Abstract

We examine the characteristics of schools preferred by parents in New Orleans, Louisiana, where a “portfolio” of school choices is available. This tests the conditions under which school choice induces healthy competition between public and private schools through the threat of student exit. Using unique data from parent applications to as many as eight different schools (including traditional public, charter, and private schools), we find that many parents include a mix of public and private schools among their preferences, often ranking public schools alongside or even above private schools on a unified application. Parents who list both public and private schools show a preference for the private sector, all else equal, and are willing to accept lower school performance scores for private schools than otherwise equivalent public options. These parents reveal a stronger preference for academic outcomes than other parents and place less value on other school characteristics such as sports, arts, or extended hours. Public schools are more likely to be ranked with private schools and to be ranked higher as their academic performance scores increase.

1.  Introduction

Among the many systemic reforms that policy makers have implemented in American school districts over the past two decades, perhaps the most fundamental have been those designed to provide families with a choice of publicly supported schools beyond the traditional, neighborhood-based option. In large urban districts across the country, many of these reforms have been based on the “portfolio” model of district management, where a number of different entities—traditional public, charter, magnet, or even private schools receiving publicly funded tuition—are administered by a centrally coordinated authority (Bulkley, Henig, and Levin 2010). Districts and charter schools officially collaborate in eighteen large urban centers that include major cities such as Baltimore, Boston, Chicago, Denver, Los Angeles, Minneapolis, Philadelphia, and New Orleans (Betts and Tang 2014). In these and other districts nationwide, many superintendents are partnering with elected and appointed city officials, school board members, and private organizations to explicitly implement portfolio models (Betts and Tang 2014). In principle, these strategies allow parents, particularly those in low-income households, to choose schools that best fit their children's needs, while promoting innovation within and autonomy for individual schools. Accountability is introduced via a combination of centralized regulation at the district level and market forces generated by competition between providers for student enrollment. In theory, these market forces will pressure public schools to improve in order to compete with charter and private schools (Friedman 1997).

For individual schools in these districts, student retention is critical. Losing students to private, charter, or suburban schools can force difficult decisions on urban school boards, often resulting in closed schools. Chicago, Detroit, Philadelphia, Kansas City, Milwaukee, Pittsburgh, and Washington, DC, have each closed at least twenty public schools over the past decade.1 According to the National Center for Education Statistics (NCES 2013), almost 1,500 public schools across the country were shuttered in 2010–11 alone. As students and families become ever more able to exercise choice over their educational opportunities, existing public schools face stiff competition for students. The purpose of this paper is to identify the implicit value parents place on different types of schools within a portfolio model, and the specific characteristics of public schools that families appear to value when they consider exiting the public school system through a voucher program. To do this, we examine a unique school choice system where there are no default neighborhood schools, and parents simultaneously rank their preferences for traditional public, charter, and voucher schools in a citywide lottery that assigns both public school seats and private school vouchers. This enables us to address the relatively unexplored question of how public schools may compete with private schools in a voucher system. More specifically, we estimate whether parents prefer private schools for qualities that public schools cannot duplicate (such as a religious emphasis) or domains across which public schools could compete (such as academics or location).

This extends a growing body of school choice literature on student selection into charter schools and voucher programs. Our unique focus is to identify the specific characteristics of public and charter schools that more successfully compete with private schools in a fully implemented portfolio model and to evaluate these characteristics as evidence, whether or not the theory of public school improvement through competition is plausible. We do this by considering the unified applications of individual families to as many as eight public, charter, and private schools in New Orleans, where, in the aftermath of Hurricane Katrina in August 2005, the state education agency eliminated the city's neighborhood-based school system and created a portfolio system of charter schools, a small number of highly performing traditional public schools, and, eventually, a publicly funded private school voucher system.

We examine two central research questions regarding the preferences of low-income parents for public and private schools in New Orleans. First, we examine the relative value parents place on private versus public schools by estimating the relationship between school sector (public or private) and application choices and rankings, controlling for other school characteristics such as standardized test performance, size, and student demographics. We also disaggregate more types of schools to compare parent preferences for charter schools, traditional public schools, and private schools with different religious emphases. Second, we examine the qualities of public schools that are attractive to potential voucher families. We estimate the effects of a rich set of school characteristics on the probability that a public school will be ranked alongside, or even higher than, a private school on a student's application.

We find that many parents who apply for private school enrollment through vouchers are also willing to enroll at public schools with strong academic indicators. Approximately half of new voucher applicants ranked at least one public school on their applications, and 12 percent ranked a public school as their first choice over all private schools. We find that parents value private schools and high-performing traditional public schools over the city's typical charter school. Religiously affiliated private schools have higher probabilities of being preferred within the set of all private schools, although all private schools are more highly valued than charter schools. In addition, parents are willing to accept lower school performance scores for private schools than they are for otherwise equivalent public options, and some are willing to travel farther to attend a private school. Public schools that compete for the same set of students as private schools are more likely to be listed on applications as their school report card scores increase. In general, extracurricular and other offerings by those public schools do not increase their likelihood of being ranked on an application. On the other hand, we also find that parents who are likely to exit public schools for voucher programs often rank some public schools as highly as private schools, sometimes even higher, which suggests that some public schools are successfully competing with private schools. Some parents also prefer only religious private schools and rank public schools above secular private schools. Overall, these parents reveal a stronger preference for academic outcomes than other parents, and they place less value on other characteristics such as sports, arts, or school facilities. Given these results, school districts hoping to retain students who have the option to exit through a voucher program might have more success if they focus on academic outcomes rather than ancillary programs or diverse extracurricular activities.

2.  Background: Choice, Selection, and Competition

Historically, most school choice literature has focused on the impact that outside schooling options have on student academic performance (usually measured by state standardized exams). Green, Peterson, and Du (1999) and Rouse (1998) were among the first to show the effectiveness of vouchers in the Milwaukee school district, as measured by improvements in students’ academic outcomes. The first federally funded voucher program in Washington, DC, generated student achievement gains by students’ third year in the program (Wolf et al. 2013), and Angrist, Bettinger, and Kremer (2006) show long-term positive effects on the wages of school voucher recipients using data from Colombia. The charter school picture is more mixed, depending on the state and charter authorizers in question (e.g., CREDO 2014; Clark et al. 2015). Nonetheless, a number of studies have indicated positive impacts. For example, Abdulkadiroğlu et al. (2011) show the positive effect of charter schools in the Boston and New York City school districts, and Dobbie and Fryer (2011) present evidence that high-quality charter schools have a large impact on lower-income students. In a meta-analysis, Betts and Tang (2014) demonstrate that charter schools, on average, outperform traditional public schools on reading and math standardized exams, and that urban charters are generally more effective than their rural or suburban counterparts. Imberman (2011) provides evidence that charter schools may improve student attendance and behavior even if the effects on achievement are more muted. In other school choice work, Cullen, Jacob, and Levitt (2006) demonstrate the value of open-enrollment programs in Chicago Public Schools, and Engberg et al. (2014) show that magnet programs can improve student achievement and behavior. Recently, a number of studies have moved beyond test-score outcomes to focus on gains in student attainment, such as high school graduation and college enrollment. Wolf et al. (2013), Cowen et al. (2013), Chingos and Peterson (2015), and Booker et al. (2011) all show consistent evidence of positive voucher or charter impacts on attainment even when test score effects are modest or nonexistent.

The most prominent theoretical motivation for establishing a school choice system is based on the idea of market-induced competition among producers. Supporters of school choice stress the idea that increased school competition for students is a tide that can “lift all boats” by spurring achievement across sectors (e.g., Hoxby 2003). Epple and Romano (1998) and Ferreyra (2007) discuss the general equilibrium effects on student quality and extent of schooling options from increased competition between public and private voucher schools, and there is recent empirical evidence that in choice-rich environments such achievement improvements may be realized (Figlio and Hart 2014). These market-based models of choice implicitly assume that school quality is the dominant criterion on which parents choose, but other perspectives differ. A psychological perspective may stress that parents seek school brands (e.g., based on the notion of, say, a Catholic education; see Trivitt and Wolf 2011) as shortcuts to make more comprehensive schooling decisions that may or may not include academic quality. Demographic and sociological perspectives stress the role of racial composition, socioeconomic conditions, or an explicitly religious identity or affiliation among school attributes (e.g., Lankford and Wyckoff 1992; Glazerman 1998; Schneider et al. 1998; Cohen-Zada 2006; Trivitt and Wolf 2011; Fleming, et al. 2015). More generally, Hastings, Kane, and Staiger (2009) present evidence that parents have heterogeneous preferences for school characteristics, and that not every family values academic ratings of schools as highly as school choice supporters often suggest. Harris and Larsen (2015) provide evidence that lower-income families in New Orleans weigh academic performance of schools less than higher-income families. For that reason, some schools may offer other perks in order to attract potential students (such as athletics, extracurriculars, and after-school day care). McMillan (2004) shows that public schools may even choose to decrease their academic productivity in the face of increased competition from other schooling options.

Although many of these studies acknowledge varying explanations for student selection—including based on preferences for schooling alternatives—selection is typically an analytic obstacle to the goal of identifying choice effects on student outcomes. There is a large literature on differences between private and public school choosers overall (e.g., Long and Toma 1988; Buddin, Cordes, and Kirby 1998; Betts and Fairlie 2001; Lankford and Wyckoff 2001; Figlio and Stone 2010; Butler et al. 2013), which generally finds private schools “cream-skim” students from more advantaged backgrounds. Only a handful of studies, however, have examined the characteristics of students selecting private schools via means-tested voucher programs. The latter studies have generally confirmed that, consistent with their policy purpose, such programs disproportionately draw students from historically disadvantaged populations according to race and income (Howell 2004; Campbell, West, and Peterson 2005; Cowen 2010; Figlio, Hart, and Metzger 2010). Of these studies, however, only Figlio, Hart, and Metzger (2010) consider the different school characteristics that may form the basis of parental preferences, finding generally that parents select schools with fewer minority students. No study of which we are aware directly tests preferences for private, traditional public, or public charter schools.

Increasingly, policy makers and analysts alike view understanding such preferences to be important in its own right. In the present context, we are able to investigate a full school-choice system within a portfolio model from the perspective of how parents choose to exit or remain in the public school system. The basic argument for why school choice will improve student outcomes has two components: first, the new schools themselves offer better curricula, pedagogy, or have other characteristics that result in greater academic achievement than their public school counterparts; second, the remaining public schools react to this competition by adopting many of the best practices of successful charter and private voucher schools, while creating niches in order to meet their students’ needs. The previous set of papers addresses the first component, but the second has been less well studied. Identifying characteristics of public schools that could retain potential exiters has important implications for the success of public schools in a choice system.

3.  The Case of New Orleans

The implementation of Louisiana's state-funded voucher program in New Orleans provides a unique opportunity to make a direct comparison between parents’ preferences for public and private schools. We are able to observe the ranked preferences for public and private schools of the parents most likely to exit the public school system through vouchers—a population that is vital to the theory of public school improvement through voucher competition. We use these rankings to measure parents’ preferences for characteristics of public and private schools.

Prior to 2005, New Orleans public schools operated as a then-typical urban school district, with centralized control of neighborhood schools by the locally elected school board. Limited school choice was available through magnet programs, a few charter schools, and district-approved transfers. At this time, the district was severely underperforming, ranked 67 out of 68 districts in student performance on state standardized tests, in a state ranked 48 out of 50 on the National Assessment of Educational Progress. In August 2005, Hurricane Katrina caused the evacuation of the city, leading to the shutdown of all public schools, many of which were severely damaged by flooding.

During the period of post-Katrina closure, state and local leaders took action to completely reform the district, including many steps that were uniquely suited to a context of a temporary school shutdown and a sudden reduction in the school-aged population of the city. All teachers and school staff were laid off, and the teachers’ collective bargaining agreement expired without renegotiation. The state's Recovery School District (RSD) was given authority over all underperforming schools in the city, leaving the elected Orleans Parish School Board (OPSB) with only a small number of high-performing district-run campuses and charter schools. As students returned, efforts at system-wide reform continued. The severe damage to entire neighborhoods altered historic residential patterns, and access to safe housing varied by neighborhood, so RSD and OPSB eliminated most school attendance zones to allow any school to immediately enroll any returning student. This created a city of system-wide choice, with enrollment managed at the school level. Between 2006 and 2013, under the oversight of an expanded RSD and diminished OPSB, schools were either permanently closed or reopened in new or renovated buildings. By 2013, RSD had contracted all of the campuses under its control (over 90 percent of public schools in the city) to nonprofit managers, creating the nation's first predominantly charter school district.

School choice in New Orleans was further enhanced in 2008, when the state of Louisiana selected the city to pilot the new statewide voucher program, known as the Louisiana Scholarship Program (LSP). In 2008, the LSP pilot provided 640 vouchers for New Orleans students to attend local private schools. The program then expanded statewide, and currently 7,000 students, including 2,850 from Orleans Parish, receive vouchers each year. Whereas many U.S. cities have voucher programs either at the state or local level, only New Orleans situates the voucher program in the context of citywide school choice, widespread school autonomy, and a market dominated by nonprofit charter schools.

Open enrollment, including lotteries for oversubscribed schools, in a choice system this large proved to be both inefficient and inequitable. RSD thus implemented a centralized school lottery and enrollment process, known as OneApp, beginning in the 2012–13 school year. This study uses data from the lottery for enrollment in the 2013–14 school year. During that year the OneApp included all RSD schools, OPSB's open-enrollment schools, and all voucher schools. Importantly for our purposes, the OneApp is used to simultaneously rank parent preferences for and sort students into traditional public schools, publicly funded charter schools, and private voucher schools.2

There are at least 46 tuition-based, privately operated schools currently operating in the city of New Orleans, and 57 more in neighboring Jefferson Parish are within commuting distance.3 To participate in the voucher program, schools must agree to accept the voucher (approximately $5,300 on average in 2013–14) as full payment and may not charge voucher parents additional tuition or fees. Voucher schools must also accept all voucher applicants without additional admissions requirements. Finally, all voucher students must be tested annually on the same state standardized tests given to public school students. Results are made public and used to determine a school's continued eligibility for the program. These program characteristics led to relatively low private school participation in LSP (Kisida, Wolf, and Rhinesmith 2015). There are 41 private schools in New Orleans and neighboring Jefferson Parish that participate in the Louisiana state school voucher program.

The LSP is open to Louisiana public school students with family income below 250 percent of the federal poverty line. The number of applicants exceeds the total number of funded vouchers, so a centralized lottery determines which students may exit public schools to enroll in voucher schools. Any entering kindergartener meeting the income requirement may apply for a voucher to attend a participating private school. Students entering grades 1–12 must be either current voucher recipients or currently attending a school that receives a grade of C or below on the state's school report card.4

Voucher applications statewide are submitted through OneApp. The application includes a numeric ranking of preferred schools.5 Students enter a lottery for each school they rank, and are placed in the highest ranked school where they win the lottery. This ensures that the lottery is strategy-proof, and parents have no incentive to distort their choice set or order of preferences. Families should list their schools in preference order to have the best chance of gaining entrance to their preferred options. In the mechanism design literature, this is called random serial dictatorship (e.g., Abdulkadiroğlu and Sönmez 1998).

In New Orleans, because the OneApp is used to place students in both public and voucher schools, voucher applications are integrated into the citywide system of public school choice. With the exception of a handful of selective admissions charter schools,6 all New Orleans public schools participate in the OneApp. Importantly, New Orleans parents rank their preferences for public and voucher schools in a single application. Many New Orleans parents will complete the OneApp and apply only to public schools, but, as long they meet LSP eligibility, they can, for example, list a private school as the first choice, a charter school as the second choice, a district-run school as the third choice, a second private school as the fourth choice, and so forth—for up to eight choices. This offers the unique opportunity to examine both the relative value to parents of private schooling and the characteristics of public school options that are ranked with, or above, private schools.

4.  Data and Summary Statistics

Our dataset, provided by the Louisiana RSD, includes all students who applied for 2013–14 enrollment at a New Orleans public (charter or traditional) or voucher school through the 2013 OneApp process. This dataset includes complete applications for approximately 37,000 students who were entering grades kindergarten through 12.7 We focus on students who are meaningfully using the OneApp to choose a new school and exclude current students who listed their current school as their only choice. We include all students who were entering kindergarten in 2013 and, for students entering grades 1–12, those who are using the OneApp to attempt to exit their current school. Approximately 28,000 applications appear to be for students wishing to return to their current public or voucher school. The remaining 8,449 applications (23 percent) express meaningful intent to enter a new school by ranking at least one school higher than the current school attended. Although we cannot observe voucher eligibility at the student level, a large majority of students completing the OneApp were likely eligible for the voucher program because the overall free and reduced-price lunch (FRPL) rate, which is a less generous cutoff than voucher eligibility, was over 80 percent in New Orleans in 2013, and 49 of 67 public schools (73 percent) in the OneApp received school report card grades of C or below.

