (2007) and the School Choice Demonstration Project’s state-authorized evaluation shows that parents consider a number of factors when choosing a school in Milwaukee. Rather than compare test scores between schools, parents might use test scores to supplement other information available about the specific school they are considering. Consider that well before the release of WKCE scores for voucher schools, more than 90% of private schools participating in the MPCP reported that they administered standardized tests (Dickman, 2004). Though the mandated public release of school-level test scores was a significant change in public policy, parents may have become accustomed to receiving standardized test scores from the schools themselves, and ignored the publishing of side-by-side comparisons.

Third, the use of WKCE test scores is an incomplete indicator of the quality of a school. If a school’s student body is primarily made up of students who entered well-below grade level, it is possible that a school looks low performing when the WKCE is really only showing the quality of students the school receives. A school with low average test scores may in fact be adding more value to students’ scores than a school with high average test scores. Some parents who appear to be choosing low-performing schools may in fact be choosing schools that add significant value to students who enter the school with low test scores.

Fourth, some parents may (and likely) do not care about standardized test scores when making schooling decisions. Parents may make choices based on school safety, location, or reputation. Or, some parents may not be savvy consumers and do not understand, or care about, standardized test score comparisons. Or, some parents may be overwhelmed by the large number of potential school choices in Milwaukee and make a decision that on the surface may appear irrational. In all of these cases, some parents would be making a schooling decision completely detached from standardized test score information.

Fifth, it may take more than 1 year of performance data to alter the behavior of consumers in a school marketplace. A parent may not be moved to act after 1 year of poor performance but may change schools if low-performance becomes an annual trend. Probably, the full impact of the voucher policy change in Milwaukee will take several years to emerge.

Last and most important, the relatively poor performance of Milwaukee schools, public, charter, and voucher, may force parents to make schooling choices that have nothing to do with performance. If no quality school option is available, a parent may choose for convenience, safety, or any other number of reasons.

None of these points undermine the presented results; there were clearly changes in school-level enrollment after the first-time release of test scores in Milwaukee’s voucher program. But the points, as well as the presented results, highlight a continuing need to fill knowledge gaps with regard to the way parents make schooling choices in Milwaukee and elsewhere. Future research answering some of these unknowns with firsthand information from parents would strengthen the understanding of the likely impact of any market-based privatization effort.

Conclusion and Lessons

Education is a public good guaranteed to every child in state constitutions across the United States. If public budgets are a sign of priorities, it is also the most important thing that states do. The privatizing of this public good in Wisconsin’s largest school district provides many lessons for other market-based privatization efforts across the country.

The largest takeaway from the Milwaukee voucher experience identified in this research is that school enrollment patterns did change after the public release of student performance data. Though this first look at 1 year of data precludes definitive conclusions, there is good reason to think that school consumer behavior was affected by the increased availability of test score data.

Most important, there are reasons to think that consumer behavior deviated in ways that promoted the positive public policy goal of maximizing enrollment in relatively higher performing schools.

However, the Milwaukee experience also provides reason to suspect that the release of public performance data alone should not be expected to maximize the quality of the privatized service. Though enrollment increases did slow at lower performing voucher schools in Milwaukee after the release of test scores, the schools were still drawing additional pupils. Part of the increase in enrollment at lower performing schools may be a sign of dissatisfaction with traditional Milwaukee Public Schools, but the migration of students from one low-performing school to another hardly justifies the well-documented political upheavals accompanying Milwaukee’s education privatization effort (Dougherty, 2004).

The evidence uncovered in this first look suggests there is little rationale for not making performance data in school voucher programs and other privatization efforts available from the time of a program’s inception. If test scores were publicly available when the Milwaukee school choice program began in 1990, it is very possible that consumer behavior would have taken a very different trajectory; parents could have become accustomed to using test scores as part of the decision process. It is likely that the 20 years parents were forced to evaluate voucher schools on a case-by-case basis lessened the impact of the first-time release of test scores; the sudden availability of new information cannot be expected to instantly alter the parental decision-making process. For a market-based privatization effort to operate as advertised, consumers need quality tools to help guide their decision making.

Overall, this analysis, despite the discussed data limitations, gives preliminary reason to conclude that the release of test scores did have a positive, albeit limited, impact. Enrollment patterns veered from historical norms, with more parents choosing higher performing schools, fewer parents choosing lower performing schools, and a significant number of low-performing schools shutting their doors. However, it is also clear that consumer decision making in a market-based privatization effort is more complex than a simple review of performance data. Accordingly, policy makers are wise to introduce reasonable barriers of entry into, and reasonable triggers for a forced exit from, any marketplace for privatized public goods. More important, policy makers and researchers in Milwaukee and elsewhere should prioritize the routine surveying of parents and create data systems that can directly link parental decision making with school performance. The creation of such data systems would allow analyses such as these to go beyond a first look and better explain the ways in which parents actually navigate a school marketplace. Understanding this would enable improved design in school voucher programs, which can in turn maximize program quality.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biography

Michael Ford teaches public budgeting and analytic methods in the master’s of public administration program at the University of Wisconsin–Oshkosh. He formerly worked for several years in state policy with a strong focus on urban education. His research interests include school board governance, privatization, local government finance, and the legislative fiscal note process.

