A new publication by Tim Sass and colleagues examines the effect of charter high schools on long-term attainment and earnings. The study builds on others by the same authors, as well as a working paper of the study released over two years ago.
The authors focus on charter high schools in Florida, where they can access a wealth of data from the state department of education’s longitudinal database. That information includes various demographic and achievement data for K–12 students, as well as data on students enrolled in community colleges and four-year universities inside and outside of Florida. (The latter info was gleaned from the National Student Clearinghouse and other sources, and employment outcomes and earnings are merged from another state database.)
The sample includes four cohorts of eighth-grade students; the first cohort enrolled in 1997–98, the last in 2000–01. They are able to observe labor outcomes for students up to twelve years removed from their eighth-grade year.
Before we get to the results, let’s address the biggest analytic hurdle to be overcome: selection bias—meaning that charter school students, by the very act of choosing an educational alternative, may be different in unobservable ways from those who attend traditional public schools (TPS). Indeed, the vast majority of the paper discusses not the findings but the various attempts to address this inherent issue in virtually all choice impact studies. Absent randomized lottery data, the authors limit their sample to students who were enrolled in charter schools in eighth grade, positing that they all possess similar unobserved characteristics. They then divide them into two groups: treatment students who enrolled in charter schools again in ninth grade and control students who switched to a TPS. The analysts also match students on observable baseline characteristics such as family income and eighth-grade test scores.
This is a very reasonable approach. But it is not without its flaws, as the analysts readily admit—especially since “back-end selection bias could occur through the comparison students’ choice to exit the charter sector after eighth grade.” By targeting a “typical transition year,” however, they reason that this bias is less significant than initial selection into the charter sector because nearly all are enrolling in a new school regardless of whether they are changing sectors.
And now for those results.
First, charter high school enrollment is positively linked to educational attainment. Specifically, there is a six-percentage-point increase in the likelihood of earning a high school diploma within five years and a nine-percentage-point increase in the likelihood of attending college. There is also a positive relationship between charter high school attendance and college persistence (defined as attending college at least one semester in consecutive years), with roughly a twelve-percentage-point boost for charter high school students. The latter also see the equivalent of a 12 percent increase in maximum earnings from age twenty-three to twenty-five (again, compared to students who attended charter middle school but transferred to a traditional high school).
The paper notes (as have others) that these results are particularly intriguing because prior studies have shown that charter schools in Florida have not impacted student test scores much. So, the logic goes, perhaps we shouldn’t pay much heed to test scores. Yet the literature is not clear on that point. (Even the study cited in the report as showing poor charter results finds that by year five, charters are on par with TPS in math and produce better reading achievement. Plus, that study does not target charter high schools, much less these charter high schools.)
Besides that, we can’t directly apply findings from one state to different charter schools in different states with different kids and different policies. (And quasi-experimental studies, especially, can’t remove every shred of bias due to unmeasured characteristics between treatment and control groups.)
Also remember that these are findings based on averages; yet kids attend specific charter schools. We simply don’t know if there are actual high schools that produce lackluster test scores but get impressive graduation, college completion, and earnings bumps.
So what’s the takeaway? First, there’s a growing body of evidence indicating that charter schools, along with Catholic schools, produce very strong real-world outcomes. Second, despite what others might imply, we should continue to judge high schools in part by their test scores—and shouldn’t shy away from shutting down the ones that, year after year, post dismal results for entire groups of kids.
SOURCE: Tim R. Sass, Ron W. Zimmer, Brian P. Gill, and T. Kevin Booker, “Charter High Schools’ Effects on Long-Term Attainment and Earnings,” Journal of Policy Analysis and Management (April 2016).