We next break down this group of 8,449 meaningful school choosers by their choice strategies. Despite widespread eligibility for the voucher program, 78 percent of these students applied only for public schools. We refer to this approach throughout as the public strategy. An additional 11 percent applied for only private schools.8 We refer to this as the private strategy. We focus our analysis in this study on the remaining 892 students (11 percent) who used the OneApp to rank both public and private schools. We refer to this as the mixed strategy. Although this subsample represents only 2.5 percent of all OneApp participants, it includes 10.5 percent of meaningful choosers, and 48.7 percent of new applicants to the voucher program. More importantly, to our knowledge, this small group reflects the only available simultaneous ranking of voucher and public schools by parents in a U.S. school district. This group represents parents who are at the margin of exiting public schools for a voucher but are also willing to remain in the public system, as evidenced by ranking both public and voucher schools. Through their exercise of choice, these parents and students are those who are theoretically responsible for ensuring that charter and voucher programs achieve improved general equilibrium effects through the competition that school choice proponents advocate. Of this group, 70 percent are identified as attempting to exit a school with a grade of D or F. It is important to reiterate that, although small relative to the number of students overall, these students are approximately equal to the number of students who apply to voucher schools only—a group that has received significant attention in the media, policy discussions, and academic literature. This suggests that in a fully developed portfolio system, a nontrivial number of families actively considers both school sectors.

Table 1 describes OneApp participants by grade level and strategy for all students and the subgroup that was either entering kindergarten or attempting to exit a public school. Overall, 10.5 percent of the OneApps include at least one voucher school. The rate of voucher application is much higher among meaningful choosers. In all grades, 21 percent of applicants include at least one voucher school. Twenty-eight percent of students entering grades 1–8, and 21 percent of students entering kindergarten, rank at least one voucher school. The voucher application rate is much lower in high school, with only 6 percent of meaningful choosers ranking a voucher school. Similarly, the mixed strategy is most popular in grades 1–8 and least popular in high school. For those using the mixed strategy, we observe a fairly even balance of public and private schools. The average mixed-strategy application ranks 6.1 choices (out of the possible 8) with 3.2 private schools and 2.8 public schools. Public schools are not necessarily the choice of last resort for these parents. Approximately 31 percent of mixed-strategy applications rank at least one public school above at least one voucher school, and 12.6 percent list a public school as their first choice above all voucher schools.

Table 1.
Frequency of Different OneApp Strategies
Public Schools OnlyVoucher Schools OnlyMixed Strategy% in Mixed Strategy
All OneApps     
Kindergarten 2,408 436 194 6.4 
Grades 1—8 22,610 2,455 644 2.5 
Grade 9—12 7,749 81 55 0.7 
All grades 32,767 2,972 893 2.4 
Students choosing a new school     
Kindergarten 2,408 436 194 6.4 
Grades 1—8 2,804 445 644 16.5 
Grade 9—12 1,423 40 55 3.6 
All grades 6,635 921 893 10.6 
Public Schools OnlyVoucher Schools OnlyMixed Strategy% in Mixed Strategy
All OneApps     
Kindergarten 2,408 436 194 6.4 
Grades 1—8 22,610 2,455 644 2.5 
Grade 9—12 7,749 81 55 0.7 
All grades 32,767 2,972 893 2.4 
Students choosing a new school     
Kindergarten 2,408 436 194 6.4 
Grades 1—8 2,804 445 644 16.5 
Grade 9—12 1,423 40 55 3.6 
All grades 6,635 921 893 10.6 

Source: Author calculations based on Louisiana Recovery School District administrative data.

Notes: Includes all first-round OneApp applications for students entering grades K—12 in Fall 2013. Strategies are based on the application's list of up to eight schools ranked by parents. Mixed-strategy applications include both public and private schools. All students entering kindergarten are considered to be choosing a new school. Students in grades 1—12 are considered to be choosing a new school if they list at least two schools on the OneApp, and the first choice school is not the student's current school.

There are some shortcomings of these data. First, the voucher schools represent a subset of New Orleans private schools. Students hoping to exit to nonparticipating private schools are not observed in the data.9 Second, selective admissions charter schools did not participate in the OneApp and managed their own admissions processes.10 If parents are attempting to exit by applying to these schools, it is also not observed in our data. Nevertheless, there is reason to believe that these omissions are relatively minor in the context of this study. Non-voucher private schools are typically quite expensive, making them inaccessible to most public school families, the majority of whom are eligible for FRPL. Selective admissions charter schools are free but admit most of their students at entry grades (either kindergarten or ninth grade) with few open slots in subsequent grades. Although the omission of these schools is important at transitions, it will have little impact on our results for other grades. We restrict much of our analysis to grades 1–8 to avoid these transition years. Thus, our results are generalizable to students from low-income families who cannot access selective admissions public schools or expensive private schools but not for other types of students. It is also likely that preferences vary for students who are currently enrolled in a school relative to those who are new entrants, due to a potential status quo bias among current students or concern about high costs of student transfers. We also present results for a subset of transitioning students in grades K–8 who have no option to return to a current school.11

We use the OneApp data to identify ranked school preferences for students in the analytic subsamples. Based on each student's grade level for the next school year, we are able to construct a full choice set of schools that each student could have ranked—that is, each public and voucher school that offers the student's upcoming grade level.12 The OneApp also identifies students who are eligible for the voucher program because they attended a school with a rating of C, D, or F. All students who applied for a voucher are assumed to be income-eligible for the program.13 The OneApp does not include any demographic characteristics for students, although it does include students’ home addresses. With this information, we calculated the distance from each student's home to each public and voucher school in the choice set.

For both public and private schools, we gathered publicly available measures of school quality from School Report Cards published annually by the Louisiana Department of Education (LDOE).14 Our measure of school performance is the School Performance Score (SPS) for public schools and the School Cohort Index (SCI) for voucher schools. The SPS is the basis for public school accountability in Louisiana and is based on state standardized testing of all students in grades 3–8, and end-of-course high school exams. The SCI is used to determine whether private schools are eligible to continue to enroll new voucher recipients. The SCI and SPS are based on the same state standardized tests, but private schools are only required to administer the tests to voucher recipients; the performance of tuition-paying students is not tested.15 Importantly, these scores are reported on the OneApp application for public schools but not private schools.16

For private schools, additional school characteristics were merged from the 2012 NCES Private Schools Universe Survey (PSS; see footnote 3). PSS variables include school enrollment, student demographics, and the religious emphasis and affiliation of the school. We coded three types of voucher schools in New Orleans: Catholic schools (N = 23), other religious schools (N = 13), and secular schools (N = 3).17 For public schools, demographics were calculated from 2012–13 student-level enrollment files provided by LDOE. We include a school's calculated total enrollment, percent white, percent FRPL, percent special education, and the number of disciplinary suspensions per student. Discipline strategies vary widely in New Orleans, ranging from positive behavioral supports to strict “no excuses” approaches. We proxy for the school discipline climate with the school's suspension rate (total suspensions per student), a statistic that likely reflects both the school discipline philosophy and the behavioral health of the student body. We also coded variables that indicated the presence of public school programs using information in the New Orleans Parents’ Guide, a catalog of public schools published annually by a local nonprofit organization to facilitate informed parent choice. We used the Parents’ Guide to create variables reflecting each public school's grade span, extracurricular activities, arts programs, sports, foreign language instruction, school hours, and afterschool care.18

Finally, we include three public school characteristics that are important in the context of New Orleans. Locally, the distinction between traditional public and charter schools is quite important, because the former escaped state takeover by being sufficiently high performing prior to Hurricane Katrina. Many of our empirical models distinguish public schools run by OPSB, which describes both the school's governance structure and a history of high performance prior to the hurricane. Second, during post-Katrina reforms, school managers had the option to rename schools or retain their pre-Katrina names. Historically, New Orleans natives place considerable importance on school names, and it is typical for adults to describe themselves in reference to the public schools they attended. We coded a dichotomous variable for whether a school retained its legacy name, which would indicate an effort to connect with the reputation and history of the pre-Katrina version of the school. Third, following Hurricane Katrina, the state and federal governments also provided funds for facilities, which have been allocated to build new schools and renovate historic sites. The quality of school facilities varies significantly between old and new or renovated buildings. To capture the attractiveness of new facilities to parents, we also include an indicator for whether the school occupies a new or recently renovated building.

Our data include all entering kindergarten students who completed the OneApp, and all students entering grades 1–12 who revealed a willingness to exit their current school by ranking two or more schools. Table 2 displays summary statistics for the 67 public schools that were ranked by these students.19 Statistics are summarized for all schools by grade level and by how schools were ranked in the mixed strategy. New Orleans schools do not follow the typical grade-level patterns for elementary, middle, and high schools, so a school is included at multiple levels if it covers a long or atypical grade span. Ignoring school capacity constraints, students have a large number of schools from which to choose when selecting their top eight choices to rank on the OneApp. There are 46 public schools offering kindergarten, 57 offering grades 1–8, and 22 offering grades 9–12, in addition to 41 voucher schools also available through the OneApp. Table 2 also displays summary statistics for all public schools that were ranked by students using the mixed strategy, as well as just those schools that were ranked above private schools in mixed strategies. The latter group reflects public schools that parents who use the mixed strategy prefer to all voucher schools. A school counts as ranked if at least one student included it in a mixed strategy, and the means displayed are not weighted by a school's overall popularity.

Table 2.
Public School Descriptive Statistics
By Grade LevelBy OneApp Rankings
KindergartenElementaryHigh SchoolAll Ranked SchoolsRanked with VouchersRanked Above Vouchers
Total enrollment 493.72 499.84 394.91 479.12 487.40 504.69 
 (168.24) (181.84) (250.38) (193.28) (188.13) (176.89) 
Percent white 0.02 0.02 0.00 0.01 0.01 0.02 
 (0.05) (0.05) (0.01) (0.05) (0.05) (0.05) 
Percent black 0.94 0.95 0.97 0.95 0.95 0.94 
 (0.11) (0.10) (0.04) (0.10) (0.10) (0.10) 
Percent FRPL 0.92 0.92 0.88 0.91 0.91 0.92 
 (0.10) (0.10) (0.09) (0.10) (0.10) (0.10) 
Percent gifted 0.02 0.02 0.03 0.02 0.02 0.02 
 (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) 
Percent SPED 0.10 0.10 0.12 0.10 0.10 0.10 
 (0.03) (0.03) (0.04) (0.04) (0.04) (0.04) 
SPS 74.67 76.36 70.29 74.67 75.27 75.61 
 (31.85) (31.55) (37.25) (32.52) (32.66) (33.29) 
Score not reported 0.13 0.11 0.09 0.11 0.11 0.12 
 (0.34) (0.31) (0.29) (0.32) (0.32) (0.32) 
# of student support staff 3.42 3.54 3.94 3.55 3.57 3.45 
 (1.16) (1.24) (1.85) (1.31) (1.31) (1.14) 
# of sports 2.89 3.37 6.06 3.58 3.57 3.18 
 (2.51) (2.59) (1.60) (2.64) (2.65) (2.46) 
# of arts activities 0.27 0.28 0.18 0.26 0.26 0.23 
 (0.62) (0.60) (0.39) (0.58) (0.58) (0.55) 
Optional after-school care 0.53 0.44 0.00 0.41 0.42 0.44 
 (0.50) (0.50) (0.00) (0.49) (0.50) (0.50) 
Extended school hours 0.83 0.84 0.86 0.84 0.84 0.84 
 (0.38) (0.37) (0.35) (0.37) (0.37) (0.37) 
Foreign language program 0.33 0.39 0.86 0.45 0.43 0.41 
 (0.47) (0.49) (0.35) (0.50) (0.50) (0.49) 
Suspension rate 0.23 0.27 0.94 0.37 0.37 0.29 
 (0.21) (0.29) (1.25) (0.62) (0.63) (0.38) 
Grade span 8.63 7.98 4.41 7.59 7.71 7.97 
 (1.89) (2.24) (1.94) (2.54) (2.45) (2.20) 
Legacy school name 0.74 0.70 0.68 0.71 0.71 0.70 
 (0.44) (0.46) (0.48) (0.45) (0.45) (0.46) 
New or renovated facility 0.24 0.21 0.05 0.19 0.20 0.24 
 (0.43) (0.41) (0.21) (0.40) (0.40) (0.43) 
Traditional public school 0.07 0.09 0.09 0.08 0.08 0.10 
 (0.25) (0.29) (0.29) (0.27) (0.28) (0.30) 
Number of schools 46 57 22 67 64 50 
By Grade LevelBy OneApp Rankings
KindergartenElementaryHigh SchoolAll Ranked SchoolsRanked with VouchersRanked Above Vouchers
Total enrollment 493.72 499.84 394.91 479.12 487.40 504.69 
 (168.24) (181.84) (250.38) (193.28) (188.13) (176.89) 
Percent white 0.02 0.02 0.00 0.01 0.01 0.02 
 (0.05) (0.05) (0.01) (0.05) (0.05) (0.05) 
Percent black 0.94 0.95 0.97 0.95 0.95 0.94 
 (0.11) (0.10) (0.04) (0.10) (0.10) (0.10) 
Percent FRPL 0.92 0.92 0.88 0.91 0.91 0.92 
 (0.10) (0.10) (0.09) (0.10) (0.10) (0.10) 
Percent gifted 0.02 0.02 0.03 0.02 0.02 0.02 
 (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) 
Percent SPED 0.10 0.10 0.12 0.10 0.10 0.10 
 (0.03) (0.03) (0.04) (0.04) (0.04) (0.04) 
SPS 74.67 76.36 70.29 74.67 75.27 75.61 
 (31.85) (31.55) (37.25) (32.52) (32.66) (33.29) 
Score not reported 0.13 0.11 0.09 0.11 0.11 0.12 
 (0.34) (0.31) (0.29) (0.32) (0.32) (0.32) 
# of student support staff 3.42 3.54 3.94 3.55 3.57 3.45 
 (1.16) (1.24) (1.85) (1.31) (1.31) (1.14) 
# of sports 2.89 3.37 6.06 3.58 3.57 3.18 
 (2.51) (2.59) (1.60) (2.64) (2.65) (2.46) 
# of arts activities 0.27 0.28 0.18 0.26 0.26 0.23 
 (0.62) (0.60) (0.39) (0.58) (0.58) (0.55) 
Optional after-school care 0.53 0.44 0.00 0.41 0.42 0.44 
 (0.50) (0.50) (0.00) (0.49) (0.50) (0.50) 
Extended school hours 0.83 0.84 0.86 0.84 0.84 0.84 
 (0.38) (0.37) (0.35) (0.37) (0.37) (0.37) 
Foreign language program 0.33 0.39 0.86 0.45 0.43 0.41 
 (0.47) (0.49) (0.35) (0.50) (0.50) (0.49) 
Suspension rate 0.23 0.27 0.94 0.37 0.37 0.29 
 (0.21) (0.29) (1.25) (0.62) (0.63) (0.38) 
Grade span 8.63 7.98 4.41 7.59 7.71 7.97 
 (1.89) (2.24) (1.94) (2.54) (2.45) (2.20) 
Legacy school name 0.74 0.70 0.68 0.71 0.71 0.70 
 (0.44) (0.46) (0.48) (0.45) (0.45) (0.46) 
New or renovated facility 0.24 0.21 0.05 0.19 0.20 0.24 
 (0.43) (0.41) (0.21) (0.40) (0.40) (0.43) 
Traditional public school 0.07 0.09 0.09 0.08 0.08 0.10 
 (0.25) (0.29) (0.29) (0.27) (0.28) (0.30) 
Number of schools 46 57 22 67 64 50 

Sources: Author calculations from Louisiana Department of Education administrative data and New Orleans Parents’ Guide, available at http://neworleansparentsguide.org/.

Notes: School means (unweighted) and standard deviations in parentheses. Elementary schools offer at least one grade K—8. High schools offer at least one grade 9-12. Some schools are included in both columns. Schools ranked with voucher schools must appear on at least one application that also includes private schools. Schools ranked above voucher schools must be ranked above the highest-ranked private school on the application. FRPL: free and reduced-price lunch; SPED: special education; SPS: School Performance Score.

