Teachers’ Unions Remain Powerful as Membership Falls and School Politics Shift

Teachers’ unions are still among the central interest groups in American education, but Michael Hartney argued that their power cannot be understood simply by asking whether a state has a collective bargaining law. Hartney’s own interest in the subject came from working in policy and government at the National Governors Association during the late-2000s period of bipartisan education reform. From that perch, he saw teachers’ unions and other established education groups as constraints on what reforms governors could plausibly pursue. But he also saw a mismatch between academic measurement and political reality: scholars often classified unions as strong or weak based mainly on the presence of collective bargaining, while governors in non-bargaining states still described teachers’ unions as major players.

Collective bargaining laws matter, Hartney said, and they are one component of teacher-union power. But they are not the whole thing. Teachers’ unions operate through national organizations, state affiliates, local affiliates, electoral activity, contract negotiations, and less formal consultation. The result is a layered political organization whose influence varies by venue.

Melissa Lyon described a similar path into the subject from the classroom side. She taught sixth grade in Houston, Texas, where collective bargaining is illegal, and then encountered unionized school environments while doing graduate fieldwork in New York City, Buffalo, and elsewhere. The contrast led her into research on teachers’ unions, right-to-work laws, cross-sector collaboration, teachers running for office, and teacher strikes.

The scale is still large. Lyon said more than two million pre-K–12 teachers are unionized today, a little over two-thirds of public school teachers in the United States. She put the current figure at about 68 percent, down from roughly 80 percent about 25 years ago. The decline is broad but uneven: 45 of 50 states have seen membership rates fall since 2007–08, with Wisconsin’s drop especially large and, in Lyon’s view, almost certainly related to the state’s 2011 Act 10. Yet in the ten strongest membership states, union membership remains above 90 percent, and in 26 states at least seven in ten teachers are union members.

That membership pattern matters because the discussion was not about unions as a single national actor. Hartney stressed that one can say there are two teachers’ unions in the United States — the National Education Association and the American Federation of Teachers — or one can say there are thousands, because the national organizations are connected to state and local affiliates. Which answer is right depends on the question. During COVID, for example, national AFT president Randi Weingarten could say she supported reopening schools, but Hartney said that position had limited bearing on whether individual schools reopened. Labor relations in the public sector are largely settled locally, through contracts, memoranda of understanding, or other local negotiations and consultations.

That structure also complicates the common distinction between “collective bargaining” and “politics.” Hartney invoked NEA lawyer Robert Chanin’s argument in litigation that it is impossible to separate the NEA’s collective-bargaining efforts from politics. Hartney’s point was not merely partisan. In public-sector bargaining, questions such as teacher pay, class size, benefits, and working conditions are also public budget and policy questions. The infrastructure of a local union — presidents, boards, building representatives — can be used for contract work one day and school-board election activity another. Hartney said trying to assign a clean percentage of union time to bargaining versus politics misunderstands how the institution works.

The Supreme Court’s Janus decision, in his telling, turned in part on that same difficulty: even bargaining over teacher pay or class size involves contested public choices. In the public sector, there is no neutral line separating workplace negotiation from political decision-making.

Tom Church pressed the researchers on where the facts come from, and the answer was mixed. Lyon said the Current Population Survey provides useful national-level information on union membership, and unionstats.com makes those data accessible at a fairly detailed level. The National Teacher and Principal Survey, formerly the Schools and Staffing Survey, also provides estimates that can be broken down by state. For longer historical state-level series, Lyon has used NEA handbooks to track membership back decades, including for work on right-to-work policies reaching to the 1940s.

The local level is different. Lyon said she would like district-level union membership data, but with roughly 14,000 school districts, many of which are not unionized, collecting local data is difficult. She did not know of a source with comprehensive local union membership data. Political activity is similarly hard to observe. She has used election-spending data from the National Institute on Money in Politics, collected original data on teacher strikes, gathered teacher-union endorsements, conducted surveys, and interviewed teachers and community members. But some of the most interesting political activity is also the hardest to measure.

