Now that we’ve finally moved on from covid19, can we also look honestly at what was driving school closures?
Most importantly, school board bureaucrats weren’t controlling a furious parent population anxious to get their kids back to school.
I knew closing schools was a horrible idea back in March 2020, and opposed everything short of return to in-person instruction. But even though the reporting on school closures was very solid, the narrative was universally wrong.
So I thought I’d go through the data to paint a more accurate picture. Schools were fully remote when a majority of parents in the school or district wanted them that way. For reasons no one entirely understands, parent choice was strongly influenced by one particular demographic factor. There is some variation on this, but it’s by far the most reliable predictor.
Where’s the rage?
NYTimes writer Jessica Grose described her surprise at recent polling showing that parents are generally happy with their public schools, even after the pandemic.
Morning Consult has been polling on education throughout the pandemic.
In October 2020, they found that most parents polled wanted remote education and, of those parents who had a choice, most chose virtual learning.
In April 2021 they found that most parents (82%) were very or somewhat satisfied with how their school responded and both the general public and parents were surprisingly satisified with local and national teachers unions (well under 30% strongly or somewhat disapproved, while 48% (public) and nearly 60% (parents) strongly or somewhat approved of local and national teachers unions.
In November 2021, Democrats won the education question so thoroughly among pollsters that Morning Consult advised Republicans to focus on the economy instead.
Morning Consult also runs a tracking poll on parental preference for instruction method, comfort with inperson instruction, and beliefs on school reopening. While they only poll 400-500 adults, the results do not in any way support the preferred narrative. Even today, 21% of parents thought schools were opened too quickly and no more than 50% at any tme believed they were opened at the right speed–and at no point in time did more than 50% of parents think in-person instruction was the best solution.
All polls reject the narrative of Goliath school districts beating down the little David parents who finally banded together to fight. Polls mostly show a very contented parent pool, consistently at 70% or over satisfaction, with relatively little unmet demand for in-person instruction. At a time when increased media attention was occupied by furious parents with megaphones, most parents were quietly satisfied with their schools’ response to a difficult situation.
Polls also seem to contradict the assertion that aggressively lenient covid19 policies were the road to political success. Covid19 hawk politicians seemed to have more voter approval than covid doves.
Gavin Newsom of California, who enforced remote instruction by mandate longer than any other state, won a recall with just 49% approval on covid19–but a lot of that disapproval was from covid19 hawks pissed he hadn’t been more restrictive. Recently he’s been polling at 60% on his covid19 handling. New Mexico’s Michelle Grisham, another governor who kept schools closed for a long time, doesn’t have great polling on crime and economy, but her covid policy approval is at 60%. Meanwhile, Florida’s DeSantis, much lauded by school opening proponents, saw his polls go up and down; while Florida is gaining a lot of transplants who like his policies, the natives are split. Texas’s Greg Abbott has tracking polls specifically on covid, and his approves were almost always lower than his disapproves.
If the policies that Abbott and DeSantis insisted on were as popular and obvious as their media fans would have it, shouldn’t they be getting raves for their policies?
Newsweek evaluated governor polls a year ago and found that “Democrats and Republican governors in blue states appear to have benefited the most from their approach to the crisis, while Republicans in deep-red states have largely suffered if they did not take strong action against the virus.”
So there’s a clear cognitive dissonance between the opinion rhetoric from all areas of the political and media arena (where even the covid19 hawks are defensive) and the public opinion polls, which seem to be largely satisfied with whatever policy their state has to offer, but not noticeably preferring aggressive re-openers.
The First Variable
Everything gets a bit clearer by tossing race into the narrative.
Race? Good lord, why think about race?
Well, for starters, pretty much everything about public schools involves race. It’s the first variable. It’s the one researchers and the media control for when they want to blame public education for its failure, and the one they ignore when it brings up questions no one wants to answer.
For another, like most issues involving education, viewing the issue through a racial prism is instructive–even if, as in this case, it’s hard to figure out why the racial aspect exists. Any poll on parent preference that controlled for race–which was most of them–showed the same results: white parents were consistently distinct from non-whites. They were the first to demand school openings, in-person instruction, and the first to reject mask mandates and vaccine mandates.