Summary statistics provide limited insight into which characteristics of schools parents prefer because parents are choosing simultaneously across many school qualities, and the information provided by relative rankings and frequency of ranking is ignored. To illustrate how schools vary simultaneously across multiple variables, table 3 displays demographics and school performance statistics for the 25 most popular public schools in the OneApp. We measure overall popularity and popularity among students using the mixed strategy as the total number of times a school is ranked across all applications. These measures reveal that parents using the mixed strategy have different aggregate ordered preferences than other parents. To illustrate the choice set of voucher schools, table 4 displays comparable statistics for the 25 most popular private schools in the OneApp, based on frequency of their being ranked on all new voucher applications. We include measures of popularity for all OneApps submitted and for the subset of mixed-strategy OneApps. Again, rankings are slightly different in the mixed strategy, which suggests that these families might have distinct preferences.

Table 3.
Top 25 Public Schools by OneApp Rank Frequency
Frequency Ranked     
RankSchool (Grade Range)Any StrategyMixed StrategyAbove Voucher SchoolEnrollment% BlackSPS ScoreGradeType
Eleanor McMain High (7—12) 1180 78 18 1,526 86 118 Traditional 
Benjamin Franklin Elementary Math & Science (PK—8) 1176 225 101 1,396 94 108 Traditional 
Dr. ML King Charter School for Science & Technology (PK—12) 1009 107 33 3,040 100 102 Charter 
McDonogh 35 High (7—12) 911 40 1,710 98 101 Traditional 
Lafayette Academy New Orleans (PK—8) 765 105 25 1,872 99 84 Charter 
Lake Area New Tech Early College High (9—12) 676 20 650 98 102 Charter 
Martin Behrman Charter School (PK—8) 663 68 23 1,424 98 112 Charter 
Gentilly Terrace Elementary (PK—8) 617 91 22 890 96 86 Charter 
KIPP Believe Primary (K—4) 591 66 14 1,234 97 100 Charter 
10 Langston Hughes Academy (PK—8) 521 62 14 1,290 99 87 Charter 
11 Sci Academy (9—12) 516 14 362 91 129 Charter 
12 Sophie B. Wright Charter (6—12) 511 24 958 97 101 Charter 
13 Mary Dora Coghill Accelerated Academy (PK—8) 505 85 17 1,190 100   Charter 
14 KIPP McDonogh 15 (K—8) 504 80 17 2,295 94 102 Charter 
15 Fannie C. Williams Charter (PK—8) 481 41 19 1,094 97 86 Charter 
16 ReNEW Schaumburg Elementary (PK—8) 479 62 1,118 98   Charter 
17 Mary Bethune Elementary (PK—6) 443 76 35 760 94 107 Traditional 
18 Arthur Ashe Charter (K—8) 422 63 24 970 95 99 Charter 
19 Medard H. Nelson (PK—8) 408 43 988 98 95 Charter 
20 Akili Academy of New Orleans (K—6) 393 71 26 770 99 86 Charter 
21 Benjamin Banneker Elementary (PK—8) 376 31 808 95 62 Charter 
22 Morris Jeff Community School (PK—5) 345 40 21 620 52 101 Charter 
23 O. Perry Walker College and Career Preparatory High (9—12) 305 885 100 109 Charter 
24 RENEW Reed Elementary at Delores T. Aaron School (PK—8) 305 27 1,366 98 74 Charter 
25 Mildred Osborne Elementary (K—6) 301 51 13 646 96     Charter 
Frequency Ranked     
RankSchool (Grade Range)Any StrategyMixed StrategyAbove Voucher SchoolEnrollment% BlackSPS ScoreGradeType
Eleanor McMain High (7—12) 1180 78 18 1,526 86 118 Traditional 
Benjamin Franklin Elementary Math & Science (PK—8) 1176 225 101 1,396 94 108 Traditional 
Dr. ML King Charter School for Science & Technology (PK—12) 1009 107 33 3,040 100 102 Charter 
McDonogh 35 High (7—12) 911 40 1,710 98 101 Traditional 
Lafayette Academy New Orleans (PK—8) 765 105 25 1,872 99 84 Charter 
Lake Area New Tech Early College High (9—12) 676 20 650 98 102 Charter 
Martin Behrman Charter School (PK—8) 663 68 23 1,424 98 112 Charter 
Gentilly Terrace Elementary (PK—8) 617 91 22 890 96 86 Charter 
KIPP Believe Primary (K—4) 591 66 14 1,234 97 100 Charter 
10 Langston Hughes Academy (PK—8) 521 62 14 1,290 99 87 Charter 
11 Sci Academy (9—12) 516 14 362 91 129 Charter 
12 Sophie B. Wright Charter (6—12) 511 24 958 97 101 Charter 
13 Mary Dora Coghill Accelerated Academy (PK—8) 505 85 17 1,190 100   Charter 
14 KIPP McDonogh 15 (K—8) 504 80 17 2,295 94 102 Charter 
15 Fannie C. Williams Charter (PK—8) 481 41 19 1,094 97 86 Charter 
16 ReNEW Schaumburg Elementary (PK—8) 479 62 1,118 98   Charter 
17 Mary Bethune Elementary (PK—6) 443 76 35 760 94 107 Traditional 
18 Arthur Ashe Charter (K—8) 422 63 24 970 95 99 Charter 
19 Medard H. Nelson (PK—8) 408 43 988 98 95 Charter 
20 Akili Academy of New Orleans (K—6) 393 71 26 770 99 86 Charter 
21 Benjamin Banneker Elementary (PK—8) 376 31 808 95 62 Charter 
22 Morris Jeff Community School (PK—5) 345 40 21 620 52 101 Charter 
23 O. Perry Walker College and Career Preparatory High (9—12) 305 885 100 109 Charter 
24 RENEW Reed Elementary at Delores T. Aaron School (PK—8) 305 27 1,366 98 74 Charter 
25 Mildred Osborne Elementary (K—6) 301 51 13 646 96     Charter 

Notes: Frequency and rankings were calculated from individual applications for all students either entering kindergarten or attempting to change schools in grades 1—12. Enrollment was calculated from student-level Louisiana Department of Education (LDOE) data. School Performance School (SPS) score and grades were obtained from LDOE school report cards. T grade indicates school is in transition after changing operator. Missing grade indicates a new school with no test data.

Table 4.
Top 25 Private Schools by OneApp Rank Frequency
Frequency Ranked    
RankSchool (Grade Range)All New Voucher ApplicantsMixed StrategyEnrollment of Voucher Students% BlackSCI ScoreVoucher Student Retention RateReligious Affiliation
St. Mary's Academy (K—5 coed, 6—9 girls) 457 232 318 100 47.6 86 Catholic 
St. Leo the Great (K—6) 354 203 185 99 72.1 81 Catholic 
St. Peter Claver (K—8) 291 172 162 100 49.1 84 Catholic 
Resurrection of Our Lord School (K—8) 289 145 336 71 71.1 80 Catholic 
St. Joan of Arc (K—8) 237 111 157 100 50.5 76 Catholic 
St. Anthony (K—8) 234 100 82 59  79 Catholic 
Bishop McManus (K—12) 232 135 109 96 21.8 76 Protestant 
St. Stephen (K—8) 211 123 60 91  81 Catholic 
St. Augustine Jr. High (6—9 boys) 207 129 27 98 49.4 95 Protestant 
10 St. Rita (K—6) 204 105 65 97 62.2 86 Catholic 
11 Holy Ghost Elementary (K—8) 157 80 119 100 41 81 Catholic 
12 Upperroom Bible Church Academy (K—8) 128 75 73 98  66 Protestant 
13 St. Paul Lutheran School (K—8) 117 71 59 85  80 Protestant 
14 St. John Lutheran School (K—6) 116 70 74 64  80 Protestant 
15 Light City Christian Academy (K—12) 115 79 55 100  82 Protestant 
16 St. Agnes School (K—8) 105 42 77 43 54.9 82 Catholic 
17 Faith Christian Academy (K—6) 94 23 53 98  53 Protestant 
18 Conquering Word Christian Academy Eastbank (K—12) 86 70 17   47 Protestant 
19 Good Shepherd Nativity Mission School (K—7) 84 54 66 100 73.4 83 Catholic 
20 Holy Rosary Academy (K—12) 84 44 40 44  64 Catholic 
21 Life of Christ Christian Academy (K—12) 83 53 64 98  66 Protestant 
22 Our Lady of Prompt Succor School (K—8) 73 32 140 44 53.5 67 Catholic 
23 St. Andrew the Apostle (K—8) 61 16 21 21  76 Catholic 
24 St. Benedict the Moor (K—4) 58 25 53 97 93.8 92 Catholic 
25 Conquering Word Christian Academy (K—12) 56 48 13 97   61 Protestant 
Frequency Ranked    
RankSchool (Grade Range)All New Voucher ApplicantsMixed StrategyEnrollment of Voucher Students% BlackSCI ScoreVoucher Student Retention RateReligious Affiliation
St. Mary's Academy (K—5 coed, 6—9 girls) 457 232 318 100 47.6 86 Catholic 
St. Leo the Great (K—6) 354 203 185 99 72.1 81 Catholic 
St. Peter Claver (K—8) 291 172 162 100 49.1 84 Catholic 
Resurrection of Our Lord School (K—8) 289 145 336 71 71.1 80 Catholic 
St. Joan of Arc (K—8) 237 111 157 100 50.5 76 Catholic 
St. Anthony (K—8) 234 100 82 59  79 Catholic 
Bishop McManus (K—12) 232 135 109 96 21.8 76 Protestant 
St. Stephen (K—8) 211 123 60 91  81 Catholic 
St. Augustine Jr. High (6—9 boys) 207 129 27 98 49.4 95 Protestant 
10 St. Rita (K—6) 204 105 65 97 62.2 86 Catholic 
11 Holy Ghost Elementary (K—8) 157 80 119 100 41 81 Catholic 
12 Upperroom Bible Church Academy (K—8) 128 75 73 98  66 Protestant 
13 St. Paul Lutheran School (K—8) 117 71 59 85  80 Protestant 
14 St. John Lutheran School (K—6) 116 70 74 64  80 Protestant 
15 Light City Christian Academy (K—12) 115 79 55 100  82 Protestant 
16 St. Agnes School (K—8) 105 42 77 43 54.9 82 Catholic 
17 Faith Christian Academy (K—6) 94 23 53 98  53 Protestant 
18 Conquering Word Christian Academy Eastbank (K—12) 86 70 17   47 Protestant 
19 Good Shepherd Nativity Mission School (K—7) 84 54 66 100 73.4 83 Catholic 
20 Holy Rosary Academy (K—12) 84 44 40 44  64 Catholic 
21 Life of Christ Christian Academy (K—12) 83 53 64 98  66 Protestant 
22 Our Lady of Prompt Succor School (K—8) 73 32 140 44 53.5 67 Catholic 
23 St. Andrew the Apostle (K—8) 61 16 21 21  76 Catholic 
24 St. Benedict the Moor (K—4) 58 25 53 97 93.8 92 Catholic 
25 Conquering Word Christian Academy (K—12) 56 48 13 97   61 Protestant 

Notes: Frequency and rankings were calculated form individual applications for all students applying for new vouchers. School Cohort Index (SCI) scores, number of voucher students, and voucher student retention were obtained from LDOE reports for the 2012—13 school year. Enrollment, percent black, and religious affiliation were obtained from the 2012 and 2010 NCES Public Schools Universe Survey (PSS). Conquering Word Christian Academy Eastbank was not included in either PSS.

5.  Models

We now turn to a ceteris paribus analysis of the particular characteristics of individual vouchers and public schools that make them more attractive to families than the private schools available to them under the LSP. The unique feature of the OneApp application, as described above, is that it allows direct observation of a choice set of all schools (up to eight private, traditional public, or charter schools, or a combination of all three types) as constructed by the families themselves.

We estimate variants of the general model:
formula
1
where
formula
2
in which private is an indicator of a school's sector and is our estimate of the relationship between school sector and the probability that a parent ranks j on the OneApp. A represents the SPS or SCI scores described above,20 and distance is the linear distance between each school j and the address listed on the OneApp for student i (we also include a squared term for this distance) and the additional vector X of available school demographic characteristics. We also consider a specification of equation 2 in which we allow the relationship between school-level achievement and distance to vary with the probability that parents prioritize certain schools by interacting private with A and distance. is a vector of student fixed effects. Thus, we estimate the effect of differences across schools available within a student's full choice set. is the random disturbance term. In all versions of equation 1, we assume is independently and identically distributed with the extreme value distribution.
Next, to consider whether particular types of private schools are more appealing than others, we also estimate
formula
3
in which we replace private with a vector of indicators for type denoting traditional public, public charter, Catholic, other religiously affiliated private schools, or secular private schools.

To estimate equation 1, we construct a dataset that includes one student--school observation for each school that was available to a student in the OneApp (i.e., the school offered the student's upcoming grade level to students of the same gender). In our primary specification, if the student ranked the school j anywhere in his up to eight OneApp choices. By estimating each variant of equation 1 within each strata identified by the unique identification number assigned to each student's OneApp, we are using a conditional logit framework suggested by McFadden (1974) and common in a number of economic applications, most notably for our purposes in models of college choice (e.g., Bettinger and Long 2004; Long 2004). In the K–12 school choice literature, variations of equation 1 are also found in Harris and Larsen (2015) and Carlson, Cowen, and Fleming (2013).21

In addition to these conditional logits, we also estimate equation 1 as a rank-ordered logit, where the probability becomes the probability that school j is ranked above others on the OneApp. The coefficients reflect the relationship between a school's observable characteristics (X, A, and distance) and its probability of selection (conditional logit) or relative ranking (rank-ordered logit) compared to other schools available to the student.

Finally, to consider how different public schools may appeal to parents who are choosing between public and voucher schools, we restrict our data to public schools in a student's choice set and estimate:
formula
4
in which the public–private distinctions are excluded; X, A, and distance are the same as in equations 2 and 3; but S is a vector of many additional school characteristics beyond demographics that are known to families choosing in the public (charter or traditional) sector through the Parents’ Guide, including the presence of sports and arts programs, special education services, after-school care, extended school days, foreign language programs, and other characteristics included in table 2. For equation 4, we also estimate specifications in which we consider the probability that j is ranked above the student's top-ranked private school, indicating a public school that is strongly preferred by parents who are willing to exit the current school. Although we must rely on equations 2 and 3 to consider the marginal value that private schools have to parents considering both sectors, equation 4 allows us to directly test the characteristics of public schools that place them into consideration with private schools in the first place. Finally, we estimate a similar specification for students using the public strategy (i.e., those who did not attempt to exit public schools through the voucher program) to determine if those who may exit public schools have different preferences from other public school families.

6.  Results

Do Parents Prefer Private Schools?

Table 5 presents results from our primary estimations for all students. Panel A estimates the influence of a private school indicator compared to the reference group of all public schools, as modeled in equation 2.22 Specifications are included for a conditional logit for being selected at any rank (column 1), a conditional logit for being selected as a first choice (column 2), and rank-ordered logit of relative ranking (column 3). Panel B repeats the same specifications for equation 3 by replacing the single private school indicator with indicators for three types of private school (Catholic, other religion, and secular) and one type of public school (OPSB direct-run), compared with the reference group of public charter schools. All specifications also control for distance to the school, distance squared, the school's numeric report card score, an indicator if no score was available, the percent white students, and logged total enrollment. Table 6 replicates these specifications for three overlapping subsets of students: entering kindergarteners (table 6, panel A), current students entering grades 1–8 including those who can choose to remain in their current school (panel B), and students entering grades K–8 who must transition to a new school because of grade progression (panel C).23 Whereas the first two groups may include students who can remain in their current school, the third group should eliminate any status quo bias by including only students who must transition to a new school because of grade progression or initial entry.

