Housing and Education Research
Search for research on the intersection of housing and education policy. Read jargon-less summaries of key findings and interesting facts, or read the full PDF and grab the citation for your own writing.
Housing Policy is School Policy
858 elementary school students in Montgomery County, Maryland who lived in public housing from 2001 to 2007.
- After 5 to 7 years, public housing students assigned to advantaged schools (as measured by free and reduced price lunch or other measures) performed better on math and reading standardized tests than public housing students attending less-advantaged schools. Public housing students attending advantaged schools cut the achievement gap in half in math and by 1/3 in reading, compared to non-poor students in Montgomery County.
- This "advantaged school" affect was strongest for public housing students attending schools where 20% or less of the population qualified for free or reduced price lunch. Once the school reached 35%, public housing students performed just as well as schools where 35% to 85% of the students qualified for free or reduced price lunch.
- Schwartz estimates that about 2/3 of the improvement of public housing children in advantaged areas was as a result of the school, and about 1/3 a result of the neighborhood.
- Students assigned to more advantaged schools performed significantly better than their peers in less-advantaged schools, even though less-advantaged schools had received increased investment because of their disadvantaged status and investments in less-advantaged schools produced statistically significant improvements in (some) test scores.
- Families who had elementary school children in the study stayed in the same unit for an average of 8 years, longer than is typically of public housing families. Montgomery County public housing families also earned about 25% more, on average, than public housing households nationally.
- Montgomery County has few schools above the 60% free and reduced price lunch threshold, so this study cannot determine whether students attending schools where 35% to 85% of the student body qualify for free or reduced price lunch perform better than students attending schools with an 85% rate or above.
- Montgomery County has a median income ($93,895 in 2008) 80% higher than the national median.
Schwartz, Heather. Housing Policy Is School Policy: Economically Integrative Housing Promotes Academic Success in Montgomery County, Maryland. Rep. New York: Century Foundation, 2010.
Changing the Geography of Opportunity by Expanding Residential Choice
One 8- to 18-year-old child was randomly selected from 118 families participating in the Gautreaux housing mobility program in 1982, with follow-up in 1989 with a 59.1% response rate.
- Compared to city movers, children of families who moved to the suburbs were less likely to drop out of school (20% to 5%) and be paid under $3.50 and hour (43% to 9%) and more likely to be on track for college (24% to 40%), attend college (21% to 54%), attend a four-year college (4% to 27%), be employed full-time if not in college (41% to 75%), be paid over $6.50 an hour (5% to 21%), and have job benefits (23% to 55%).
- At baseline in 1982, families that moved to the suburbs were more satisfied with teachers and courses at the new schools, and reported smaller class sizes than city movers.
- Survey results indicated that suburban movers were just as likely to be accepted by their peers as city movers, and had almost as many friends, though suburban movers spent more time with whites outside of class than did city movers.
Rosenbaum, James E. "Changing the Geography of Opportunity by Expanding Residential Choice: Lessons from the Gautreaux Program." Housing Policy Debate 6.1 (1995): 231-69.
Moving to Opportunity for Fair Housing Demonstration Program: Final Impacts Evaluation
From 1994 to 1998, 4,604 public housing households were enrolled in five cities - Baltimore, Boston, Chicago, Los Angeles, and New York and assigned to the experimental, section 8 only, or control groups. Final follow-up occurred 10 to 15 years after initial enrollment, where the response rate was 90% for household heads and 89% for youth.
- At final follow-up, found no impacts on reading or math scores for youth ages 13 to 20, even for children less than 6 years old at the time of enrollment. About 1/4 of youth ages 15 to 20 were enrolled in a postsecondary institution across all groups.
- At final follow-up, youth in all three groups had similar levels of tardiness, absenteeism, and behavior problems, though males in the experimental and section 8 group were more likely to be suspended or expelled.
- On average, students at baseline attended schools at the 15th percentile on state assessments. At final follow-up, the average for all three groups (control, section 8, and experimental) was between the 19th and 22nd percentiles. Experimental group students had slightly better impressions of their new schools, as measured by a school climate index.
- The authors conclude that MTO demonstrates the impact neighborhood change can have without changing school quality much.