For once the journalist side of the media did its job in reporting the parental preference. There are dozens of articles like Angry White Parents vs. the Public School System:
Belying the issue of urban school districts’ hesitance to reopen for in-person learning amid outcries from mostly white and upper-middle class parents is a complex racial dynamic that underscores how the coronavirus pandemic was experienced by and has affected groups of people differently.
White parents are least happy with online learning. Only 34% of white families prefer fully remote school, compared with 58% of Hispanic, 59% of black, and 66% of Asian families.
Every state in the country has produced articles or research discussing the racial imbalance of parent preference, that non-white parents were reluctant to return to inperson instruction and whites were eager and angry at any delays. National surveys told the same story. By November 2020, 45% of “low-minority” districts throughout the country were offering full-time in-person instruction, while only 28% of “high minority” districts were–and “low minority” and “high minority” cover a huge range.
Combining this clear racial distinction in education model preferences, and then adding in the fact that most school districts don’t reflect America’s demography on a percentage basis, and much becomes clear. From the earliest days of the pandemic to today, white parents are disproportionately the covid-19 doves. This pattern is found throughout the country and holds up regardless of state covid19 policy, political affiliation, or relative union strength.
The race factor is often obscured to those who don’t go looking for it. Regional and school demographics don’t always run in tandem. Many majority white counties have majority non-white school districts. For example, Arlington County is 60% white, but the school district is 46% white. Fairfax County is 50% white but its school district is 39% white. San Francisco’s school district is 34% Asian, 31% Hispanic, 14% white, and 8% black, while the city itself is 39% white, 34% Asian, 16% Hispanic, and 5% black. These differences matter when the racial pattern of support for strict covid19 policies is so clearly predictive.
Sometimes seemingly inexplicable decisions make more sense when district demographics are taken into consideration. For example, Michael Brendan Daugherty was furious earlier in the year when two neighboring districts in Westchester County ended mask mandates while his district was still mandating them. But the two neighboring districts are 75% white and if MBD still lives in Mt. Kisco, his school district is just over 50% white.
Similarly, urban charters closed for as long as their public counterparts. In California, charter students were more likely to be in remote instruction than publics (63 to 55) and less likely to be in hybrid or full-time, a finding that Martin West’s excellent survey confirmed nationwide as well.
That charters were more likely to be remote than publics is very consistent with the racial preference in covid policy argument, as the charter student population is majority non-white, and thus more likely to favor remote instruction.
Polls were the very first sign of different racial attitudes towards opening schools. But revealed preferences show the same story. Whenever parents were given a choice for in-person instruction–at least where we can track the data–white parents were far more likely to take that choice. Non-white parents were far more likely to stay in remote mode, even when state leadership was aggressively in favor of in-person instruction.
Florida and Texas were hailed by the conservative media and chastised by the mainstream media for their bold approach to opening schools. Meanwhile, New York City and Chicago were targeted by the conservative media and, eventually, the mainstream media for their delayed reopenings.
At various points in the year, Chicago, New York City, Houston, and a wide swath of central Florida either a) surveyed its entire parent population for enrollment intent or b) tracked each student by actual enrollment choice. All four areas provided datasets revealing parental preferences for in-person or remote instruction and included a wide variety of family demographics.
In September 2020, Houston polled all the parents in its districts on whether they were coming to school in person or remote. A central Florida newspaper collected school data attendance, remote vs in-person, for several districts in November 2020. In March 2021, Chicago queryed its parents in k-8 schools on intent to enroll. New York City collected enrollment choices throughout the pandemic year.
Only one of these data sources coded parental choice by race. But all district and school demographic data is publicly available information and while it’s boring to look up every district and school for that information, it’s not difficult. That makes it possible to compare percentage of white students to percentage of students opting for in-person instruction.
Logic will aid in matching graph to region. Politics will not. In all cases, white students are positively correlated with in-person instruction. Governance or unions or even time in pandemic cycle doesn’t affect the strong white preference for in-person instruction. (It’s theoretically possible that more white students in a school led more non-white kids to opt in and the white students to opt out, but I’m hoping it’s obvious that’s not the most likely explanation.)
Graph A: 5 central Florida counties (November 2020, actual enrollment, source). While the data patterns have the widest range, I looked up several of the outliers and found a few inconsistencies. For example, Hegerty High School has a 66% white population with only 3% in-person enrollment, according to the data record. But several stories written just a couple months after the data record makes it clear that a much larger percentage is in school. So some outliers are probably genuine, others might be errors.