Table 5.
Conditional Logit and Rank-Ordered Logit Predictions of Preferences for Private Schools—All Mixed-Strategy Students
Panel A: Comparing All Private School Types to All Public School TypesPanel B: Comparing Five Sectorial School Types (Public Charter Schools Omitted)
(1) Choice(2) First(3) Ranking(4) Choice(5) First(6) Ranking
Private school 2.022* 3.454* 1.925*    
 (0.066) (0.155) (0.061)    
OPSB direct-run    1.607* 2.251* 1.519* 
    (0.070) (0.166) (0.064) 
Catholic    1.884* 3.466* 1.795* 
    (0.064) (0.140) (0.059) 
Other religion    2.291* 4.195* 2.163* 
    (0.104) (0.248) (0.099) 
Secular private    0.930* 2.741* 0.812* 
    (0.286) (0.749) (0.282) 
Distance −0.356* −0.374* −0.328* −0.362* −0.379* −0.333* 
 (0.024) (0.047) (0.021) (0.024) (0.046) (0.021) 
Distance squared 0.010* 0.012* 0.010* 0.010* 0.011* 0.009* 
 (0.002) (0.004) (0.002) (0.002) (0.004) (0.002) 
State report card score 0.018* 0.023* 0.018* 0.009* 0.012* 0.009* 
 (0.001) (0.004) (0.001) (0.001) (0.003) (0.001) 
Score not reported 1.290* 1.706* 1.255* 0.554* 0.710* 0.566* 
 (0.119) (0.309) (0.116) (0.113) (0.246) (0.108) 
Percent white students −2.116* −3.035* −2.070* −1.727* −2.403* −1.690* 
 (0.127) (0.310) (0.123) (0.135) (0.313) (0.130) 
Total enrollment (logged) 0.812* 1.247* 0.767* 0.827* 1.259* 0.771* 
 (0.035) (0.080) (0.033) (0.033) (0.069) (0.031) 
Observations 71,069 66,602 71,069 71,069 66,602 71,069 
Panel A: Comparing All Private School Types to All Public School TypesPanel B: Comparing Five Sectorial School Types (Public Charter Schools Omitted)
(1) Choice(2) First(3) Ranking(4) Choice(5) First(6) Ranking
Private school 2.022* 3.454* 1.925*    
 (0.066) (0.155) (0.061)    
OPSB direct-run    1.607* 2.251* 1.519* 
    (0.070) (0.166) (0.064) 
Catholic    1.884* 3.466* 1.795* 
    (0.064) (0.140) (0.059) 
Other religion    2.291* 4.195* 2.163* 
    (0.104) (0.248) (0.099) 
Secular private    0.930* 2.741* 0.812* 
    (0.286) (0.749) (0.282) 
Distance −0.356* −0.374* −0.328* −0.362* −0.379* −0.333* 
 (0.024) (0.047) (0.021) (0.024) (0.046) (0.021) 
Distance squared 0.010* 0.012* 0.010* 0.010* 0.011* 0.009* 
 (0.002) (0.004) (0.002) (0.002) (0.004) (0.002) 
State report card score 0.018* 0.023* 0.018* 0.009* 0.012* 0.009* 
 (0.001) (0.004) (0.001) (0.001) (0.003) (0.001) 
Score not reported 1.290* 1.706* 1.255* 0.554* 0.710* 0.566* 
 (0.119) (0.309) (0.116) (0.113) (0.246) (0.108) 
Percent white students −2.116* −3.035* −2.070* −1.727* −2.403* −1.690* 
 (0.127) (0.310) (0.123) (0.135) (0.313) (0.130) 
Total enrollment (logged) 0.812* 1.247* 0.767* 0.827* 1.259* 0.771* 
 (0.035) (0.080) (0.033) (0.033) (0.069) (0.031) 
Observations 71,069 66,602 71,069 71,069 66,602 71,069 

Notes: Results are displayed as logit coefficients from conditional logit estimation of the probability of being listed as any choice (choice) or first choice (first), and rank-ordered logit for relative choice ranking (ranking). The analytic sample includes all students entering grades K—12 with mixed-strategy applications. Estimates on first exclude any schools that were never ranked first. Two private schools are omitted due to missing information—one serving grades K—3, and one serving grades K—12. Students who ranked either of these schools first are omitted from estimations on first. Score is the numeric School Performance School for public schools and Scholarship Cohort Index for private schools. Distance is measured in miles (as the crow flies) from the student's home address to the school. For estimates on ranking, rankings were reordered beginning with one for the remaining schools on applications where students ranked the omitted private schools. All estimates include student fixed effects and robust standard errors. OPSB: Orleans Parish School Board.

*p < 0.05.

Table 6.
Conditional Logit and Rank-Ordered Logit Predictions of Preferences for Private Schools–—Mixed-Strategy Student Subgroups
Panel A: Kindergarten
 Choice (1)First (2)Rank (3)Choice (4)First (5)Rank (6)
Private school 2.654* 3.629* 2.486*    
 (0.150) (0.322) (0.136)    
Five sectoral types (omitted group is charter schools) 
OPSB direct-run    1.694* 2.111* 1.594* 
    (0.155) (0.308) (0.141) 
Catholic    2.610* 3.609* 2.440* 
    (0.150) (0.311) (0.135) 
Other religion    2.643* 3.709* 2.450* 
    (0.228) (0.491) (0.216) 
Secular private    2.541* 3.691* 2.342* 
    (0.466) (1.172) (0.450) 
Other school characteristics       
Distance −0.425* −0.524* −0.391* −0.441* −0.546* −0.402* 
 (0.040) (0.057) (0.036) (0.042) (0.058) (0.038) 
Distance squared 0.015* 0.021* 0.014* 0.014* 0.021* 0.013* 
 (0.003) (0.003) (0.002) (0.003) (0.004) (0.003) 
State report card score 0.029* 0.035* 0.027* 0.019* 0.023* 0.018* 
 (0.003) (0.007) (0.003) (0.003) (0.006) (0.003) 
Score not reported 1.967* 2.437* 1.834* 1.301* 1.636* 1.241* 
 (0.218) (0.601) (0.207) (0.238) (0.610) (0.228) 
Percent white students −2.479* −3.178* −2.438* −2.411* −3.021* −2.374* 
 (0.290) (0.660) (0.282) (0.339) (0.743) (0.328) 
Total enrollment (logged) 0.985* 1.132* 0.924* 1.025* 1.098* 0.947* 
 (0.076) (0.167) (0.069) (0.071) (0.149) (0.064) 
Observations 15,714 14,985 15,714 15,714 14,985 15,714 
Panel B: Grades 1—8 
  Choice (7) First (8) Rank (9) Choice (10) First (11) Rank (12) 
Private school 1.953* 3.316* 1.843*    
 (0.072) (0.181) (0.067)    
Five sectoral types (omitted group is charter schools) 
OPSB direct-run    1.629* 2.449* 1.556* 
    (0.081) (0.202) (0.075) 
Catholic    1.797* 3.336* 1.698* 
    (0.069) (0.164) (0.064) 
Other religion    2.110* 4.085* 1.977* 
    (0.120) (0.296) (0.115) 
Secular private    0.489 2.190* 0.346 
    (0.388) (1.039) (0.385) 
Other school characteristics       
Distance −0.350* −0.344* −0.321* −0.356* −0.348* −0.326* 
 (0.028) (0.050) (0.024) (0.027) (0.050) (0.025) 
Distance squared 0.010* 0.009* 0.009* 0.009* 0.008* 0.008* 
 (0.002) (0.004) (0.002) (0.002) (0.004) (0.002) 
State report card score 0.014* 0.020* 0.014* 0.005* 0.008* 0.005* 
 (0.002) (0.005) (0.002) (0.002) (0.004) (0.002) 
Score not reported 1.016* 1.497* 0.996* 0.331* 0.484 0.354* 
 (0.149) (0.380) (0.148) (0.136) (0.282) (0.132) 
Percent white students −2.124* −2.842* −2.054* −1.747* −2.201* −1.685* 
 (0.148) (0.353) (0.142) (0.152) (0.348) (0.146) 
Total enrollment (logged) 0.790* 1.188* 0.734* 0.777* 1.208* 0.713* 
 (0.041) (0.093) (0.038) (0.040) (0.081) (0.038) 
Observations 52,284 48,941 52,284 52,284 48,941 52,284 
Panel C: No Guaranteed School 
  Choice (13) First (14) Rank (15) First Choice (16) Rank (17) (18) 
Private school 2.481* 3.767* 2.362*    
 (0.150) (0.292) (0.136)    
Five sectoral types (omitted group is charter schools) 
OPSB direct-run    1.684* 2.158* 1.534* 
    (0.132) (0.272) (0.118) 
Catholic    2.378* 3.738* 2.260* 
    (0.148) (0.269) (0.133) 
Other religion    2.806* 4.307* 2.643* 
    (0.217) (0.472) (0.202) 
Secular private    2.186* 4.039* 2.050* 
    (0.451) (1.144) (0.437) 
Other school characteristics       
Distance −0.376* −0.459* −0.350* −0.392* −0.478* −0.363* 
 (0.036) (0.050) (0.033) (0.038) (0.052) (0.035) 
Distance squared 0.014* 0.020* 0.013* 0.013* 0.020* 0.013* 
 (0.002) (0.003) (0.002) (0.003) (0.003) (0.002) 
State report card score 0.032* 0.032* 0.030* 0.021* 0.020* 0.020* 
 (0.003) (0.006) (0.002) (0.003) (0.005) (0.002) 
Score not reported 2.097* 1.990* 1.968* 1.239* 0.928* 1.171* 
 (0.196) (0.522) (0.186) (0.206) (0.469) (0.199) 
Percent white students −2.438* −3.496* −2.415* −2.097* −2.966* −2.080* 
 (0.273) (0.721) (0.265) (0.317) (0.754) (0.308) 
Total enrollment (logged) 0.969* 1.404* 0.932* 1.008* 1.354* 0.948* 
 (0.072) (0.159) (0.065) (0.063) (0.136) (0.056) 
Observations 19,359 18,711 19,359 19,359 18,711 19,359 
Panel A: Kindergarten
 Choice (1)First (2)Rank (3)Choice (4)First (5)Rank (6)
Private school 2.654* 3.629* 2.486*    
 (0.150) (0.322) (0.136)    
Five sectoral types (omitted group is charter schools) 
OPSB direct-run    1.694* 2.111* 1.594* 
    (0.155) (0.308) (0.141) 
Catholic    2.610* 3.609* 2.440* 
    (0.150) (0.311) (0.135) 
Other religion    2.643* 3.709* 2.450* 
    (0.228) (0.491) (0.216) 
Secular private    2.541* 3.691* 2.342* 
    (0.466) (1.172) (0.450) 
Other school characteristics       
Distance −0.425* −0.524* −0.391* −0.441* −0.546* −0.402* 
 (0.040) (0.057) (0.036) (0.042) (0.058) (0.038) 
Distance squared 0.015* 0.021* 0.014* 0.014* 0.021* 0.013* 
 (0.003) (0.003) (0.002) (0.003) (0.004) (0.003) 
State report card score 0.029* 0.035* 0.027* 0.019* 0.023* 0.018* 
 (0.003) (0.007) (0.003) (0.003) (0.006) (0.003) 
Score not reported 1.967* 2.437* 1.834* 1.301* 1.636* 1.241* 
 (0.218) (0.601) (0.207) (0.238) (0.610) (0.228) 
Percent white students −2.479* −3.178* −2.438* −2.411* −3.021* −2.374* 
 (0.290) (0.660) (0.282) (0.339) (0.743) (0.328) 
Total enrollment (logged) 0.985* 1.132* 0.924* 1.025* 1.098* 0.947* 
 (0.076) (0.167) (0.069) (0.071) (0.149) (0.064) 
Observations 15,714 14,985 15,714 15,714 14,985 15,714 
Panel B: Grades 1—8 
  Choice (7) First (8) Rank (9) Choice (10) First (11) Rank (12) 
Private school 1.953* 3.316* 1.843*    
 (0.072) (0.181) (0.067)    
Five sectoral types (omitted group is charter schools) 
OPSB direct-run    1.629* 2.449* 1.556* 
    (0.081) (0.202) (0.075) 
Catholic    1.797* 3.336* 1.698* 
    (0.069) (0.164) (0.064) 
Other religion    2.110* 4.085* 1.977* 
    (0.120) (0.296) (0.115) 
Secular private    0.489 2.190* 0.346 
    (0.388) (1.039) (0.385) 
Other school characteristics       
Distance −0.350* −0.344* −0.321* −0.356* −0.348* −0.326* 
 (0.028) (0.050) (0.024) (0.027) (0.050) (0.025) 
Distance squared 0.010* 0.009* 0.009* 0.009* 0.008* 0.008* 
 (0.002) (0.004) (0.002) (0.002) (0.004) (0.002) 
State report card score 0.014* 0.020* 0.014* 0.005* 0.008* 0.005* 
 (0.002) (0.005) (0.002) (0.002) (0.004) (0.002) 
Score not reported 1.016* 1.497* 0.996* 0.331* 0.484 0.354* 
 (0.149) (0.380) (0.148) (0.136) (0.282) (0.132) 
Percent white students −2.124* −2.842* −2.054* −1.747* −2.201* −1.685* 
 (0.148) (0.353) (0.142) (0.152) (0.348) (0.146) 
Total enrollment (logged) 0.790* 1.188* 0.734* 0.777* 1.208* 0.713* 
 (0.041) (0.093) (0.038) (0.040) (0.081) (0.038) 
Observations 52,284 48,941 52,284 52,284 48,941 52,284 
Panel C: No Guaranteed School 
  Choice (13) First (14) Rank (15) First Choice (16) Rank (17) (18) 
Private school 2.481* 3.767* 2.362*    
 (0.150) (0.292) (0.136)    
Five sectoral types (omitted group is charter schools) 
OPSB direct-run    1.684* 2.158* 1.534* 
    (0.132) (0.272) (0.118) 
Catholic    2.378* 3.738* 2.260* 
    (0.148) (0.269) (0.133) 
Other religion    2.806* 4.307* 2.643* 
    (0.217) (0.472) (0.202) 
Secular private    2.186* 4.039* 2.050* 
    (0.451) (1.144) (0.437) 
Other school characteristics       
Distance −0.376* −0.459* −0.350* −0.392* −0.478* −0.363* 
 (0.036) (0.050) (0.033) (0.038) (0.052) (0.035) 
Distance squared 0.014* 0.020* 0.013* 0.013* 0.020* 0.013* 
 (0.002) (0.003) (0.002) (0.003) (0.003) (0.002) 
State report card score 0.032* 0.032* 0.030* 0.021* 0.020* 0.020* 
 (0.003) (0.006) (0.002) (0.003) (0.005) (0.002) 
Score not reported 2.097* 1.990* 1.968* 1.239* 0.928* 1.171* 
 (0.196) (0.522) (0.186) (0.206) (0.469) (0.199) 
Percent white students −2.438* −3.496* −2.415* −2.097* −2.966* −2.080* 
 (0.273) (0.721) (0.265) (0.317) (0.754) (0.308) 
Total enrollment (logged) 0.969* 1.404* 0.932* 1.008* 1.354* 0.948* 
 (0.072) (0.159) (0.065) (0.063) (0.136) (0.056) 
Observations 19,359 18,711 19,359 19,359 18,711 19,359 

Notes: Results are displayed as logit coefficients from conditional logit estimation of the probability of being listed as any choice (choice) or first choice (first) and rank-ordered logit for relative choice ranking (rank). Estimates on first exclude any schools that were never ranked first. Two private schools are omitted due to missing information–—one serving grades K—3, and one serving grades K—12. Students who ranked either of these schools first are omitted from estimations on first. Score is the numeric School Performance School for public schools and Scholarship Cohort Index for private schools. Distance is measured in miles (as the crow flies) from the student's home address to the school. The analytic subsamples include students with mixed-strategy application who were: entering kindergarten only (panel A), entering grades 1—8 (panel B), and entering grades K—8, and must choose a new school due to having no guaranteed seat in a current school (panel C). All estimates include student fixed effects and robust standard errors. OPSB: Orleans Parish School Board.

*p < 0.05.

The results for all students and all three student subgroups unambiguously indicate that among parents who are considering both public and private schools, private schools are preferred. The coefficient estimate on from equation 2 is positive and significant. Translated to log odds, among all students, a private school is over 7 times more likely to be ranked than a public school and over 31 times more likely to be the student's first choice. This effect is even larger for students with no guaranteed school who are 11 times more likely to rank a private school than a public school, and 43 times more likely to rank the private school first. This remains the case even though significant relationships between distance and school achievement are also apparent and taken into account. Parents demonstrate a strong preference for schools nearer to their homes, and schools with higher SPS/SCI scores, all else equal. Converted to log odds, an additional mile to school reduces the likelihood of ranking by approximately 30 percent. The squared term on the distance coefficient indicates that at some point, the negative relationship between distance and school preference flattens out, perhaps because parents do not have an absolute preference for their very nearest schools. The estimated relationship between percent white and being ranked is negative, while the relationship between total enrollment and being ranked is positive.

