One of the most interesting and significant findings about charter schools in the last decade—outside of the fact that they tend to outperform traditional public schools (TPS)—is that growing charter enrollment is tied to significant increases, on average, in the achievement of poor, Black, and Hispanic students across entire metro areas (particularly larger ones). In other words, a rising tide of charters lifts all boats, regardless of whether disadvantaged students are enrolled in charter or traditional schools.
A recent working paper from David Figlio and colleagues adds some additional nuance and methodological innovation to these findings. The new wrinkles utilize sibling groups and access to birth data to enable better controls and attempt to assess whether concurrent competition from private schools mutes or magnifies competition from charters.
The researchers examine how an increase in access to charter schools in twelve districts in Florida affects a number of outcomes for students remaining in traditional public schools. Charters in these dozen districts serve about 13 percent of the students in their jurisdictions, while the entire Florida charter sector serves almost 12 percent of students in the state, thus raising the likelihood that results are applicable to the Sunshine State as a whole. The analysis merges student-level administrative data for students in the dozen districts from the 2000–01 through 2016–17 school years with birth record data (including parental education and income) for all Florida students who were born in the state between 1992 and 2002.
Figlio and team measure competition primarily through a charter-density approach, which captures the number of charters within a five-mile radius of a given TPS. The study uses three different analytical methods.
The first model uses each student as their own comparison group. That is, a student’s relative performance in a year where their charter school faces little competition is compared to their own performance in a year where their charter faces more competition (due to charter closings or openings).
The second model takes particular care to control for student sorting. For each student, it makes comparisons to a single average competition measure based on the predicted level of charter competition that the student will be exposed to (without going into great detail, that prediction is loosely based on students born in a given ZIP code in a given academic cohort).
The last method has siblings serving as comparisons for one other. Specifically, it compares the outcomes of two or more siblings, each attending a given grade level in the same TPS in different years. Analysts determine whether the outcomes of a sibling who attends a given grade in a given school are systematically better (or worse) when the school experiences more charter competition, compared to the sibling who attended the same grade in the same school under conditions of lesser charter competition. The siblings analysis is their preferred method because it controls for unobserved family characteristics, such as parental expectations or the ability to competently help with homework and helps to eliminate typical criticisms, including the possibility that new charters take only low-performing students and leave high performers at the district schools.
In looking across all three models, the researchers find that competition stemming from the opening of new charter schools improves reading but not math performance, and that it also decreases absenteeism among students who remain in the TPS. The siblings analysis is the more conservative set of results and finds that an increase of ten charter schools within five miles would be associated with a 0.035 standard deviation (SD) increase in reading scores. Since an expansion of ten charter schools is not a likely scenario in real life, they recalculate their results in terms of a single-school increase. For math, results are not significant. For reading, they find a statistically significant effect size of 0.35 SD and statistically significant reductions in absenteeism of 0.74 percent of the sample mean. Although those reading impacts seem modest, analysts say that they nonetheless compare to the effects of the voucher program in Florida (based on an earlier study). Subgroup analysis shows that absenteeism results are null for Black students, but that charter competition is associated with reductions in absences for White and Hispanic students. Test score results by subgroup depend largely on the model, at least one of which shows statistically significant negative effects for White students.
Finally, the researchers look at whether competitive effects of charter schools are influenced by private schools. For math, results are statistically insignificant regardless of the level of private school saturation (or the selection of model). For reading, they find positive results between charter competition and achievement that are driven by the schools facing more private school competition (but only in the siblings analysis). By contrast, for absences, they find that charter effects are bigger when there is less private school competition in the local area, but here results differ based on the approach, so analysts advise caution.
These findings largely reinforce those from previous studies on charter competitive effects and make them even more compelling, given the stronger controls inherent in the siblings analysis. What’s more, adding more private school competition in places with lots of charter competition doesn’t harm student outcomes. Still, we need replication studies in other states, since the market share of charters in the Sunshine State is not comparable to most others due to its large size and relative diversity.
SOURCE: David Figlio, Cassandra Hart, and Krzysztof Karbownik, “Competitive Effects of Charter Schools,” Annenberg Institute at Brown University (February 2024).