Sanbonmatsu, Lisa, Jens Ludwig, Lawrence F. Katz, Lisa A. Gennetian, Greg J. Duncan, Ronald C. Kessler, Emma Adam, Thomas W. McDade, and Stacy T. Lindau. Moving to Opportunity for Fair Housing Demonstration Program: Final Impacts Evaluation. Rep. Washington, D.C.: U.S. Department of Housing and Urban Development, 2011.
Does Moving to Better Neighborhoods Lead to Better Schooling Opportunities?
The mothers and children that participated in the Baltimore site of the Moving to Opportunity experiment (recruited from public housing between 1994 and 1998). For the quantitative section, the study uses data from 249 children in 2001. The qualitative section uses a stratified random subsample of 149 families (55 interviews with control-group families and 35 with experimental families that relocated) in July 2003 and June 2004.
- Baltimore experimental students in MTO voucher households attended schools at the 27th and 25th percentile in state reading and math tests after their first move, compared to schools at the 14th percentile pre-move. However, many experimental children still attended schools in the inner city.
- About 70% of experimental children attended high-poverty (>80% FARM) schools at baseline, but just 16% did after first move. And while 7% attended schools with test scores above the state median, 17% did after the first move. There was also a large reduction in students attending schools below the 10th percentile in state test scores.
- Almost no families moved to school catchment areas with schools with high test scores (over one standard deviation above the mean for the states). Most moved to areas within one standard deviation of the mean. The school zones they moved to averaged 45% of students qualifying for free and reduced price lunch.
- Why did MTO vouchers not lead to greater changes in schools? Some mothers felt a change would be traumatic (33% of all mothers mentioned this), others felt they lacked the necessary information (38% of experimental mothers), and many had low expectations for schools in general (46% of experimental mothers).
- Many parents believed that school characteristics were not important, emphasizing hard work and good attitude as necessary attributes to succeed, no matter the school (about 67% of all mothers believed in "the accomplishment of natural growth").
- For about 70% of the mothers in the sample, what makes a good school has a lot to do with distance to work, the existence of uniforms and school discipline, and if the teachers care about the children.
- Many of the interviewees suffered with what the authors term the "context of chaos", including "needy extended family members", difficulties with odd hours and other issues with low-wage work, and landlord problems that could force a tenant into an involuntary or hasty move.
Deluca, Stefanie, and Peter Rosenblatt. "Does Moving to Better Neighborhoods Lead to Better Schooling Opportunities? Parental School Choice in an Experimental Housing Voucher Program." Teachers College Record 112.5 (2010): 1443-491.
Why Poor People Move (and Where They Go)
Semi-structured interviews with a stratified sample of 140 low-income African-American parents earning less than 50% AMI with at least one child under 18 in Baltimore, MD and Mobile, AL in 2010.
- About 70% of the sample described their last move as out of their control, or a reactive move. More than 80% of the sample had at least one reactive move at some time. For the 30% that decided to move, they made the decision because of family issues, safety, or housing costs.
- The most prevalent cause of reactive moves was "unit failure", or dwelling-related issues, which often lead to problems with vermin and caused the unit to fail the HUD Section 8 inspection. While public housing demolition was another factor in reactive moves, another was receipt of public housing assistance, which almost always lead to a move from the residence. Another major factor in reactive moves was when landlords sold the property to new owners.
- The most prevalent cause of voluntary moves was to leave family and friends and form a new household, commonly by young mothers after the birth of a child.
- Families used a number of strategies to find housing. The most common were using kin networks and choosing the first available from the Section 8 list or rental signs, revealing expediency as a very important factor in many moves. Preventing homelessness is a key concern for some families searching for units in a restricted time span.
- Families didn't really choose neighborhoods as much as they chose housing units, with many parents able to detail specific features they want in a property without care for the neighborhood as a whole. Families preferred to live in a single-family detached home, and looked for affordable units with desired features like a basement, backyard, and the necessary number of bathrooms. Putting up with bad neighborhoods in order to secure a desired unit was common.
- When families did assess neighborhoods, they looked primarily at safety, and defined it for the most part as the absence of large housing projects. Many respondents only required that neighborhood safety meet a minimum acceptable threshold.
- Families rarely mentioned racial criteria as a reason for choosing a specific neighborhood, and when probed, said that race didn't matter.