The results with multiple indicators for school type add nuance to the results with a single private school indicator. Here we identify five types of schools: charter schools (omitted), OPSB direct-run traditional public schools, Catholic schools, other religious schools, and secular private schools. In kindergarten, all types of private schools are preferred. A kindergartener is approximately 13 times more likely to rank a Catholic school than a charter school, and 14 times more likely to rank a non-Catholic religious school. In grades 1–8, only Catholic and Protestant schools are preferred to charter schools. Traditional public schools—those that remained under school district control due to higher performance pre-Katrina—are also preferred to the charters, across all specifications and grade levels, controlling for distance, SPS, and school demographics. These results are consistent for transitioning students (table 6, panel C) indicating they are not influenced by status quo bias among parents of current students. A student with no guaranteed school is five times more likely to rank a traditional public school than a charter school overall, and more than eight times more likely to rank the traditional public school first. Among the students who cannot remain in their current schools (table 6, panel C), students are also five times more likely to rank a traditional public school, and eleven times more likely to rank a traditional public school first. With the exception of secular private schools in high grades, parents who consider both public and private schools appear to prefer any option to the typical charter school in New Orleans.24

Next, we investigate whether parents weigh different factors equally for public and private schools by interacting the private indicator with distance and SPS/SCI. These results are provided in table 7. The main effects in each column indicate again that private schools are preferred (marginal effects show the predicted probability of being listed increases by 0.041) and that parents still prioritize shorter distances and higher school SPS in choosing or ranking their schools on the OneApp. For parents of kindergarteners and transitioning students, the interaction of distance and private is significant and positive, indicating that these parents are willing to travel farther for private school, ceteris paribus. For those who can remain at their current school, the response to distance is similar for public and private schools. The importance of state report card grades is greatly diminished for private schools, across all specifications and grade levels. Although parents prefer schools with higher achievement, they are less responsive to SCI scores for private schools than SPS scores for public school. There may be four explanations for this difference. First, as Trivitt and Wolf (2011) note, parents may simply prefer private schools as a “brand” of education that encompasses a number of different preferences. In that framework, parents may be willing to give private schools a sort of academic “benefit of the doubt” that the gains to be had from private education trump whatever academic benefits may be reflected in higher SCI scores overall. It could also be true, however, that parents interested in sending their children to private schools are simply the sort of parents who are less concerned about proficiency rates on standardized test scores. Third, although the SCI is publicly available in the state's annual LSP report, it is not provided on the OneApp. The SPS score for public schools is printed on the application. Parents may simply not be acting on private school information that they do not have. To the extent that they realize the SCI score is based only on voucher recipients, parents actually demonstrate reduced preference for SCI scores in schools with more voucher students. Finally, the issue of bias within the SCI, which was discussed earlier and stems from the fact that only voucher recipients are included in its computation, could be partially responsible for these results if parents interpret the SCI as a limited representation of a private school's effects on student performance.

Table 7.
Predictions of Preferences for Private Schools Interacted with Distance and School Performance
Panel A: All GradesPanel B: KindergartenPanel C: Grades 1—8Panel D: No Guaranteed School
Choice (1)First (2)Rank (3)Choice (4)First (5)Rank (6)Choice (7)First (8)Rank (9)Choice (10)First (11)Rank (12)
Private school 5.437* 11.276* 5.198* 6.270* 11.603* 5.925* 5.138* 11.819* 4.917* 6.994* 10.204* 6.546* 
 (0.290) (0.775) (0.264) (0.553) (1.398) (0.506) (0.363) (1.030) (0.330) (0.484) (1.012) (0.434) 
Other characteristics             
Distance −0.481* −0.453* −0.458* −0.627* −0.639* −0.583* −0.462* −0.423 −0.441* −0.574* −0.573* −0.514* 
 (0.080) (0.228) (0.071) (0.071) (0.153) (0.065) (0.113) (0.320) (0.104) (0.061) (0.137) (0.052) 
Distance squared 0.024* 0.025 0.023* 0.037* 0.040* 0.035* 0.021* 0.023 0.020* 0.035* 0.038* 0.031* 
 (0.007) (0.019) (0.006) (0.006) (0.012) (0.006) (0.010) (0.027) (0.010) (0.005) (0.011) (0.004) 
State report card score 0.044* 0.087* 0.043* 0.059* 0.100* 0.055* 0.039* 0.091* 0.039* 0.065* 0.082* 0.060* 
 (0.002) (0.005) (0.002) (0.005) (0.012) (0.005) (0.003) (0.007) (0.003) (0.004) (0.008) (0.004) 
Score not reported 3.868* 7.786* 3.766* 5.062* 8.781* 4.760* 3.451* 8.239* 3.407* 5.541* 6.974* 5.109* 
 (0.220) (0.605) (0.208) (0.475) (1.274) (0.437) (0.263) (0.798) (0.253) (0.424) (0.920) (0.387) 
Percent white students −1.722* −2.520* −1.681* −1.965* −2.471* −1.921* −1.776* −2.340* −1.718* −1.865* −2.727* −1.826* 
 (0.118) (0.280) (0.113) (0.253) (0.572) (0.245) (0.140) (0.322) (0.134) (0.236) (0.636) (0.228) 
Total enrollment (logged) 0.599* 0.973* 0.562* 0.758* 0.775* 0.706* 0.590* 0.918* 0.544* 0.693* 1.078* 0.661* 
 (0.031) (0.069) (0.029) (0.066) (0.142) (0.060) (0.036) (0.078) (0.033) (0.062) (0.146) (0.057) 
Interactions             
Private×distance 0.160 0.058 0.169 0.244* 0.043 0.232* 0.146 0.085 0.163 0.231* 0.072 0.193* 
 (0.100) (0.250) (0.090) (0.083) (0.173) (0.077) (0.140) (0.359) (0.129) (0.073) (0.175) (0.069) 
Private×distance squared −0.018* −0.015 −0.018* −0.029* −0.018 −0.028* −0.015 −0.018 −0.016 −0.027* −0.019 −0.024* 
 (0.009) (0.022) (0.008) (0.007) (0.013) (0.007) (0.013) (0.031) (0.012) (0.006) (0.014) (0.006) 
Private×score −0.050* −0.095* −0.048* −0.052* −0.096* −0.049* −0.048* −0.103* −0.046* −0.064* −0.078* −0.059* 
 (0.003) (0.007) (0.003) (0.007) (0.014) (0.006) (0.003) (0.009) (0.003) (0.006) (0.010) (0.005) 
Private×not reported −4.305* −8.073* −4.128* −4.836* −8.528* −4.536* −4.016* −8.746* −3.893* −5.701* −6.981* −5.226* 
 (0.250) (0.666) (0.237) (0.575) (1.414) (0.529) (0.292) (0.869) (0.280) (0.491) (1.044) (0.446) 
Observations 71,069 66,602 71,069 15,714 14,985 15,714 52,284 48,941 52,284 19,359 18,711 19,359 
Panel A: All GradesPanel B: KindergartenPanel C: Grades 1—8Panel D: No Guaranteed School
Choice (1)First (2)Rank (3)Choice (4)First (5)Rank (6)Choice (7)First (8)Rank (9)Choice (10)First (11)Rank (12)
Private school 5.437* 11.276* 5.198* 6.270* 11.603* 5.925* 5.138* 11.819* 4.917* 6.994* 10.204* 6.546* 
 (0.290) (0.775) (0.264) (0.553) (1.398) (0.506) (0.363) (1.030) (0.330) (0.484) (1.012) (0.434) 
Other characteristics             
Distance −0.481* −0.453* −0.458* −0.627* −0.639* −0.583* −0.462* −0.423 −0.441* −0.574* −0.573* −0.514* 
 (0.080) (0.228) (0.071) (0.071) (0.153) (0.065) (0.113) (0.320) (0.104) (0.061) (0.137) (0.052) 
Distance squared 0.024* 0.025 0.023* 0.037* 0.040* 0.035* 0.021* 0.023 0.020* 0.035* 0.038* 0.031* 
 (0.007) (0.019) (0.006) (0.006) (0.012) (0.006) (0.010) (0.027) (0.010) (0.005) (0.011) (0.004) 
State report card score 0.044* 0.087* 0.043* 0.059* 0.100* 0.055* 0.039* 0.091* 0.039* 0.065* 0.082* 0.060* 
 (0.002) (0.005) (0.002) (0.005) (0.012) (0.005) (0.003) (0.007) (0.003) (0.004) (0.008) (0.004) 
Score not reported 3.868* 7.786* 3.766* 5.062* 8.781* 4.760* 3.451* 8.239* 3.407* 5.541* 6.974* 5.109* 
 (0.220) (0.605) (0.208) (0.475) (1.274) (0.437) (0.263) (0.798) (0.253) (0.424) (0.920) (0.387) 
Percent white students −1.722* −2.520* −1.681* −1.965* −2.471* −1.921* −1.776* −2.340* −1.718* −1.865* −2.727* −1.826* 
 (0.118) (0.280) (0.113) (0.253) (0.572) (0.245) (0.140) (0.322) (0.134) (0.236) (0.636) (0.228) 
Total enrollment (logged) 0.599* 0.973* 0.562* 0.758* 0.775* 0.706* 0.590* 0.918* 0.544* 0.693* 1.078* 0.661* 
 (0.031) (0.069) (0.029) (0.066) (0.142) (0.060) (0.036) (0.078) (0.033) (0.062) (0.146) (0.057) 
Interactions             
Private×distance 0.160 0.058 0.169 0.244* 0.043 0.232* 0.146 0.085 0.163 0.231* 0.072 0.193* 
 (0.100) (0.250) (0.090) (0.083) (0.173) (0.077) (0.140) (0.359) (0.129) (0.073) (0.175) (0.069) 
Private×distance squared −0.018* −0.015 −0.018* −0.029* −0.018 −0.028* −0.015 −0.018 −0.016 −0.027* −0.019 −0.024* 
 (0.009) (0.022) (0.008) (0.007) (0.013) (0.007) (0.013) (0.031) (0.012) (0.006) (0.014) (0.006) 
Private×score −0.050* −0.095* −0.048* −0.052* −0.096* −0.049* −0.048* −0.103* −0.046* −0.064* −0.078* −0.059* 
 (0.003) (0.007) (0.003) (0.007) (0.014) (0.006) (0.003) (0.009) (0.003) (0.006) (0.010) (0.005) 
Private×not reported −4.305* −8.073* −4.128* −4.836* −8.528* −4.536* −4.016* −8.746* −3.893* −5.701* −6.981* −5.226* 
 (0.250) (0.666) (0.237) (0.575) (1.414) (0.529) (0.292) (0.869) (0.280) (0.491) (1.044) (0.446) 
Observations 71,069 66,602 71,069 15,714 14,985 15,714 52,284 48,941 52,284 19,359 18,711 19,359 

Notes: Results are displayed as logit coefficients from conditional logit estimation of the probability of being listed as any choice (choice) or first choice (first), and rank-ordered logit for relative choice ranking (rank). Estimates on first exclude any schools that were never ranked first. Two private schools are omitted due to missing information–—one serving grades K—3, and one serving grades K—12. Students who ranked either of these schools first are omitted from estimations on first. Score is the numeric School Performance School for public schools and Scholarship Cohort Index for private schools. Distance is measured in miles (as the crow flies) from the student's home address to the school. The analytic subsamples include students with mixed-strategy applications who were: entering all grades K-12 (panel A), entering kindergarten only (panel B), entering grades 1-8 (panel C), and entering grades K-8, and must choose a new school due to having no guaranteed seat in a current school (panel D). All estimates include student fixed effects and robust standard errors.

*p < 0.05.

Characteristics of Competing Public Schools

The specifications in tables 5, 6, and 7 estimate whether parents prefer public or private schools overall. We now consider which public school characteristics appeal to parents enough to place them into consideration with private schools in the first place. We estimate equation 4 for the subset of public schools that are ranked by mixed-strategy parents. Table 8 presents results for all mixed-strategy applicants, as well as the three subgroups described above. Column 1 of each panel in table 8 predicts whether the school appears on the OneApp. Column 2 presents the probability that the school is actually ranked higher than the top-ranked private school, and column 3 is the rank-ordered logit results, where relative rank on the OneApp is predicted. As before, we estimate each of these for all grades (panel A), for kindergarten only (panel B), for students who are entering grades 1–8 and exiting a C, D, or F school (panel C), and for students who must transition to a new school due to grade progression (panel D).