While I’m on the topic of Florida, nowhere was the narrative more egregiously skewed than on the fabled state where “the schools are open”. As the data shows, many non-white schools were in remote mode. The two largest counties had similar patterns. Broward County (65% nonwhite) and Miami Dade County (90% nonwhite) refused to open their schools at all until October and November. By the end of the 2021 school year, half of Miami’s public school students and 55% of Broward County’s were still in remote. Both counties were described by Edweek as “open for wide-scale, inperson learning”. Nat Malkus of AEI introduced an instructional status tracker that scraped school district websites and determined how many districts in each state were offering in-person instruction. His tracker showed Florida’s districts as 100% in-person from its inception.
Both Edweek and Malkus may have been technically accurate, but when half of the two largest Florida districts were remote for the entire year and large chunks of central Florida were likewise choosing remote, perhaps these statements were a tad misleading.
30% of all Florida students were still in remote instruction in February 2021. Given the pattern above, it’s very likely most of those students were non-white.
Graph B: Houston (September 2020, enrollment intent, source). Texas governor Greg Abbott was very aggressive about opening schools, believing long before it became common knowledge that remote education was bad for students. In fact, the open nature of Texas schools was a compare-contrast case to New Mexico in Pro Publica’s heartrending story of a high school football star’s suicide in New Mexico. But even though “the schools were open in Texas”, Texas students of color returned to person in below average rates: ” 56% percent of Texas students on average returned to on-campus instruction during the school year, including 75% of white students, about 53% of Black students, 49% of Hispanic students and 31% of Asian students.”
Graph C: New York City (Jan 2021, actual enrollment, source). NYC has the least impressive trendline in my graphs given the fact that almost no school had a majority white population. But the NYC data source has something that none of the others have: a breakdown of all students’ in-person or hybrid selection by race, which makes another easy graph possible (I only do easy graphs):
One of every two white NYC students chose inperson, as opposed to 1 in 3 Hispanics, 3 in 10 blacks, and 1 in 5 Asians.
Graph D: Chicago (March 2021, enrollment intent, source). As Keeanga-Yamahtta Taylor wrote in the New Yorker back in February, the first time the Chicago teachers refused to return to class, roughly a third of Asians, Hispanics, and blacks signaled an intent to return to in-person instruction and of those numbers less than half of blacks and Hispanics and only 60% of Asians actually showed up. Meanwhile, two-thirds of whites opted in and 75% of that group actually showed.
Four regions. Two different covid19 policies. Same pattern. These are parents making choices based on identical policies offered, often in the same schools.
Was Race a Proxy for Politics?
Nate Silver has speculated that the racial skew in education preferences is just reflecting political preferences.
Indisputably, Democrat governors were far more covid-hawkish than GOP governors, and GOP-run states were more likely to allow full-time in-person instruction. State political control was the second biggest factor in school opening decisions and in my telling, Democrat governors bear a huge amount of the blame for the angry white parents (a topic for a second piece I’m working on, assuming I ever finish this one).
But why were Democrats more cautious and controlling? Why were Democrats stressing safety long after the media had determined that covid19 didn’t pose much of a risk to young people? Perhaps because there weren’t millions of angry parents.
To disentangle this, begin with the fact that the progressive left went totally nuts on covid. I don’t know if it began as a reaction to Donald Trump’s desire to play down the pandemic or a genuine fear of a virus that wasn’t all that dangerous for most of us, but even today thousands, possibly millions of people are still covid hysterics freaked out that mask mandates are ending, and they all are the sort who post their pronouns.
The moderate left did not go totally nuts, or at least got less scared much faster. That’s why Democrat politicians are so terrified that a midterm meltdown awaits them, right?
But who is the moderate left? Non-whites aren’t particularly progressive. It’s well-established of late that progressive, “woke” Democrats are much more likely to be white, that non-white Democrats are more centrist and moderate.
Building out from that fact: if politics controlled choices, then centrist non-whites should behave more like moderate white Democrats, favoring in-person instruction and less restrictive policies. But the evidence shows exactly the opposite. So either nonwhites are more progressive than is currently thought, or non-white centrist Democrats are behaving differently from white centrist Democrats and politics is a less effective predictor than race.
To separate politics from race, I looked at three cases.
Case 1: how did Republican politicians respond to the demands of majority non-white districts? If these districts wanted in-person instruction, the politicians themselves would owe no allegiance to the teachers unions and would readily open the schools.