Deluca, Stefanie, Peter Rosenblatt, and Holly Wood. Why Poor People Move (and Where They Go): Residential Mobility, Selection and Stratification. Rep. New York: NYU, 2013. DRAFT: PLEASE DO NOT CITE WITHOUT PERMISSION
Public Schools, Public Housing
The public housing data consists of 286 non-senior public housing developments with 169,105 units in the New York City Housing Authority in 2008. School data consists of 736,274 public elementary and middle school students in 2002-03, 84,526 of whom were matched to a non-senior NYCHA development.
- NYCHA students perform a little more than 0.3 standard deviations below the citywide mean on fifth grade math and reading assessments, while non-NYCHA students perform 0.06 standard deviations better than the average.
- One half of all public housing students are concentrated in 10% of the schools, and 2/3 are concentrated in just 15% of all schools.
- In a typical school attended by public housing students, 38% of kids live in a public housing development and 85% qualify for free or reduced price lunch, compared to 8% and 70% in the typical school attended by non-public housing students. While attendance rates are similar in both groups, about 40% of students pass reading and math tests in a typical school for a public housing student, while about 50% pass in the typical school attended by other students.
- Teachers in the typical school attended by NYCHA students have only marginally less education and experience than the typical school attended by other students, and the typical school attended by NYCHA students receive approximately 12% greater per pupil expenditures and a slightly smaller student teacher ratio.
- Even controlling for observable student demographics and school fixed effects, the authors are unable to explain about a standard deviation of difference between NYCHA and non-NYCHA students, which they suggest is a result of unobservable individual-level or family characteristics of public housing families. Schools, do appear to play a role, though funneling more resources does not appear to be enough to close the achievement gap.
- NYCHA public housing holds approximately 5% of all New York City residents. 31% are younger than 18. The average public elementary and middle school in the city has about 14% of its students living in public housing.
- In the average New York City public elementary and middle school, 15% of students are White, 35% are Black, 38% are Hispanic, and 74% qualify for free or reduced price meals.
Schwartz, Amy E., Brian J. McCabe, Ingrid G. Ellen, and Colin C. Chellman. "Public Schools, Public Housing: The Education of Children Living in Public Housing." Urban Affairs Review 46.1 (2010): 68-89.
Do Federally Assisted Households Have Access to High Performing Public Schools?
Datasets used include national data on subsidized housing tenants in 2008, Low Income Housing Tax Credit data up to 2009, Department of Education proficiency scores in Math and English for all public school students, and the Common Core of Data (public schools), presumably from the 2008-09 school year. This study focuses on elementary schools.
- Public elementary schools closest to public housing tenants have a median test score ranking in the 19th percentile, compared to the 26th percentile for Housing Choice Voucher (HCV) households, the 28th percentle for Project-based Section 8 households, the 31st percentile for Low-Income Housing Tax Credit (LIHTC) households, the 30th percentile for all poor households, and the 37th percentile for all renters.
- Similar patterns emerge when looking at the share of households with children living nearest an elementary school ranked in the bottom 10th percentile and in the top 50th percentile. 32.5% of public housing households live nearest a bottom 10th percentile school, while only 23.2% of LIHTC households do (compared to 21.6% of poor households). And while 19.4% of public housing households live nearest a top 50th percentile school, 33% of LIHTC households do (compared to 31.6% of poor households). On both measures, Project-Based Section 8 and HCV households fall somewhere in between.
- The same pattern emerges when using poverty rate as a school quality measure. 53.3% of public housing households live nearest a school where over 80% of students qualify for Free and Reduced price Lunch (FRPL), compared to 34.1% of LIHTC households (and 40.6% of poor households). While 5.7% of public housing households live nearest a school where less than 20% of students qualify for FRPL, 10.3% of LIHTC households do (compared to 10.2% of poor households). On both measures, Project-Based Section 8 and HCV households fall somewhere in between.
- There are large racial disparities in these results. Looking at the difference in locational outcomes among voucher holders, who can choose their own housing to a certain degree, the authors find a 20 percentile point gap in the ranking of the median school nearest white and black voucher holders, and 15 percentile point gap between white and Hispanic voucher holders (white being higher in both cases).