Table 8.
Predictions of Preferences for Public School Characteristics—Mixed-Strategy Students and Subgroups
Panel A: All GradesPanel B: KindergartenPanel C: Grades 1—8Panel D: No Guaranteed School
Choice (1)Above Private (2)Rank (3)Choice (4)Above Private (5)Rank (6)Choice (7)Above Private (8)Rank (9)Choice (10)Above Private (11)Rank (12)
Total enrollment (logged) 1.232* 1.236* 1.102* 2.084* 2.163* 1.932* 1.300* 1.074* 1.189* 1.390* 2.063* 1.194* 
 (0.101) (0.192) (0.096) (0.213) (0.457) (0.200) (0.121) (0.245) (0.117) (0.224) (0.399) (0.205) 
Percent white students −1.262 −0.885 −1.838 0.655 4.781 0.400 −1.266 −3.768 −1.773 0.241 3.320 −0.568 
 (1.037) (2.374) (0.992) (2.366) (4.984) (2.294) (1.260) (2.733) (1.206) (1.951) (4.254) (1.817) 
Percent FRPL −1.689* −2.269 −1.791* −2.489 −1.715 −2.532* −1.466* −3.240* −1.603* −1.999* −1.627 −1.886* 
 (0.545) (1.236) (0.520) (1.272) (2.652) (1.230) (0.659) (1.458) (0.634) (1.016) (2.332) (0.945) 
Percent gifted 1.735 7.363 3.154 −7.695 −4.821 −7.354 1.168 10.372 2.016 −1.648 −3.342 0.695 
 (1.781) (4.501) (1.697) (4.358) (9.662) (4.107) (2.281) (5.587) (2.162) (3.757) (8.443) (3.480) 
Percent SPED −3.029* −6.220* −3.001* −6.542* −8.285 −5.875* −2.841* −5.975 −2.898* −5.295* −9.945* −4.272* 
 (1.102) (2.731) (1.072) (2.436) (5.064) (2.376) (1.371) (3.472) (1.348) (2.127) (4.629) (2.062) 
SPS 0.029* 0.061* 0.028* 0.036* 0.055* 0.035* 0.029* 0.063* 0.029* 0.049* 0.054* 0.047* 
 (0.003) (0.008) (0.003) (0.010) (0.019) (0.009) (0.004) (0.010) (0.004) (0.007) (0.014) (0.007) 
Score not reported 2.356* 4.424* 2.303* 2.377* 3.795* 2.308* 2.489* 4.651* 2.474* 3.681* 3.266* 3.499* 
 (0.285) (0.711) (0.273) (0.944) (1.830) (0.900) (0.415) (0.924) (0.400) (0.583) (1.207) (0.571) 
Distance −0.580* −0.631* −0.546* −0.640* −0.617* −0.587* −0.586* −0.643* −0.556* −0.628* −0.642* −0.566* 
 (0.036) (0.072) (0.034) (0.072) (0.159) (0.065) (0.044) (0.081) (0.042) (0.063) (0.145) (0.056) 
Distance squared 0.030* 0.034* 0.028* 0.033* 0.032* 0.030* 0.029* 0.034* 0.028* 0.039* 0.039* 0.034* 
 (0.003) (0.006) (0.003) (0.006) (0.014) (0.006) (0.004) (0.007) (0.004) (0.005) (0.012) (0.004) 
# student support staff −0.091* −0.058 −0.077* −0.274* −0.246 −0.251* −0.060 −0.019 −0.045 −0.131* −0.253 −0.118* 
 (0.033) (0.082) (0.032) (0.100) (0.204) (0.097) (0.039) (0.095) (0.037) (0.066) (0.141) (0.058) 
# of sports −0.008 −0.033 −0.019 0.094* 0.101 0.086* −0.011 −0.061 −0.019 −0.010 0.033 −0.030 
 (0.015) (0.031) (0.014) (0.039) (0.077) (0.038) (0.017) (0.035) (0.016) (0.029) (0.067) (0.027) 
# of arts activities −0.199* −0.519* −0.195* −0.261 −0.814 −0.271 −0.194* −0.415 −0.202* −0.408* −0.677* −0.341* 
 (0.068) (0.193) (0.065) (0.191) (0.423) (0.183) (0.083) (0.227) (0.079) (0.152) (0.296) (0.138) 
Optional aftercare −0.342* −0.292 −0.280* −0.923* −0.799 −0.869* −0.362* −0.268 −0.315* −0.519* −0.521 −0.403 
 (0.096) (0.218) (0.090) (0.217) (0.542) (0.207) (0.124) (0.272) (0.117) (0.220) (0.449) (0.211) 
Extended school hours −0.141 −0.023 −0.122 −0.150 0.144 −0.077 −0.157 −0.050 −0.128 0.115 0.118 0.056 
 (0.083) (0.203) (0.079) (0.256) (0.461) (0.249) (0.100) (0.241) (0.097) (0.160) (0.351) (0.138) 
Foreign language 0.116 0.145 0.142* −0.039 −0.036 −0.013 0.159 0.275 0.179* 0.057 0.042 0.104 
 (0.068) (0.163) (0.066) (0.141) (0.276) (0.138) (0.086) (0.204) (0.084) (0.130) (0.249) (0.124) 
Suspension rate 0.221 0.341 0.231* −0.514 0.815 −0.419 0.394* 0.354 0.397* 0.418 −0.012 0.345 
 (0.122) (0.276) (0.108) (0.499) (0.948) (0.478) (0.181) (0.445) (0.175) (0.220) (0.410) (0.184) 
Suspensions not reported −0.617* −1.327* −0.630* −0.646 −0.210 −0.580 −0.603* −1.816* −0.613* −0.573 −0.415 −0.599* 
 (0.159) (0.383) (0.158) (0.360) (0.850) (0.355) (0.206) (0.436) (0.205) (0.308) (0.712) (0.303) 
Grade span 0.031 0.044 0.018 0.110 0.073 0.107 0.039 0.049 0.032 0.062 0.039 0.040 
 (0.025) (0.060) (0.024) (0.073) (0.141) (0.071) (0.033) (0.078) (0.032) (0.060) (0.110) (0.058) 
Legacy school name −0.123 −0.227 −0.085 −0.267 −0.328 −0.217 −0.149 −0.217 −0.125 −0.085 −0.224 −0.037 
 (0.083) (0.202) (0.080) (0.221) (0.401) (0.213) (0.105) (0.253) (0.102) (0.184) (0.304) (0.178) 
New facility 0.289* 0.937* 0.296* 0.889* 1.162* 0.847* 0.128 0.860* 0.141 0.805* 1.379* 0.749* 
 (0.072) (0.172) (0.068) (0.148) (0.365) (0.144) (0.090) (0.212) (0.085) (0.126) (0.316) (0.120) 
OPSB district-run 0.940* 0.802 0.721* 2.661* 3.298* 2.537* 0.886* 0.110 0.756* 1.578* 2.769* 1.076* 
 (0.185) (0.520) (0.174) (0.573) (1.123) (0.539) (0.231) (0.632) (0.218) (0.399) (0.889) (0.361) 
Observations 35,526 9,557 36,575 8,685 2,835 8,730 26,031 6,647 27,035 9,659 3,321 10,082 
Panel A: All GradesPanel B: KindergartenPanel C: Grades 1—8Panel D: No Guaranteed School
Choice (1)Above Private (2)Rank (3)Choice (4)Above Private (5)Rank (6)Choice (7)Above Private (8)Rank (9)Choice (10)Above Private (11)Rank (12)
Total enrollment (logged) 1.232* 1.236* 1.102* 2.084* 2.163* 1.932* 1.300* 1.074* 1.189* 1.390* 2.063* 1.194* 
 (0.101) (0.192) (0.096) (0.213) (0.457) (0.200) (0.121) (0.245) (0.117) (0.224) (0.399) (0.205) 
Percent white students −1.262 −0.885 −1.838 0.655 4.781 0.400 −1.266 −3.768 −1.773 0.241 3.320 −0.568 
 (1.037) (2.374) (0.992) (2.366) (4.984) (2.294) (1.260) (2.733) (1.206) (1.951) (4.254) (1.817) 
Percent FRPL −1.689* −2.269 −1.791* −2.489 −1.715 −2.532* −1.466* −3.240* −1.603* −1.999* −1.627 −1.886* 
 (0.545) (1.236) (0.520) (1.272) (2.652) (1.230) (0.659) (1.458) (0.634) (1.016) (2.332) (0.945) 
Percent gifted 1.735 7.363 3.154 −7.695 −4.821 −7.354 1.168 10.372 2.016 −1.648 −3.342 0.695 
 (1.781) (4.501) (1.697) (4.358) (9.662) (4.107) (2.281) (5.587) (2.162) (3.757) (8.443) (3.480) 
Percent SPED −3.029* −6.220* −3.001* −6.542* −8.285 −5.875* −2.841* −5.975 −2.898* −5.295* −9.945* −4.272* 
 (1.102) (2.731) (1.072) (2.436) (5.064) (2.376) (1.371) (3.472) (1.348) (2.127) (4.629) (2.062) 
SPS 0.029* 0.061* 0.028* 0.036* 0.055* 0.035* 0.029* 0.063* 0.029* 0.049* 0.054* 0.047* 
 (0.003) (0.008) (0.003) (0.010) (0.019) (0.009) (0.004) (0.010) (0.004) (0.007) (0.014) (0.007) 
Score not reported 2.356* 4.424* 2.303* 2.377* 3.795* 2.308* 2.489* 4.651* 2.474* 3.681* 3.266* 3.499* 
 (0.285) (0.711) (0.273) (0.944) (1.830) (0.900) (0.415) (0.924) (0.400) (0.583) (1.207) (0.571) 
Distance −0.580* −0.631* −0.546* −0.640* −0.617* −0.587* −0.586* −0.643* −0.556* −0.628* −0.642* −0.566* 
 (0.036) (0.072) (0.034) (0.072) (0.159) (0.065) (0.044) (0.081) (0.042) (0.063) (0.145) (0.056) 
Distance squared 0.030* 0.034* 0.028* 0.033* 0.032* 0.030* 0.029* 0.034* 0.028* 0.039* 0.039* 0.034* 
 (0.003) (0.006) (0.003) (0.006) (0.014) (0.006) (0.004) (0.007) (0.004) (0.005) (0.012) (0.004) 
# student support staff −0.091* −0.058 −0.077* −0.274* −0.246 −0.251* −0.060 −0.019 −0.045 −0.131* −0.253 −0.118* 
 (0.033) (0.082) (0.032) (0.100) (0.204) (0.097) (0.039) (0.095) (0.037) (0.066) (0.141) (0.058) 
# of sports −0.008 −0.033 −0.019 0.094* 0.101 0.086* −0.011 −0.061 −0.019 −0.010 0.033 −0.030 
 (0.015) (0.031) (0.014) (0.039) (0.077) (0.038) (0.017) (0.035) (0.016) (0.029) (0.067) (0.027) 
# of arts activities −0.199* −0.519* −0.195* −0.261 −0.814 −0.271 −0.194* −0.415 −0.202* −0.408* −0.677* −0.341* 
 (0.068) (0.193) (0.065) (0.191) (0.423) (0.183) (0.083) (0.227) (0.079) (0.152) (0.296) (0.138) 
Optional aftercare −0.342* −0.292 −0.280* −0.923* −0.799 −0.869* −0.362* −0.268 −0.315* −0.519* −0.521 −0.403 
 (0.096) (0.218) (0.090) (0.217) (0.542) (0.207) (0.124) (0.272) (0.117) (0.220) (0.449) (0.211) 
Extended school hours −0.141 −0.023 −0.122 −0.150 0.144 −0.077 −0.157 −0.050 −0.128 0.115 0.118 0.056 
 (0.083) (0.203) (0.079) (0.256) (0.461) (0.249) (0.100) (0.241) (0.097) (0.160) (0.351) (0.138) 
Foreign language 0.116 0.145 0.142* −0.039 −0.036 −0.013 0.159 0.275 0.179* 0.057 0.042 0.104 
 (0.068) (0.163) (0.066) (0.141) (0.276) (0.138) (0.086) (0.204) (0.084) (0.130) (0.249) (0.124) 
Suspension rate 0.221 0.341 0.231* −0.514 0.815 −0.419 0.394* 0.354 0.397* 0.418 −0.012 0.345 
 (0.122) (0.276) (0.108) (0.499) (0.948) (0.478) (0.181) (0.445) (0.175) (0.220) (0.410) (0.184) 
Suspensions not reported −0.617* −1.327* −0.630* −0.646 −0.210 −0.580 −0.603* −1.816* −0.613* −0.573 −0.415 −0.599* 
 (0.159) (0.383) (0.158) (0.360) (0.850) (0.355) (0.206) (0.436) (0.205) (0.308) (0.712) (0.303) 
Grade span 0.031 0.044 0.018 0.110 0.073 0.107 0.039 0.049 0.032 0.062 0.039 0.040 
 (0.025) (0.060) (0.024) (0.073) (0.141) (0.071) (0.033) (0.078) (0.032) (0.060) (0.110) (0.058) 
Legacy school name −0.123 −0.227 −0.085 −0.267 −0.328 −0.217 −0.149 −0.217 −0.125 −0.085 −0.224 −0.037 
 (0.083) (0.202) (0.080) (0.221) (0.401) (0.213) (0.105) (0.253) (0.102) (0.184) (0.304) (0.178) 
New facility 0.289* 0.937* 0.296* 0.889* 1.162* 0.847* 0.128 0.860* 0.141 0.805* 1.379* 0.749* 
 (0.072) (0.172) (0.068) (0.148) (0.365) (0.144) (0.090) (0.212) (0.085) (0.126) (0.316) (0.120) 
OPSB district-run 0.940* 0.802 0.721* 2.661* 3.298* 2.537* 0.886* 0.110 0.756* 1.578* 2.769* 1.076* 
 (0.185) (0.520) (0.174) (0.573) (1.123) (0.539) (0.231) (0.632) (0.218) (0.399) (0.889) (0.361) 
Observations 35,526 9,557 36,575 8,685 2,835 8,730 26,031 6,647 27,035 9,659 3,321 10,082 

Notes: Results are displayed as logit coefficients from conditional logit estimation of the probability of being listed as any choice (choice) or being ranked above the top-ranked private school (above private) and rank-ordered logit for relative choice ranking (rank). Score is the numeric School Performance Score (SPS) for public schools. Distance is measured in miles (as the crow flies) from the student's home address to the school. The analytic subsamples include students with mixed-strategy applications who were: entering all grades K—12 (panel A), entering kindergarten only (panel B), entering grades 1—8 (panel C), and entering grades K—8 and must choose a new school due to having no guaranteed seat in a current school (panel D). All estimates include student fixed effects and robust standard errors. FRPL: free and reduced-price lunch; SPED: special education; OPSB: Orleans Parish School Board.

*p < 0.05.

There are a number of patterns related to demographics, school performance, and school services or extracurricular activities. The least ambiguous result, and the most consistent across specifications in table 8, as well as in the earlier models, is the link between SPS and school selection. The higher the SPS, the higher the probability that the school is ranked and ranked higher on the OneApp. Plainly, academics matter to these parents in the relative weight they assign to their school choices. Similarly consistent and straightforward to interpret is the relationship between home-to-school distance and school ranking. Across all specifications in table 8, parents prefer schools that are closer to home—at least, as the squared term indicates, to a point. Racial demographics of the schools appears to be insignificant in most specifications, but in most results there is a negative relationship between the percentage of FRPL students and the probability that a school is ranked. Similarly, in most specifications schools with higher percentages of special education students are less likely to be ranked and are ranked lower.

Parents appear to prefer schools with some extracurricular activities or services but are less likely to prioritize schools with others. The presence of sports programs increases the probability that schools are ranked higher on the OneApp for kindergarten families but are not related overall or for exiters (grades 1–8), and sports do no elevate schools to the first choice. The presence of a foreign language program is positively predictive of a higher rank for both the overall sample and for exiting families but insignificant for kindergarteners and students transitioning to a new school. Arts programs, after-school care, and extended school hours are all either negative or insignificant influences on parent rankings. Taken alongside our results relating SPS to the probability that a school is preferred, the evidence here suggests that parents who consider leaving the public school system are generally looking for higher academic quality, perhaps by rejecting schools with more extracurricular activities or special programs in the process.25

Finally, legacy schools do not appear to be preferred one way or the other, all else equal, and most specifications indicate stronger preferences for traditional public schools and new facilities, with stronger effects in kindergarten. For exiting families, the number of suspensions—our proxy for school climate regarding discipline—positively predicts both appearance on the OneApp and a school's ranking, indicating that, controlling for student demographics, exiting families may be looking for schools with a stronger disciplinary climate.

The difference between kindergarten and the results for other grades suggests that parent preferences may vary based on prior experiences in the public schools. We test this further by restricting the sample to students in grades 1–8 who are exiting the lowest-performing schools, those with a D or F SPS score, with results in table 9. Preferences for SPS and shorter distance are also evident in this subgroup. Other results indicate that D/F exiters prefer schools with fewer students with special academic needs (similar to the results overall). In general, families exiting D/F schools are even less likely to be influenced by special programs than the larger sample. Characteristics such as after-school care, arts, sports, suspension rate, and legacy name are all either not significant or negative influences on ranking and relative rank. These parents are more likely to select a public school as the first choice if it offers language instruction or has a new facility, and rankings are higher for traditional public schools, ceteris paribus.

Table 9.
Predictions of Preferences for Public School Characteristics–—Mixed-Strategy Students Exiting Failing Schools
Exiting a D/F School
Choice (1)Above Private (2)Ranking (3)
Total enrollment (logged) 1.004* 2.100* 0.930* 
 (0.138) (0.498) (0.134) 
Percent white students −1.250 −6.279 −1.658 
 (1.523) (5.373) (1.463) 
Percent FRPL −0.855 −2.805 −0.947 
 (0.761) (2.565) (0.733) 
Percent gifted 3.913 8.321 4.670 
 (2.534) (11.845) (2.463) 
Percent SPED −4.231* −12.600* −4.393* 
 (1.647) (5.702) (1.618) 
SPS 0.020* 0.101* 0.021* 
 (0.005) (0.022) (0.005) 
Score not reported 1.999* 6.982* 2.005* 
 (0.480) (1.814) (0.464) 
Distance −0.594* −0.753* −0.560* 
 (0.055) (0.158) (0.051) 
Distance squared 0.031* 0.050* 0.030* 
 (0.005) (0.011) (0.004) 
# of student support staff −0.065 −0.434* −0.045 
 (0.045) (0.161) (0.044) 
# of sports 0.012 −0.233* −0.001 
 (0.022) (0.081) (0.021) 
# arts activities −0.072 −0.491 −0.088 
 (0.091) (0.330) (0.088) 
Optional aftercare −0.304* −0.978 −0.270* 
 (0.144) (0.607) (0.136) 
Extended school hours −0.326* −0.096 −0.281* 
 (0.114) (0.369) (0.113) 
Foreign language program 0.122 1.095* 0.149 
 (0.101) (0.471) (0.099) 
Suspension rate 0.346 0.252 0.354 
 (0.210) (0.945) (0.202) 
Suspensions not reported −0.517* −0.777 −0.502* 
 (0.225) (0.945) (0.225) 
Grade span 0.041 0.152 0.033 
 (0.040) (0.146) (0.039) 
Legacy school name −0.165 −0.343 −0.145 
 (0.120) (0.473) (0.117) 
New facility −0.081 1.082* −0.050 
 (0.105) (0.499) (0.097) 
District-run 0.781* −0.149 0.662* 
 (0.253) (1.232) (0.243) 
Observations 17,882 2,540 18,886 
Exiting a D/F School
Choice (1)Above Private (2)Ranking (3)
Total enrollment (logged) 1.004* 2.100* 0.930* 
 (0.138) (0.498) (0.134) 
Percent white students −1.250 −6.279 −1.658 
 (1.523) (5.373) (1.463) 
Percent FRPL −0.855 −2.805 −0.947 
 (0.761) (2.565) (0.733) 
Percent gifted 3.913 8.321 4.670 
 (2.534) (11.845) (2.463) 
Percent SPED −4.231* −12.600* −4.393* 
 (1.647) (5.702) (1.618) 
SPS 0.020* 0.101* 0.021* 
 (0.005) (0.022) (0.005) 
Score not reported 1.999* 6.982* 2.005* 
 (0.480) (1.814) (0.464) 
Distance −0.594* −0.753* −0.560* 
 (0.055) (0.158) (0.051) 
Distance squared 0.031* 0.050* 0.030* 
 (0.005) (0.011) (0.004) 
# of student support staff −0.065 −0.434* −0.045 
 (0.045) (0.161) (0.044) 
# of sports 0.012 −0.233* −0.001 
 (0.022) (0.081) (0.021) 
# arts activities −0.072 −0.491 −0.088 
 (0.091) (0.330) (0.088) 
Optional aftercare −0.304* −0.978 −0.270* 
 (0.144) (0.607) (0.136) 
Extended school hours −0.326* −0.096 −0.281* 
 (0.114) (0.369) (0.113) 
Foreign language program 0.122 1.095* 0.149 
 (0.101) (0.471) (0.099) 
Suspension rate 0.346 0.252 0.354 
 (0.210) (0.945) (0.202) 
Suspensions not reported −0.517* −0.777 −0.502* 
 (0.225) (0.945) (0.225) 
Grade span 0.041 0.152 0.033 
 (0.040) (0.146) (0.039) 
Legacy school name −0.165 −0.343 −0.145 
 (0.120) (0.473) (0.117) 
New facility −0.081 1.082* −0.050 
 (0.105) (0.499) (0.097) 
District-run 0.781* −0.149 0.662* 
 (0.253) (1.232) (0.243) 
Observations 17,882 2,540 18,886 

Notes: Results are displayed as logit coefficients from conditional logit estimation of the probability of being listed as any choice (choice) or being ranked above the top-ranked private school (above private) and rank-ordered logit for relative choice ranking (ranking). Score is the numeric School Performance Score (SPS) for public schools. Distance is measured in miles (as the crow flies) from the student's home address to the school. The analytic subsamples include students with mixed-strategy applications who were currently enrolled in a D- or F-rated public school. All estimates include student fixed effects and robust standard errors. FRPL: free and reduced-price lunch; SPED: special education.