But data above from Florida and Texas show nonwhites actively opted for remote instruction. Couple that evidence with that from Mississippi, a state with weak unions and a governor who aggressively opened schools (and oh, by the way, had pretty terrible polls as a result). The six poorest school districts in Mississippi, all entirely African American students, were online for the entire 20-21 school year. Then consider Tennessee, where just 1 in 4 students started the 2021 year in remote. Yet in two largest schools districts, the remote rates were 46% (Nashville, 75% non-white) and 67% (Shelby County Schools, 88% nonwhite–and by the way, when allowed to go mask optional, just 2% took the deal in the early days).
GOP governor Larry Hogan of Maryland, a Democrat state, had to beg schools to open by March, and Baltimore City schools were still primarily remote for most of the year. Massachusetts governor Charlie Baker, also a Republican running a blue state, first started pushing in November 2020 to convince districts to open, but 23% of students were still in full-time remote instruction in February. Boston, the largest district, (15% white) didn’t start opening schools to the general population until March while Springfield, the second largest district (10% white), delayed opening even elementary schools until April.
One of the more interesting and instructive examples of politicians responding to race-related covid hawkishness better than politics just took place in Virginia. Glenn Youngkin ran on ending the mask mandate. Virginia’s counties are school districts, for the most part, so it’s very easy to map county demographics, governor support, and–thanks to the Washington Post–whether or not the district went mask optional after Youngkin’s declaration but before the lawsuit was decided. The first graph shows the county’s decision to go mask optional or mask mandate based on the percentage of white students in the district. The second breaks down the decisions further by the district governor support. (Forgive the weird colors.)
As the Washington Post story noted, many districts were simply holding onto the mandate until the court case was decided, for all sorts of legal reasons I’ll try to discuss in my next piece on this. This probably explains the outliers. Still, many Republican voting districts with majority non-white student populations held onto the mandate.
Case 2: What did schools do in majority white Democrat states? If politics drove school policy, then Vermont (90% white) and Washington (63% white) should have had the same school closure patterns as more diverse Democrat states.
In Vermont, just 14% of students were in remote and 30% of students were in school full time by February 2021. At no point in the entire 20-21 year were Vermont students more than 18% remote. High school students were rarely allowed to return full-time, but were overwhelmingly in hybrid mode.
In Washington, statewide data shows that much of the state had open schools (light and dark blue).
The map on the left outlines the Trump voting counties. The northwest and a southeast corner of dark blue did not vote for Trump, but have plenty of dark blue. The map on the right is all the majority white districts which account for the dark and medium blue areas that the Trump voting regions miss.
Only with the white majority district map do you see how many districts had lukewarm Trump support and decent to high in-person instruction.
Thus, in at least two majority white states dominated by Democrats, school options were far more similar to Republican states and very unlike diverse Democrat states.
The final case, and the most difficult to hunt down, is the inverse of majority white Democrat regions. What did majority non-white counties that voted Republican do? I found two, one that went each way. Dodge City School District, a Republican area with a school districts that’s 80% Hispanic, saw 95% of kids choose in-person instruction despite very high covid19 rates. Another, Zapata County, despite governor Greg Abbott’s stated priority for open schools and inperson instruction, was one of the school districts that begged for more time in remote instruction and pushed the time to even offer in-person instruction, much less open the schools to everyone, to December. No helpful conclusions there.
Recently, Report Card podcast’s Nat Malkus discussed his Return to Learn tracker that’s been updated to reflect mask mandates, and he goes on for half an hour about political views and never once mentions race. It’s depressing how often people attempted to correlate political positions with covid19 policies. Many public schools are filled with students whose immigrant parents can’t vote. Moreover, school districts only occasionally overlap with counties. But school districts faithfully report their racial demographics. It’s much easier to correlate racial demographics than political ones. So easy a teacher could do it! For some reason few researchers wanted to go there.
It Wasn’t Unions
(I originally had a lot more rebuttal to the “unions closed the schools” narrative, but that was probably my obsession talking.)
The most common culprit offered up for continued remote education are teachers unions–although most union leaders think of their responsibility as credit, not blame. In fact, unions were at best tertiary.