- The same is true when looking at FRPL eligibility. In the median school nearest to white voucher holders, 57% of students qualified for FRPL, compared to 81% and 80% at the nearest school to black and Hispanic voucher holders, respectively. Both gaps are smaller than the racial gaps for the population as a whole.
- One of four households eligible to receive a housing subsidy from the federal government actually receives one.
- 1.2 million households live in public housing, 360,000 of which have children under 18. 1.5 million households live in private units subsidized by many federal programs, the largest of which is the project-based section 8 program, with 1.2 million households, 400,000 of which have children under 18. The largest program is the LIHTC program, which houses about 2.5 million households. This study estimates that about 900,000 of these have children under 18. Housing Choice Vouchers serve over 2 million households, 1.2 million of which have children under 18.
Ellen, Ingrid G., and Keren M. Horn. Do Federally Assisted Households Have Access to High Performing Public Schools? Rep. Washington, D.C.: PRRAC, 2012.
Welcome to the Neighborhood?
Youth ages 8 to 18 (1 to 11 at relocation) in 189 families in public housing or on the waiting list in Yonkers, NY randomly selected to new "dispersed" public housing in middle-income neighborhoods between 1992 and 1994. These "movers" were compared to 145 comparable families eligible to move but who did not (the authors did not have access to the waiting list - only half had actually entered the lottery), with interviews 2 (315 Black and Latino families) and 7 (247 of these families) years after relocation. In both groups, a total of 221 youth were interviewed at the 7 year follow up (a 22% loss between the interviews).
- Movers reported statistically significant lower math and reading test scores and lower levels of educational engagement than stayers, however the actual test scores of both groups were not significantly different, indicating movers perceived their performance differently than stayers.
- Movers reported experiencing a greater incidence of anxious and depressed symptoms, partially as a result of fewer episodes of informal contact in their new neighborhoods.
- Because of school choice in Yonkers, both the movers and stayers attended about the same schools, as measured by English and math test scores, school types (themes), share minority, and share eligible for free lunch. In both groups, most attended public schools outside of the students' neighborhoods (72%). However, mover neighborhoods had higher median family income, a lower share of minority residents, lower poverty rates, and a higher proportion of units that were single family homes.
- Parents in mover families reported monitoring their children less than stayers, which seems to have mediated increased substance abuse in the 15 to 18 year-old age group.
Fauth, Rebecca C., Tama Leventhal, and Jeanne Brooks-Gunn. "Welcome to the Neighborhood? Long-Term Impacts of Moving to Low-Poverty Neighborhoods on Poor Children's and Adolescents' Outcomes." Journal of Research on Adolescence 17.2 (2007): 249-84.
The Impact of School Choice on Student Outcomes
Follows three groups of 9th graders in Chicago Public Schools in the fall of 1993, 1994, and 1995, for a total of 76,563 students. Because of missing 8th grade test scores, missing subsequent outcomes, and other missing data, the final total is 60,623.
- Controlling for a number of variables, the authors find that opting out of an assigned school is associated with a higher rate of 10th grade completion (a 6.5% increase), 11th grade completion (8% increase) and graduating on time (8% increase). This implies that most of the effect occurs by 10th grade. Because this does not reconcile with plausible models of choice, the authors point to this (and other findings) as indicating that this association may be spurious.
- The gains from opting out were slightly higher for top quartile students (8 percentage points) than bottom quartile students (6 percentage points), but the gains were highest in all ability groups if the student opted to attend a high-acheiving school (12 to 15 percentage points). Again, the authors posit that these associations are likely spurious.
- The authors find that students who opt out differ significantly from those who don't (a selection effect). Using an instrumental variable approach (distance to alternative schools), the authors find that only career academies seem to provide true improvement in graduation rates - opting out to other schools only seems to improve outcomes through selection effects.
- This cohort tests on average 1 grade level below the national average in math and reading. More than 60% the the students are black and over 25% are Hispanic. 25% drop out before the end of 10th grade and half graduate in 4 years.
- The lowest quartile of students perform at about the 6th grade level in math and 5th grade level in reading, and only about 1/3 of these students graduate, as opposed to about 70% of top quartile students. Top quartile students live in areas with "higher achieving peers, lower poverty rates, fewer female-headed households, greater home ownership, and lower crime rates."