*p < 0.05.

Finally, we compare the preferences of our subsample of mixed-strategy parents with the more typical parents who ranked only public schools. This provides a contrast between parents willing to exit to voucher schools and those who are not. Table 10 displays results for the subsample of parents of students entering kindergarten and those entering grades 1–8 who are willing to exit their public school without applying for voucher schools. As above, those who rank only public schools are more likely to select higher SPS schools and schools that are closer to their local neighborhoods. However, SPS is not statistically significant for kindergarten parents in the public strategy. Unlike mixed-strategy parents, there is a large, significant relationship between gifted enrollment and ranking, which is negative in kindergarten and positive in grades 1–8. Schools with arts programs and optional after-school care are all less preferred, and sports are preferred among public schools for entering kindergarten families only. The relationship between suspensions and school preferences varies by grade level, with only kindergarten families indicating a preference for lower rates. Parents committed to public schools also appear to prefer OPSB schools at all levels, just like their mixed-strategy counterparts.

Table 10.
Predictions of Preferences for Public School Characteristics–—Public-Strategy Students
Panel A: All StudentsPanel B: KindergartenPanel C: Grades 1—8
Choice (1)Rank (2)Choice (3)Rank (4)Choice (5)Rank (6)
Enrollment (logged) 0.840* 0.748* 1.414* 1.327* 1.026* 0.953* 
 (0.034) (0.031) (0.086) (0.081) (0.051) (0.047) 
Percent white −0.297 −0.685 0.164 −0.093 −0.850 −1.042* 
 (0.380) (0.358) (0.746) (0.719) (0.548) (0.520) 
Percent FRPL −1.143* −1.122* −2.059* −2.077* −0.702* −0.721* 
 (0.189) (0.178) (0.454) (0.438) (0.261) (0.247) 
Percent gifted 4.202* 4.951* −3.575* −3.636* 5.464* 5.208* 
 (0.581) (0.537) (1.183) (1.129) (0.904) (0.845) 
Percent SPED −4.122* −3.880* 2.221* 2.229* −3.465* −3.087* 
 (0.370) (0.348) (0.836) (0.805) (0.566) (0.545) 
SPS 0.018* 0.016* 0.004 0.005 0.017* 0.017* 
 (0.001) (0.001) (0.002) (0.002) (0.002) (0.002) 
Score not reported 1.281* 1.168* −0.046 0.048 1.223* 1.232* 
 (0.087) (0.081) (0.242) (0.232) (0.171) (0.161) 
Distance −0.576* −0.518* −0.785* −0.744* −0.596* −0.556* 
 (0.013) (0.012) (0.027) (0.026) (0.018) (0.017) 
Distance squared 0.030* 0.027* 0.040* 0.038* 0.028* 0.026* 
 (0.001) (0.001) (0.002) (0.002) (0.002) (0.001) 
# of student support staff −0.046* −0.033* −0.152* −0.130* 0.004 0.017 
 (0.010) (0.009) (0.031) (0.029) (0.015) (0.014) 
# of sports 0.012 0.003 0.041* 0.034* 0.008 0.000 
 (0.006) (0.006) (0.015) (0.015) (0.008) (0.007) 
# of arts activities −0.212* −0.204* −0.195* −0.214* −0.175* −0.193* 
 (0.024) (0.022) (0.051) (0.049) (0.030) (0.029) 
Optional aftercare −0.089* −0.030 −0.395* −0.369* −0.118* −0.097* 
 (0.031) (0.029) (0.079) (0.075) (0.047) (0.044) 
Extended school hours 0.139* 0.117* −0.112 −0.078 0.077 0.077* 
 (0.028) (0.027) (0.079) (0.075) (0.040) (0.038) 
Foreign language program −0.058* −0.046 0.174* 0.174* −0.033 −0.023 
 (0.026) (0.025) (0.049) (0.048) (0.034) (0.032) 
Suspension rate 0.311* 0.270* −0.545* −0.477* 0.280* 0.275* 
 (0.027) (0.024) (0.153) (0.146) (0.076) (0.072) 
Rate not reported −0.262* −0.284* −0.075 −0.046 −0.409* −0.396* 
 (0.049) (0.047) (0.101) (0.099) (0.078) (0.074) 
Grade span −0.027* −0.039* 0.023 0.024 −0.015 −0.014 
 (0.008) (0.007) (0.020) (0.019) (0.013) (0.012) 
Legacy school name 0.035 0.077* −0.003 0.006 0.012 0.024 
 (0.029) (0.028) (0.065) (0.063) (0.046) (0.045) 
New facility 0.298* 0.298* 0.127* 0.112* 0.166* 0.162* 
 (0.026) (0.025) (0.056) (0.053) (0.038) (0.035) 
District-run 0.618* 0.425* 1.491* 1.422* 0.461* 0.394* 
 (0.065) (0.057) (0.168) (0.161) (0.100) (0.092) 
Observations 208,342 208,342 89,820 89,820 100,357 100,357 
Panel A: All StudentsPanel B: KindergartenPanel C: Grades 1—8
Choice (1)Rank (2)Choice (3)Rank (4)Choice (5)Rank (6)
Enrollment (logged) 0.840* 0.748* 1.414* 1.327* 1.026* 0.953* 
 (0.034) (0.031) (0.086) (0.081) (0.051) (0.047) 
Percent white −0.297 −0.685 0.164 −0.093 −0.850 −1.042* 
 (0.380) (0.358) (0.746) (0.719) (0.548) (0.520) 
Percent FRPL −1.143* −1.122* −2.059* −2.077* −0.702* −0.721* 
 (0.189) (0.178) (0.454) (0.438) (0.261) (0.247) 
Percent gifted 4.202* 4.951* −3.575* −3.636* 5.464* 5.208* 
 (0.581) (0.537) (1.183) (1.129) (0.904) (0.845) 
Percent SPED −4.122* −3.880* 2.221* 2.229* −3.465* −3.087* 
 (0.370) (0.348) (0.836) (0.805) (0.566) (0.545) 
SPS 0.018* 0.016* 0.004 0.005 0.017* 0.017* 
 (0.001) (0.001) (0.002) (0.002) (0.002) (0.002) 
Score not reported 1.281* 1.168* −0.046 0.048 1.223* 1.232* 
 (0.087) (0.081) (0.242) (0.232) (0.171) (0.161) 
Distance −0.576* −0.518* −0.785* −0.744* −0.596* −0.556* 
 (0.013) (0.012) (0.027) (0.026) (0.018) (0.017) 
Distance squared 0.030* 0.027* 0.040* 0.038* 0.028* 0.026* 
 (0.001) (0.001) (0.002) (0.002) (0.002) (0.001) 
# of student support staff −0.046* −0.033* −0.152* −0.130* 0.004 0.017 
 (0.010) (0.009) (0.031) (0.029) (0.015) (0.014) 
# of sports 0.012 0.003 0.041* 0.034* 0.008 0.000 
 (0.006) (0.006) (0.015) (0.015) (0.008) (0.007) 
# of arts activities −0.212* −0.204* −0.195* −0.214* −0.175* −0.193* 
 (0.024) (0.022) (0.051) (0.049) (0.030) (0.029) 
Optional aftercare −0.089* −0.030 −0.395* −0.369* −0.118* −0.097* 
 (0.031) (0.029) (0.079) (0.075) (0.047) (0.044) 
Extended school hours 0.139* 0.117* −0.112 −0.078 0.077 0.077* 
 (0.028) (0.027) (0.079) (0.075) (0.040) (0.038) 
Foreign language program −0.058* −0.046 0.174* 0.174* −0.033 −0.023 
 (0.026) (0.025) (0.049) (0.048) (0.034) (0.032) 
Suspension rate 0.311* 0.270* −0.545* −0.477* 0.280* 0.275* 
 (0.027) (0.024) (0.153) (0.146) (0.076) (0.072) 
Rate not reported −0.262* −0.284* −0.075 −0.046 −0.409* −0.396* 
 (0.049) (0.047) (0.101) (0.099) (0.078) (0.074) 
Grade span −0.027* −0.039* 0.023 0.024 −0.015 −0.014 
 (0.008) (0.007) (0.020) (0.019) (0.013) (0.012) 
Legacy school name 0.035 0.077* −0.003 0.006 0.012 0.024 
 (0.029) (0.028) (0.065) (0.063) (0.046) (0.045) 
New facility 0.298* 0.298* 0.127* 0.112* 0.166* 0.162* 
 (0.026) (0.025) (0.056) (0.053) (0.038) (0.035) 
District-run 0.618* 0.425* 1.491* 1.422* 0.461* 0.394* 
 (0.065) (0.057) (0.168) (0.161) (0.100) (0.092) 
Observations 208,342 208,342 89,820 89,820 100,357 100,357 

Notes: Results are displayed as logit coefficients from conditional logit estimation of the probability of being listed as any choice (choice) and rank-ordered logit for relative choice ranking (rank). Score is the numeric School Performance Score (SPS). Distance is measured in miles (as the crow flies) from the student's home address to the school. Includes students with public-strategy applications who were: entering all grades K—12 (panel A), kindergarten only (panel B), and grades 1—8 (panel C). All estimates include student fixed effects and robust standard errors. FRPL: free and reduced-price lunch; SPED: special education; SPS: School Performance Score.

*p < 0.05.

7.  Discussion

Although portfolio-style models are used in many of the nation's largest cities—Boston, Chicago, Los Angeles, and New York City among them—no urban area has implemented the portfolio model as extensively as New Orleans. Since the city reconstituted its school system after Hurricane Katrina, public schools have faced competition from charter and private schools, many of which accept vouchers to allow low-income students to attend. Our samples of students that are either choosing an initial schooling option (kindergarten), choosing potentially to exit a public school (grades 1–8), or transitioning to a new school because of grade progression, generally exhibit discernible patterns in their preferences. Our results consistently demonstrate that many parents strongly prefer private schools to public schools, but there are characteristics of public schools that appear to make them more competitive. Nevertheless, the results do not point to a consistent strategy that public schools should pursue to retain students.

School type appears to influence parents’ preferences. In our first models, parents unambiguously prefer private schools overall, whether Catholic, affiliated with another religious tradition, or secular. These preferences are strong relative to the city's charter schools, but parents also prefer traditional public schools, suggesting that they are not necessarily seeking private schools above all but are also willing to “hedge their bets” by listing public schools with a history of strong academic performance alongside their preferred private choices.

There is also some evidence that socioeconomic characteristics of schools influence parents’ preferences for schools in New Orleans, as other research has indicated. Generally speaking, race played little discernible role in school choice on average in New Orleans, but there is little variation in school populations, as the vast majority of families are nonwhite. Socioeconomic status, however, may play a role along with the academic characteristics of a school's students. In a number of our specifications, schools with higher levels of FRPL students and higher levels of students with special needs were less likely to be selected by choosing parents. A strong preference for high academic performance of public schools is clearly shown across all students and in nearly all models and specifications. The higher the school's performance score, the higher the probability that the school is ranked higher on the OneApp.

Evidence regarding whether parents prefer niche offerings of public schools, related either to academics (arts or foreign language programs, extended school days) or nonacademic programs (sports and other extracurricular activities), is mixed. Although public schools might be tempted to add extracurricular activities to attract families away from private schools, this appears to be a poor strategy if not combined with academic improvements. Not surprisingly, distance plays an important role as well: parents prefer schools nearer to them, whether public or private.

These results have potentially significant implications for policy making in settings where there are multiple alternatives to traditional public schools. Such “high choice” environments are increasingly common across the country (Cowen and Toma 2015), regardless of whether they are formal portfolio systems. How parents make their choices, and on what basis, are still questions that only a handful of recent studies have considered. We are able to focus on a particularly important type of parent who is unsatisfied with a public school but is open to both private and public alternatives. That socioeconomic status of students and academic performance influence the preferences of parents who are considering exiting the public school system is a reminder that school choice may not necessarily improve—and may even exacerbate—historical patterns of inequality within the public school system. On the other hand, that higher-performing public schools with high minority and FRPL populations in New Orleans do appear able to compete with private schools suggests that parents who opt out of the public system, or at least consider opting out, are doing so primarily to improve the academic experiences of their children. There is nothing about public schools per se to which these parents object, and they will choose high-performing public schools when available. In fact, in New Orleans, exiting parents often show a preference for district-run schools, keeping other characteristics constant, over a larger selection of charter schools. Finally, because our evidence is drawn directly from parents’ rankings of schools, we are able to show (rather than simply infer or assume) that school differences observable to parents play a role in the choices they make. Such evidence suggests that parents do make use of publicly available information about potential schools and consider clear tradeoffs between public and private school alternatives when making their decisions.

Acknowledgments

This study was conducted at the Education Research Alliance for New Orleans at Tulane University with support from the Laura and John Arnold Foundation and William T. Grant Foundation. We thank Paula Arce-Trigatti, Nathan Barrett, Matthew Larsen, and Jon Valant for data analysis, and Douglas Harris, Betheny Gross, and many other colleagues for helpful feedback on prior drafts. Institutional support was provided by Tulane University's Murphy Center and School of Liberal Arts, Michigan State University, and the University of North Alabama. All errors are our own.

