Public School Battles, Mapped
To examine how the kinds of controversies that paralyzed Coachella Valley and Dover schools affect academic achievement, I combined information from the Cato Institute’s Public Schooling Battle Map with test score data from the Stanford Election Data Archive.
Cato’s catalogue is limited to sufficiently high-profile controversies that attract media attention and misses many smaller conflicts in districts with less media coverage. It is thus useful to think of the “treatment” that causes a district to enter the Cato dataset as a bundle that includes both the event as well as the media attention that follows.
The number and types of controversies that plague local school districts is surprising, as a few examples help illustrate. One, for example, involved an Ohio district where middle school students sang “Cotton Needs Pickin” at the fall concert, to the outrage of many African American parents in the audience. Another was the tragic case from a Southern California district that disciplined students who skipped school to attend an immigration reform rally. One of these students, apparently traumatized by his punishment, ended up taking his own life, prompting a high-profile lawsuit from his family.
Many others involve controversial school mascots—in particular, the “Redskins,” a popular team name that fell out of favor in recent years. Some controversies deal with free speech questions, often involving school-uniform or dress-code infractions.
Critically, the Cato sample includes controversies initiated by activists from both sides of the ideological spectrum.
An obvious challenge is that such events are unlikely to be random. For example, larger districts located in the core of their newspaper circulation area or television media market are more likely to see a run-of-the-mill, small-scale conflict escalate after attracting press coverage.
To isolate the effect of the controversies tracked by Cato from the types of districts most likely to create them, I leverage variation in the precise timing of each event, comparing changes in student achievement over time—before vs. after—to a set of school districts that never appear in the Cato dataset, a variant of the “difference-in-differences” quasi-experimental design often used in education policy evaluations.
Overall, I find that student achievement declines in the years after a local political controversy, with the effects concentrated in math and in elementary and lower middle school grades. The impacts are not huge, translating to about 10 days of learning in a typical 180-day school year, but they are meaningful (see Figure 1).
Although there is some evidence of pre-existing downward trends in districts that become the focus of political conflicts, a variety of robustness checks—including comparing only districts that experience controversies in different years and modern statistical models that adjust for pre-trends—suggest that the declines are real.
Interestingly, most types of controversies have little if any effect on student learning. This is also true for controversies surrounding instructional curriculum and those related to reading materials, suggesting that transition costs related to changes in pedagogy or teaching materials are unlikely to be driving the overall findings. Instead, large negative effects appear only for two types of controversies: those dealing with the teaching of evolution and disputes surrounding race, which generated declines of almost 1.5 months of learning in mathematics and somewhat smaller but still meaningful declines in reading. In addition, there are smaller but statistically significant declines following controversies dealing with freedom of expression—the most frequent type of conflict in the Cato dataset.