- Less than 50% of students enroll in their assigned school - this figure is 26% for top quartile students and 63% for bottom quartile students. Of those who opt out, about 16% choose to attend a high-achieving school (top quintile of schools by 8th grade test scores of incoming students) - this figure is 47% for top quartile students and 2% for bottom quartile students.
Cullen, Julie Berry, Brian A. Jacob, and Steven D. Levitt. "The Impact of School Choice on Student Outcomes: An Analysis of the Chicago Public Schools." Journal of Public Economics 89.5-6 (2005): 729-60.
Opening Doors: How Low-Income Parents Search for the Right School
Targeted telephone interviews of 800 low-income parents that had recently picked a school for their child in Washington, D.C., Milwaukee, and Denver in late fall of 2005.
- Across the three cities, about 43% of parents found out they had a choice through friends or other parents (22%) or school personnel (21%). 14% found out from school letters or other written materials. Parents do not rely on one source of information, but rather work hard to get all the information they need - 46% talked to five or more people, excluding their immediate family, while only 20% talked to no one outside the family.
- When forced to choose their best source of information, parents chose talking with people (friends, parents and school personnel) over written material by 2 to 1. When forced to choose the best source of information from people, respondent chose other parents over teachers by about 2 to 1.
- 45% of parents cited academic quality as the most important factor in choosing schools, followed by curriculum or thematic focus (19%) and location and convenience (11%). When pressed on how important location was from 1 to 5, 14% said it was the most important factor, 22% very important, 27% somewhat important, 10% slightly important, and 26% not important at all. The level of importance declines for children in higher grades.
- Parents chose among a small number of schools. Half considered only two schools and applied to just one, while the other half considered at least three schools and applied to at least two (though few considered four or more and applied to three or more).
- Parents gather information from a large number of sources. Of a list of 10 information gathering activities, 77% of parents engage in at least 5, and 10% engage in at least 9. The most common are visiting the school (85%), talking to administrators (77%), talking to teachers (76%), having the child visit the school (74%), seeing printed information (73%), and talking to family or friends (68%). When pressed on the "single most important source", 38% said teachers or administrators and 34% said family or friends. Only 17% of parents felt they were missing important information.
- Older children were much more likely to be involved in the school choice process than younger students. Almost all parents with children in 8th grade involved their child, while only about 1/3 of parents with children in grades K to 4 involved their children in the decision. Parents involving their children were more likely to consider more schools and participate in more information gathering activities.
- When trying to match their child to a good school, 40% of parents tried to match their children to gifted programs, while 18% cited positive social issues (good peers) and 14% cited negative social issues (bad peers). About 88% of parents were "very" or "somewhat" satisfied with their choice. There was a statistically significant relationship between the number of information gathering activities and satisfaction.
- Parental income was roughly evenly distributed between $0 and $50,000, though 90% of parents were women (about 2/3 of the sample were single mothers - the remainder were the primary school decisionmakers). Parents were asked about their "most recent choice" - most choices were found at Kindergarten and 1st grade (33% of the sample), suggesting interest in selecting a school is highest at the elementary school level.
- Of the total sample records that were not disconnected or establishments, the survey team could not make human contact with 22.4% of households in Denver, 17.4% in D.C., and 3.9% in Milwaukee.
- Of those the survey was able to contact, 70% of parents considered a school other than their child's zoned school in Milwaukee, compared to 73% in D.C. and 56% in Denver. Children of respondents were much more likely to be in private school - 26% of respondents in Milwaukee, compared to 20% in D.C. and 14% in Denver. A large portion — 37% in Milwaukee, 44% in D.C., and 45% in Denver — attended the "closest public school".
Teske, Paul, Jody Fitzpatrick, and Gabriel Kaplan. Opening Doors: How Low-Income Parents Search for the Right School. Rep. Seattle, WA: Center on Reinventing Public Education, 2007.
All Choices Created Equal?
Three interviews of 48 parents of children before and after they began attending sixth or ninth grade from January to November, 2003. The metro area is presented as a typical metro in the Midwest with abundant school choice, though the author does not name it due to confidentiality concerns. Potential participants were purposely sampled from failing and non-failing schools of different school types (public, public charter, etc.). Using lists of fifth and eighth grade students, participants were randomly selected from three income categories and contacted by telephone. The positive response rate was 60%, and parents with similar characteristics replaced parents that declined.