Randi Weingarten has no power to influence the teachers in her union–and by the way, her union is an also-ran of the Big Two. (Quick, name the president of the really powerful union, the NEA. No googling.) Weingarten’s primary job is as fundraiser and showboat, not teacher representative. What Weingarten advises the CDC has absolutely no impact on school closures. Governors weren’t bound to follow the CDC, and many didn’t.
Several studies purported to find union strength as predictive of remote education, most notably Corey De Angelis’s study of September 2020. DeAngelis used four metrics of union strength: right to work status, Fordham Institute ranking, percentage of union members in labor force and increase in union menbers at county level. If I’m reading this correctly, only the Fordham Institute ranking was directly related to teacher union strength. All but four of the right-to-work states are over 60% white. Two of the states with right to work laws, Texas and Florida, have clear patterns of large percentages of non-white students opting out of in-person instruction.
The Fordham Institute rankings are good, and when DeAngelis used this ranking, he found that a10% increase in “union power” made a school 1.3% more likely to close, which doesn’t seem to be all that much.
In any event, exceptions to the union story abound.
For all the complaints, New York City, considered to have strong unions, opened schools relatively early in December, February, and March for elementary, middle, and high school. Similarly union-strong Chicago opened schools by March. California didn’t open most of its schools until April and only after Gavin Newsom essentially bribed the school boards , not the unions, to open. Even then, at least three large districts stayed in remote education all year.
New York, California, and Illinois are ranked in the top ten for most powerful teacher unions states but had wildly different reopening schedules. One possibility: New York City and Chicago need their white parents for their tax base, and assuaging them to the extent possible would have to be a high priority. Meanwhile, California has different power bases depending on the location. In the northeastern part of the state, just a third of all students were in remote mode. (Quick, guess what blue stands for in this California county map.)
Meanwhile, Montana, third on the list for most powerful teachers unions and with a Democrat governor, opened schools in May 2020 and by September 2020, most schools–not school districts, but schools–were open for full-time instruction. Ohio has strong unions but its schools were mostly opened in September, as were the aforementioned Vermont and Washington, both with strong unions.
Texas and New Mexico, whose school policies were compared in the Pro Publica football piece, both have very weak unions yet had entirely different re-opening policies at the state level. (These two states are an interesting contrast, more on this some other time.)
North Carolina has very weak unions, and is over 60% white. School boards were given the choice at the beginning of the 2020 school year to go full remote, hybrid, or full-time in person. Last amateur graph, I promise:
Another study finding that union strength correlated with closures found that districts with Catholic schools were more likely to open for in-person learning. As it happens, North Carolina isn’t terribly Catholic, but its recent Hispanic influx has increased the numbers over the past 20 years. There are 46 Catholic private schools in the state. Forty of them are in districts that chose remote; six were in districts that opted for in-person. Of those six schools, 5 were in counties that were 50-80% white. One was an outlier, just 34% white.
I don’t know why white parent preferences varied so distinctly from the others. Nor do I “blame” non-white parents for taking the remote option, even though that preference clearly appeared to have damaged educational outcomes for low income black and Hispanic children.
As Andy Smarick wrote a year ago, making many similar observations to the ones here: “the story hasn’t been politics, unions, laziness, risk-aversion, or ignorance. It’s been parents, pluralism, and self-government.”
A year ago. His article was largely ignored, as were the many polls and stories reflecting the fact that schools were in fact responding to local parent demand.
How would things have changed if the debate had honestly acknowledged the reason for school closures? Both sides were actively distorting reality. For every “open the schools” advocate, there was a matching “covid19 is dangerous” proponent. But in fact, the correct answer to “open the schools” was “the majority of parents in your district disagree with you”–a response the covid19 hawks didn’t want to give because they wanted to present school closures as “science”.
Perhaps an honest discussion would have gone into when and how it was easier and less expensive to default to closed schools, and when the odds went the other way. Perhaps acknowledging the schools never should have been closed would have been helpful and, having made this mistake, that parents should never have been given a choice to keep their kids home once schools were open again.
The irony, of course, is that most of the loudest voices advocating for opening schools are often in favor of giving parents a choice.
I’m not interested in blame. I’m interested in making sure people understand what happened because lord knows I don’t want this to happen again.
But that’s for the next article. (At least, I hope so. I’m a slow writer.)
(Note: I edited this after I posted it, focusing all the union content in one area, and I added a review of the Corey DeAngelis study. No other changes were made. I just realized I was getting a tad obsessive about the “union closed the schools” narrative).