REFERENCES

Abdulkadiroğlu
,
Atila
,
Joshua D.
Angrist
,
Susan M.
Dynarski
,
Thomas J.
Kane
, and
Parag A.
Pathak
.
2011
.
Accountability and flexibility in public schools: Evidence from Boston's charters and pilots
.
Quarterly Journal of Economics
126
(
2
):
699
748
. doi:10.1093/qje/qjr017.
Abdulkadiroğlu
,
Atila
, and
Tayfun
Sönmez
.
1998
.
Random serial dictatorship and the core from random endowments in house allocation problems
.
Econometrica
66
(
3
):
689
701
. doi:10.2307/2998580.
Ahmed-Ullah
,
Noreen S.
,
John
Chase
, and
Bob
Secter
.
2013
.
CPS approves largest school closure in Chicago's history
.
Chicago Tribune
,
23 May
.
Angrist
,
Joshua
,
Eric
Bettinger
, and
Michael
Kremer
.
2006
.
Long-term educational consequences of secondary school vouchers: Evidence from administrative records in Colombia
.
American Economic Review
96
(
3
):
847
862
. doi:10.1257/aer.96.3.847.
Bettinger
,
Eric
, and
Bridget Terry
Long
.
2004
.
Shape up or ship out: The effects of remediation on students at four-year colleges
.
NBER Working Paper No. 10369
.
Betts
,
Julian R.
, and
Rober W.
Fairlie
.
2001
.
Explaining ethnic, racial, and immigrant differences in private school attendance
.
Journal of Urban Economics
50
(
1
):
26
51
. doi:10.1006/juec.2000.2207.
Betts
,
Julian R.
, and
Y. Emily
Tang
.
2014
.
A meta-analysis of the literature on the effect of charter schools on student achievement.
Available
www.crpe.org/sites/default/files/CRPE_meta-analysis_charter-schools-effect-student-achievement_workingpaper.pdf.
Accessed 1 June 2015
.
Booker
,
Kevin
,
Tim R.
Sass
,
Brian
Gill
, and
Ron
Zimmer
.
2011
.
The effects of charter high schools on educational attainment
.
Journal of Labor Economics
29
(
2
):
377
415
. doi:10.1086/658089.
Brown
,
Emma
.
2013
.
D.C. to close 15 underenrolled schools
.
Washington Post
,
17 January
.
Buddin
,
Richard J.
,
Joseph J.
Cordes
, and
Sheila Nataraj
Kirby
.
1998
.
School choice in California: Who chooses private schools
?
Journal of Urban Economics
44
(
1
):
110
134
. doi:10.1006/juec.1997.2063.
Bulkley
,
Katrina E.
,
Jeffrey R.
Henig
, and
Henry M.
Levin
.
2010
.
Between public and private: Politics, governance, and the new portfolio models for urban school reform
.
Cambridge, MA
:
Harvard Education Press
.
Butler
,
J. S.
,
Douglas A.
Carr
,
Eugenia F.
Toma
, and
Ron
Zimmer
.
2013
.
Choice in a world of new school types
.
Journal of Policy Analysis and Management
32
(
4
):
785
806
. doi:10.1002/pam.21711.
Campbell
,
David E.
,
Martin R.
West
, and
Paul E.
Peterson
.
2005
.
Participation in a national, means‐tested school voucher program
.
Journal of Policy Analysis and Management
24
(
3
):
523
541
. doi:10.1002/pam.20113.
Carlson
,
Deven
,
Joshua M.
Cowen
, and
David J.
Fleming
.
2013
.
Life after vouchers: What happens to students who leave private schools for the traditional public sector?
Educational Evaluation and Policy Analysis
35
(
2
):
179
199
. doi:10.3102/0162373712461852.
Center for Research on Education Outcomes (CREDO)
.
2014
.
National charter school study 2013
.
Available
https://credo.stanford.edu/documents/NCSS%202013%20Final%20Draft.pdf.
Accessed 10 June 2015
.
Chingos
,
Matthew M.
, and
Paul E.
Peterson
.
2015
.
Experimentally estimated impacts of school vouchers on college enrollment and degree attainment
.
Journal of Public Economics
122
:
1
12
. doi:10.1016/j.jpubeco.2014.11.013.
Chute
,
Eleanor
.
2014
.
Allegheny county school districts resize, close schools as population shifts
.
Pittsburgh Post-Gazette
,
31 August
.
Clark
,
Melissa A.
,
Philip M.
Gleason
,
Christina Clark
Tuttle
, and
Marsha K.
Silverberg
.
2015
.
Do charter schools improve student achievement
?
Educational Evaluation and Policy Analysis
37
(
4
):
419
436
. doi:10.3102/0162373714558292.
Cohen-Zada
,
Danny
.
2006
.
Preserving religious identity through education: Economic analysis and evidence from the U.S
.
Journal of Urban Economics
60
(
3
):
372
398
. doi:10.1016/j.jue.2006.04.007.
Cowen
,
Joshua
.
2010
.
Who chooses, who refuses? Learning more from students who decline private school vouchers
.
American Journal of Education
117
(
1
):
1
24
. doi:10.1086/656344.
Cowen
,
Joshua M.
,
David J.
Fleming
,
John F.
Witte
,
Patrick J.
Wolf
, and
Brian
Kisida
.
2013
.
School vouchers and student attainment: Evidence from a state‐mandated study of Milwaukee's parental choice program
.
Policy Studies Journal
41
(
1
):
147
168
. doi:10.1111/psj.12006.
Cowen
,
Joshua M.
, and
Eugenia F.
Toma
.
2015
.
Emerging alternatives to neighborhood-based public schooling
. In
Handbook of research in education finance and policy
, 2nd ed.,
edited by
Helen F.
Ladd
and
Margaret E.
Goertz
, pp.
468
488
.
New York
:
Routledge
.
Cullen
,
Julie Berry
,
Brian A.
Jacob
, and
Steven
Levitt
.
2006
.
The effect of school choice on participants: Evidence from randomized lotteries
.
Econometrica
74
(
5
):
1191
1230
. doi:10.1111/j.1468-0262.2006.00702.x.
Dobbie
,
Will
, and
Roland G.
Fryer
, Jr.
2011
.
Are high-quality schools enough to increase achievement among the poor? Evidence from the Harlem Children's Zone
.
American Economic Journal: Applied Economics
3
(
3
):
158
187
. doi:10.1257/app.3.3.158.
Dowdall
,
Emily
.
2011
.
Closing public schools in Philadelphia: Lessons from six urban districts
.
Philadelphia, PA
:
The Pew Charitable Trusts
.
Engberg
,
John
,
Dennis
Epple
,
Jason
Imbrogno
,
Holger
Sieg
, and
Ron
Zimmer
.
2014
.
Evaluating education programs that have lotteried admission and selective attrition
.
Journal of Labor Economics
32
(
1
):
27
63
. doi:10.1086/671797.
Epple
,
Dennis
, and
Richard E.
Romano
.
1998
.
Competition between private and public schools, vouchers, and peer-group effects
.
American Economic Review
88
(
1
):
33
62
.
Ferreyra
,
Maria Marta
.
2007
.
Estimating the effects of private school vouchers in multidistrict economies
.
American Economic Review
97
(
3
):
789
817
. doi:10.1257/aer.97.3.789.
Figlio
,
David
, and
Cassandra M. D.
Hart
.
2014
.
Competitive effects of means-tested school vouchers
.
American Economic Journal
6
(
1
):
133
156
.
Figlio
,
David
,
Cassandra M. D.
Hart
, and
Molly
Metzger
.
2010
.
Who uses a means-tested scholarship, and what do they choose?
Economics of Education Review
29
(
2
):
301
317
. doi:10.1016/j.econedurev.2009.08.002.
Figlio
,
David N.
, and
Joe A.
Stone
.
2010
.
Can public policy affect private school cream skimming
?
Journal of Urban Economics
49
(
2
):
240
266
. doi:10.1006/juec.2000.2189.
Fleming
,
David J.
,
Joshua M.
Cowen
,
John F.
Witte
, and
Patrick J.
Wolf
.
2015
.
Similar students, different choices: Who uses a school voucher in an otherwise similar population of students
?
Education and Urban Society
47
(
7
):
785
812
. doi:10.1177/0013124513511268.
Friedman
,
Milton
.
1997
.
Public schools: Make them private
.
Education Economics
5
(
3
):
341
344
. doi:10.1080/09645299700000026.
Glazerman
,
Steve
.
1998
.
School quality and social stratification: The determinants and consequences of parental school choice
.
Paper presented at the Annual Meeting of the American Educational Research Association
,
San Diego, CA, April
.
Green
,
Jay
,
Paul
Peterson
, and
Jiangtao
Du
.
1999
.
Effectiveness of school choice: The Milwaukee experiment
.
Education and Urban Society
31
(
2
):
190
213
. doi:10.1177/0013124599031002005.
Harris
,
Douglas N.
, and
Matthew
Larsen
.
2015
.
What schools do families want (and why)? Technical report
.
Available
http://educationresearchalliancenola.org/files/publications/Technical-Report-Final-Combined.pdf.
Accessed 10 June 2015
.
Hastings
,
Justine
,
Thomas J.
Kane
, and
Douglas
Staiger
.
2009
.
Heterogeneous preferences and the efficacy of public school choice
.
NBER Working Paper No. 2145
.
Howell
,
William G.
2004
.
Dynamic selection effects in means‐tested, urban school voucher programs
.
Journal of Policy Analysis and Management
23
(
2
):
225
250
. doi:10.1002/pam.20002.
Hoxby
,
Caroline M.
2003
.
School choice and school productivity: Could school choice be a tide that lifts all boats?
In
The economics of school choice
,
edited by
Caroline M.
Hoxby
, pp.
287
342
.
Chicago
:
University of Chicago Press
. doi:10.7208/chicago/9780226355344.003.0009.
Hurdle
,
John
.
2013
.
Philadelphia officials vote to close 23 schools
.
New York Times
,
7 March
.
Imberman
,
Scott A.
2011
.
Achievement and behavior in charter schools: Drawing a more complete picture
.
Review of Economics and Statistics
93
(
2
):
416
435
. doi:10.1162/REST_a_00077.
Kisida
,
Brian
,
Patrick J.
Wolf
, and
Evan
Rhinesmith
.
2015
.
Views from private schools: Attitudes about school choice programs in three states
.
Available
www.aei.org/wp-content/uploads/2015/01/Views-from-Private-Schools-7.pdf.
Accessed 10 June 2015
.
Lankford
,
Hamilton
, and
James
Wyckoff
.
1992
.
Primary and secondary school choice among public and religious alternatives
.
Economics of Education Review
11
(
4
):
317–337
. doi:10.1016/0272-7757(92)90040-A.
Lankford
,
Hamilton
, and
James
Wyckoff
.
2001
.
Who would be left behind by enhanced private school choice
?
Journal of Urban Economics
50
(
2
):
288
312
. doi:10.1006/juec.2001.2218.
Long
,
Bridget Terry
.
2004
.
How have college decisions changed over time? An application of the conditional logistic choice model
.
Journal of Econometrics
121
(
1
):
271
296
. doi:10.1016/j.jeconom.2003.10.004.
Long
,
James E.
, and
Eugenia F.
Toma
.
1988
.
The determinants of private school attendance,
1970
1980
.
Review of Economics and Statistics
70
(
2
):
351
357
. doi:10.2307/1928323.
McFadden
,
Daniel
.
1974
.
Conditional logit analysis of qualitative choice behavior
. In
Frontiers in econometrics
,
edited by
Paul
Zarembka
, pp.
105
142
.
New York
:
Academic Press
.
McMillan
,
Robert
.
2004
.
Competition, incentives, and public school productivity
.
Journal of Public Economics
88
(
9–10
):
1871
1892
. doi:10.1016/j.jpubeco.2003.06.003.
National Center for Education Statistics (NCES)
.
2013
.
Digest of education statistics, 2012
.
Washington, DC
:
U.S. Department of Education, NCES
.
Rouse
,
Cecilia Elena
.
1998
.
Private school vouchers and student achievement: An evaluation of the Milwaukee parental choice program
.
Quarterly Journal of Economics
113
(
2
):
553
602
. doi:10.1162/003355398555685.
Saulny
,
Susan
.
2010
.
Board's decision to close 28 Kansas City schools follows years of inaction
.
New York Times
,
11 March
.
Schneider
,
Mark
,
Melissa
Marschall
,
Paul
Teske
, and
Christine
Roch
.
1998
.
School choice and culture wars in the classroom: What different parents seek from education
.
Social Science Quarterly
79
(
3
):
489
501
.
Trivitt
,
Julie R.
, and
Patrick J.
Wolf
.
2011
.
School choice and the branding of Catholic schools
.
Education Finance and Policy
6
(
2
):
202
245
.
Wolf
,
Patrick J.
,
Brian
Kisida
,
Babette
Gutmann
,
Michael
Puma
,
Nada
Eissa
, and
Lou
Rizzo
.
2013
.
School vouchers and student outcomes: Experimental evidence from Washington, DC
.
Journal of Policy Analysis and Management
32
(
2
):
246
270
. doi:10.1002/pam.21691.

Notes

1. 

As discussed by Saulny (2010), Dowdall (2011), Ahmed-Ullah, Chase, and Secter (2013), Brown (2013), Hurdle (2013), and Chute (2014).

2. 

Throughout this paper, we refer to traditional public schools and publicly funded charter schools collectively as “public schools.” We refer to tuition-funded schools in general as “private schools,” and to those that participate in the LSP as “voucher schools.”

3. 

Author calculations from the 2012 NCES Private Schools Universe Survey (see https://nces.ed.gov/surveys/pss).

4. 

In the assignment process, students exiting D or F schools receive a higher priority for winning a voucher than students exiting a C school.

5. 

Efforts are made to predetermine eligibility before voucher applications are submitted, although formal documentation of income eligibility is not collected until enrollment. In rare cases, parents cannot provide the required documentation and must withdraw after applying.

6. 

The subset of schools that do not participate in the OneApp include selective admissions charter schools, language immersion schools that require a fluency exam, and a specialized performing arts school. The Louisiana RSD does not allow charter schools to use selective admissions schools, and no OPSB direct-run schools are selective. All selective admissions schools in New Orleans are charter schools under the authority of either the local school board (OPSB) or the state Board of Elementary and Secondary Education.

7. 

To ensure that all current students are reassigned for the next school year, the OneApp process automatically creates an application for all current enrollees that ranks only the current school. These students receive a guaranteed seat at their current school. Parents considering exiting their current school would complete OneApp and rank at least one other school.

8. 

Because of the limited availability of vouchers, parents were encouraged by RSD to apply for at least one public school. Approximately 26 percent of students in the private strategy group were already voucher recipients who were guaranteed a seat at their current private school. The other 74 percent had no guaranteed school to attend, and if they failed to win a voucher, they would be required to participate in a subsequent OneApp round for assignment. Subsequent rounds provide considerably fewer public school options and no private school options. We cannot observe whether these students have access to other ways to finance private school enrollment, such as private aid or scholarships, if they fail to win a voucher.

9. 

The national Private Schools Universe Survey reports that in 2012 approximately 15,000 students attended 46 private schools in Orleans Parish. This includes over two dozen private schools that do not participate in the voucher program. The voucher rate is set quite low, and nonparticipating schools tend to charge higher tuition than participating schools.

10. 

There are 22 schools that did not participate in OneApp for 2013–14. They enrolled 26 percent of students overall and 19 percent in grades 1–8.

11. 

This includes kindergarteners who did not attend a public school pre-K program, students whose current school does not include their upcoming grade, and students who are entirely new to the system. This subset includes 248 students in the mixed strategy.

12. 

In addition to grade-level restrictions, the choice sets are limited by two single-sex voucher schools (one for boys and one for girls) that participated in the voucher program.

13. 

Families are screened for eligibility at the time of application. Verification is not required until registration for students who win a voucher, but based on conversations with RSD staff, it is extremely rare for a family to fail to verify eligibility after the initial screening.

14. 

Publicly available at www.louisianabelieves.com/data/reportcards.

15. 

For this reason, the SCI may be biased compared with SPS. It will be biased in a positive direction if voucher students at a given private school are less disadvantaged than the typical New Orleans public school student, but in a negative direction if voucher students at a given private school are more disadvantaged than the typical public school student. We do not have sufficient information to determine the likely size of direction of bias due to selection into voucher schools. Conditional on being enrolled and tested, the two scores are mathematically comparable measures of test proficiency rates. Measured on a scale of 0–150, schools achieving 100 points or above receive an A grade, which is roughly equivalent to having 100 percent of students reach basic proficiency, with extra points available for students who achieve higher proficiency levels.

16. 

The SCI is reported in a separate annual report on the voucher program published by LDOE. Additionally, schools might not have a publicly reported score for several reasons. For public schools, scores are not reported when campuses are new or in transition (for example, if a charter school has been transferred to a new operator). This is fairly common in New Orleans’s largely charter school system, and in our 2012 data, eight charter schools have a reported grade of “T” (for transition). For private schools, publication of scores depends on the number of tested children. If fewer than ten children are tested on a campus, scores are suppressed for student privacy reasons, so private schools with no published score are typically those that enroll only a small number of students in tested grades.

17. 

The other religious category includes several Protestant Christian denominations and one Jewish day school. The secular category includes a language immersion school, an early childhood/kindergarten center, and a college prep academy.

18. 

Schools completed an annual survey covering topics such as instructional focus, extracurricular activities, transportation, discipline policies, and school hours, and this information was validated and summarized by Parents’ Guide staff. The guide was provided to parents online, and hard copies were freely distributed at school choice fairs and OneApp enrollment centers. It is publicly available at www.neworleansparentsguide.org.

19. 

New schools opening in 2013–14 have missing student demographics and school performance data for the prior year.

20. 

Schoolwide math value-added measures, or VAMs, were also included in A in some specifications. They were never statistically significant and did not substantially change the coefficients on SPS/SCI, so we do not include them in our results. It should be noted that parents do not have access to any publicly available VAMs for these schools.

21. 

Note that this framework assumes independence of irrelevant alternatives, which, if appropriate, implies that the estimates in equation 1 are consistent even if the decision to opt out of a school graded below a C (i.e., fill out a OneApp) is endogenous (Long 2004). Because we are essentially estimating equation 1 across a set of student–school pairwise combinations in the data but within the set of available options to each student, we are not able to include individual student or family characteristics other than distance, which varies within each student record according to the distance between each address and school j.

22. 

Tables 510 present results as coefficient estimates from different specifications of equations 2 through 4. Equivalent full tables providing log odds are available upon request from the authors.

23. 

We cannot reliably estimate exiters from grades 9–12 as a subgroup because of the limited number of observations.

24. 

Because we have access to student addresses, we also estimated effects based on the median income of the student's census block. Although all voucher applicants have income below 250 percent of the poverty line (∼$49,000 for a family of four), they vary in the quality of neighborhood residence, as evidenced by median census block income. Results by income block tercile (not shown) suggest that parents from all income blocks prefer religious private schools and OPSB-run schools to the typical New Orleans charter school.

25. 

Harris and Larsen (2015) look at the full sample of OneApps, including students who opt to remain in the same school, and find that sports and after-school care do influence relative school rankings for this population. By comparison to our results, the subset of parents in the mixed strategy appear to be more sensitive to academics and less to extracurricular programs than the average New Orleans parent.