- 40 of the 48 parents had previously chosen for their child to attend a school other than their neighborhood school, however many fewer parents opted out of the customary feeder patterns.
- 15 parents chose not to conduct a search, 9 of whom said there were no other alternatives to the neighborhood school that "offered what they wanted". These parents were not ignorant — they had information from previous searches with older children. Some parents thought the neighborhood school was a good one or were at least comfortable with it, and would reassess their decision the following year. Social class did not appear to be related to choice — 61% of middle-class parents and 75% of poor and working-class parents conducted a search.
- The author finds parents made 2 types of searches. 22 parents conducted an open search, in which they began with a large set of schools (7.5 on average) and narrowed down the selection over time through "interaction in the schooling market". 11 parents conducted a closed search, in which the parent began a small number of schools in mind (3.4 on average) and made preparations for the child to attend these schools (i.e., exams, finances, and transportation arrangement).
- In selecting their child's final school, 69% of parents cited factors concerning the child's overall well-being (e.g., "thriving where they are") and 58% cited academic reasons (e.g., "good teachers"), while only 33% cited social reasons, 27% logistical reasons, and 25% administrative reasons.
- The share of selective, non-failing, and tuition-based schools in middle-class parents' choice sets were higher than those in poor and working-class parents' choice sets. 58% of middle-class parents had at least two non-failing schools in their choice sets, compared to 16% for poor and working-class parents. 53% of middle-class parents chose a non-failing school, compared to 36% of poor and working-class parents. Middle class parents were also more likely to choose selective and tuition-based schools.
- All but 3 parents used social networks to learn about schools, but middle-class parents' networks lead them to more non-failing, selective, and tuition-based schools. Though parents could select any school, schools in the customary feeder pattern for their child made up 49% of schools in parents' choice sets, a pattern that held across social classes.
- Parents with children with weak academic histories tended to want to avoid setting their children up for failure by sending them to "challenging" schools, indicating the child's academic background played an important role in the decision-making process.
- 45 of the 48 parents were mothers, 67% were African American, 27% were White, and 4% were Hispanic, similar to the racial profile of the city.
- Many parents said they wanted more diverse schools, but in practice the schools in parents' choice sets had very little variation in racial composition, suggesting "race issues played out in more subtle ways".
Bell, Courtney A. "All Choices Created Equal? The Role of Choice Sets in the Selection of Schools." Peabody Journal of Education 84.2 (2009): 191-208.
Parental Preferences and School Competition
Parents in Charlotte-MecklenBurg School District (CMS), North Carolina, in 2002, after school choice was implemented. Parents submitted the top three choices for schools for their children - about 95% (105,000 of 110,000 students) of children submitted choices. This study narrows that population to parents of students entering grades 4 through 8. In addition, CMS drew new school boundaries, forcing about half of parents to be assigned to a new school (43% elementary, 52% middle and 35% high school), allowing the researchers to differentiate a desire for proximity to the school from other, unobservable, neighborhood characteristics.
- Higher-income children (not qualifying for Free and Reduced Price Lunch (FRL)) had a substantially higher preference for test scores (measure by a school's mean test score) than low-income students. Additionally, the authors found that the higher the student's academic ability and neighborhood income, the more they preferred schools with high test scores. After controlling for racial school preferences, Blacks and Whites had similar preferences for school test scores.
- Parents placed a high value on proximity, and proximity was strongly negatively correlated with parents' test score preferences. This remains true when only analyzing the student's who were reassigned to a new neighborhood school as a result of redistricting.
- Demand is much more elastic (more responsive to small changes in test scores) at high-performing schools than low-performing schools. The students responding to these test score changes are much more likely to have higher academic performance and be higher income (not on FRL).
- Of the total school population, 43% were White and about the same share were Black, though 10% of White students receive FRL, compared to more than 60% of Blacks. Black or FRL students were much more likely to fill out all three choices (66% of Black FRL students compared to 29% of White non-FRL students), because White non-FRL students were more likely to have a guaranteed neighborhood school with higher test scores and because longer distances in suburban areas may have left limited choices.