References

Carlson

Cowen

J. M.

Fleming

D. J.

(2013). Life after vouchers: What happens to students who leave private schools for the traditional public sector? Educational Evaluation and Policy Analysis, 35, 179-199.

Carnoy

Adamson

Chudgar

Luschei

Witte

(2007). Vouchers and public school performance. Washington, DC: Economic Policy Institute.

Chakrabarti

(2008). Can increasing private school participation and monetary loss in a voucher program affect public school performance? Evidence from Milwaukee. Journal of Public Economics, 92, 1371-1393.

Chubb

Moe

(1988). Politics, markets, and the organization of schools. The American Political Science Review, 82, 1066-1087.

Cowen

J. M.

Fleming

D. J.

Witte

J. F.

Wolf

P. J.

Kisida

(2013). School vouchers and student attainment: Evidence from a state-mandated study of Milwaukee’s Parental Choice Program. Policy Studies Journal, 41, 147-168.

Dickman

A. M.

(2004). Standardized tests taken by nine out of ten voucher schools: Reporting results would increase accountability. Milwaukee, WI: Public Policy Forum.

Dougherty

(2004). More than one struggle: The evolution of Black school reform in Milwaukee. Chapel Hill, NC: The University of North Carolina Press.

Ford

(2011). School Exits in the Milwaukee Parental Choice Program: Evidence of a Marketplace? Journal of School Choice, 5(2), 182-204.

Friedman

(1955). The role of government in public education. In Solow

R. A.

(Ed.), Economics and the public interest (pp. 123-144). New Brunswick, NJ: Rutgers Press.

10.

Friedman

(1962). Capitalism and freedom. Chicago, IL: University of Chicago Press.

11.

Greene

J. P.

Marsh

R. H.

(2009). The effect of Milwaukee’s Parental Choice Program on student achievement in Milwaukee public schools. Fayetteville: School Choice Demonstration Project, University of Arkansas.

12.

Greene

J. P.

Peterson

P. E.

(1999). Effectiveness of school choice: The Milwaukee experiment. Education and Urban Society, 31, 190-213.

13.

Howell

(2012). March 2012 poll of Milwaukee residents. Hartland, UK: Wisconsin Policy Research Institute.

14.

Hoxby

C. M.

(2003). School choice and school competition: Evidence from the United States. Swedish Economic Policy Review, 10, 11-67.

15.

Jacob

A. M.

Wolf

P. J.

(2012). Milwaukee longitudinal school choice evaluation: Annual School Testing Summary Report 2010-11. Fayetteville: School Choice Demonstration Project at the University of Arkansas.

16.

Kava

(2013). Milwaukee and Racine Parental Choice Programs (Informational Paper 29). Madison: Wisconsin Legislative Fiscal Bureau.

17.

Kisida

Jensen

L. I.

Wolf

P. J.

(2011). The Milwaukee Parental Choice Program: Descriptive Report on Participating Schools 2009 - 2010. Fayetteville, AK: The School Choice Demonstration Project at the University of Arkansas.

18.

Lubienski

(2007). Marketing schools: Consumer goods and competitive incentives for consumer information. Education and Urban Society, 40, 118-141.

19.

National Center for Education Statistics. (2011a). The nation’s report card: Trial urban district assessment mathematics 2011. Washington, DC: Institute of Education Sciences, U.S. Department of Education.

20.

National Center for Education Statistics. (2011b). The nation’s report card: Trial urban district assessment reading 2011. Washington, DC: Institute of Education Sciences, U.S. Department of Education.

21.

Richards

(2011, March 29). Choice schools not outperforming MPS. Milwaukee Journal Sentinel. Retrieved from http://www.jsonline.com/news/education/118820339.html

22.

Rouse

C. E.

(1998). Private school vouchers and student achievement: An evaluation of the Milwaukee parental choice program. The Quarterly Journal of Economics, 113, 553-602.

23.

Teske

Fitzpatrick

Kaplan

(2007). Opening doors: How low-income parents search for the right school. Seattle, WA: Center for Reinventing Public Education.

24.

Van Dunk

Dickman

(2003). School choice and the question of accountability: The Milwaukee experience. New Haven, CT: Yale University Press.

25.

Witte

J. F.

(2000). The market approach to education: An analysis of America’s first voucher program. Princeton, NJ: Princeton University Press.

26.

Witte

J. F.

Carlson

Cowen

J. M.

Fleming

D. J.

Wolf

P. J.

(2012). MPCP Longitudinal Educational Growth Study: Fifth year report. Fayetteville: School Choice Demonstration Project at the University of Arkansas.

27.

Witte

J. F.

Sterr

T. D.

Thorn

C. A.

(1995). Fifth-year report Milwaukee Parental Choice Program. Madison: Department of Political Science and The Robert M. La Follette Institute of Public Affairs, University of Wisconsin–Madison.

28.

Witte

J. F.

Wolf

Cowen

Fleming

Lucas-McLean

(2008). The MPCP longitudinal educational growth study first year report: SCDP Milwaukee evaluation report. Fayetteville, AK: The School Choice Demonstration Project at the University of Arkansas.

29.

Wolf

P. J.

Stewart

(2012). Detroit school shoppers study. Fayetteville: School Choice Demonstration Project at the University of Arkansas.