- The final sample of students in fourth through eighth grade consisted of 36,816 students - 12,755 with only a first choice specified, 6,701 with two choices, and 17,360 with three choices. Those applying to specialized programs (e.g., disability, LEP) were removed from the analysis.
- There was little consensus among parents as to a top school or schools. Parents with kids in 2nd through 5th grade chose 93 different schools as a first choice, and no choice accounted for more than 2.7% of all parents. Some of this can be explained by travel time to school, but other choices seem to represent heterogeneity in preferences. After controlling for previous school assignments, elementary school boundaries had 10.4 different first choice schools on average.
- Measured by travel distance to a top-quartile school (as measured by test scores), Black, White, FRL, and non-FRL students have about the same travel distance to the closest top-quartile school, just over 2 miles. Within the FRL and non-FRL categories, Blacks had lower average test scores and neighborhood incomes than Whites.
Hastings, Justine S., Thomas J. Kane, and Douglas O. Staiger. Parental Preferences and School Competition: Evidence From a Public School Choice Program. Working paper no. 11805. Cambridge, MA: National Bureau of Economic Research, 2005.
Information, School Choice, and Academic Achievement
Parents of children in the Charlotte-Mecklenburg Public School District (CMS) in NCLB-sanctioned schools (16 schools with 6,695 students) who had already made school choice selections, were re-sent choice forms with a simplified 3-page listing of schools with test scores in the summer of 2004. Additionally, parents at NCLB and non-NCLB schools in CMS in 2006 were sent a one-page sheet of schools, with either test scores or odds of admission (randomized) as part of the school choice process — non-NCLB schools were randomly selected, while NCLB schools were required to send out information to all students. This second experiment included 6,328 non-NCLB students and 10,134 NCLB students.
- Compared to controls, both experiments caused the share of parents choosing "non-guaranteed" schools to increase by 5 to 7 percentage points. The schools chosen had test scores 0.05 to 0.1 standard deviations higher than those chosen by controls. The one-page format did not perform significantly better than the three-page NCLB format.
- 16% of parents in 2004 chose a different school for their child with test scores at least 0.5 standard deviations above the schools they had chosen earlier that year. Proximity to these schools was a key indicator in parents selecting these higher performing schools.
- Using an instrumental variable approach, the authors find that choosing to a attend a school with test scores one standard deviation higher than average leads to a 0.37 to 0.41 standard deviation improvement in that student's test scores, on average.
- Families in both experiments were more likely to be Black, qualify for Free and Reduced Price Lunch (FRL), and have lower test scores and lower income levels.
- In the 2004 NCLB experiment, 1,092 of 6,695 parents responded and chose a different school than their current NCLB school. While 11.2% (about 750) of those parents had already chosen an alternative school in the choice process earlier that spring, the NCLB information increased that figure to 16.3% (342 parents, 5.1 percentage points), and increased the average test score from 0.053 standard deviations above average to 0.1.
- Almost all of the parents who chose new schools chose schools with higher test scores than their first choice schools, with these choosers selecting schools with 0.62 standard deviations above their previous NCLB schools on average. Most of this comes from parents acting on new information — if parents randomly selected a non-NCLB school within 5 miles, the average increase would only have been 0.075 standard deviations higher than their first choice.
- In the 2004 NCLB experiment, key predictors of choosing a new, non-NCLB school were the (higher) average test scores of schools within 5 miles, if the parent had a single child in CMS, if the student was entering Kindergarten or 6th grade in the fall, if the student was Black, and if the parent had already chosen a non-NCLB school earlier in the spring. Key indicators of whether the child went to a higher performing school were the average test score of schools within 5 miles, if the child was not Black or FRL, and if the parent had already chosen a non-NCLB school earlier in the spring.
- In the second experiment, compared to parents in the control group, many parents receiving information picked fewer schools with lower test scores than their neighborhood school. Receiving information also doubled the number of parents choosing schools with test scores one or more standard deviations above their neighborhood school.
Hastings, Justine S., and Jeffrey M. Weinstein. Information, School Choice, and Academic Achievement: Evidence from Two Experiments. Working paper no. 13623. Cambridge, MA: National Bureau of Economic Research, 2007.