Hartney made the data gap a governance issue, not just a research inconvenience. If an interest group has the legal right to bargain collectively with government, he argued, that gives it a special role in policymaking. During COVID, school districts were not generally required to sit down with a local PTA president before deciding whether to reopen. In many places, they did have to sit down with labor partners. Whether one supports or opposes collective bargaining, Hartney said, basic public information should be collected: which districts have collective bargaining, how many teachers are employed, how many are union members, and what salary increases appear in the labor contract.

Indiana, he said, provides an example of the kind of transparency he has in mind: taxpayers can go to a state website and see district teacher employment, union membership for the exclusive representative, prior-year salary increases in the labor contract, and related information. He compared this to lobbying disclosure — not inherently partisan, but basic information about how policymaking works.

Hartney added an important symmetry condition. If transparency is required for teachers’ unions, he said, it should also apply to police and fire unions. Any public-sector union participates in policymaking, and the transparency principle should not be applied selectively. He credited some Republican governors with improving disclosure in the education context, but also said such reforms are sometimes implemented in a partisan way.

Lyon’s teacher-strike research began with the problem of building a national dataset. She went back to 2007–08, used administrative data where states had it, and then developed a process for identifying strikes and their stated reasons through local news sources. She found evidence of almost 800 teacher strikes since that period.

The reasons were not mutually exclusive, but compensation dominated. Lyon said nine out of ten strikes were for compensation. About 60 percent involved working conditions, and about 10 percent involved “common good” reasons — demands framed around broader public or community concerns rather than only narrow pay-and-benefit issues.

The first published paper from that project, “Teacher Strikes as Public Signals,” examined how strikes affect political discourse. Lyon said the study found that teacher strikes cause political elites to talk much more about education. The effect is especially large in the period before elections, when education becomes more prominent in political advertising. The strikes also have downstream fiscal effects: they increase education spending.

In related work, Lyon said she and collaborators are examining compensation, working conditions, student achievement, and school-board election turnout. Compensation rises after strikes. Pupil-teacher ratios go down. For student achievement, she said they find no long-run effects from the typical strike.

That combination complicates the simplest pro- or anti-union story. In Lyon’s account, strikes are effective political signals and can increase spending and improve teacher compensation and staffing conditions. At the same time, the typical strike does not appear, in her work, to produce long-run student-achievement effects.

Hartney and Lyon both treated COVID school closures as a critical case for understanding union power, but they emphasized different cautions.

Hartney began with the learning-loss backdrop. He said the academic slide in the United States predated COVID and appears to correlate with the decline of the bipartisan accountability era associated with No Child Left Behind. He was careful not to claim a settled causal explanation; he noted that many theories are circulating, from cell phones to accountability changes. What is clearer, he said, is that there was enormous variation in reopening. Private schools, including low-cost Catholic parochial schools, were much more likely to return to in-person learning than traditional public schools. Within the public sector, many large urban districts in California, Oregon, and other West Coast states kept students out of a regular in-person cadence for 16 to 18 months, while other places did not.

Hartney said peer-reviewed research shows learning loss was steeper where remote learning lasted longer. He described that as depressing but not surprising: if children are not in school, they are less likely to learn. He also noted the reversal in pre-COVID ideological expectations. Before the pandemic, conservatives were often more favorable toward virtual learning because they saw potential cost savings, while unions were skeptical. During COVID, the partisan valence shifted.

To get at reopening variation, Hartney surveyed about 5,000 school-board members in 2023, most of whom had served during COVID. He asked them why private and Catholic schools reopened while their districts did not. According to Hartney, school-board members rarely identified resources, class size, or facility quality as the main issue; those explanations drew only about 10 to 15 percent. More than half instead mentioned bargaining with a teachers’ union resistant to quick reopening, partisan politics, or public-system bureaucracy.

Hartney did not say unions were the only factor. He called them “nontrivial,” in part because their institutional role is to represent members’ preferences. He also resisted a too-easy separation between “teachers” and “teachers’ unions.” While conservatives sometimes say union positions do not reflect teachers, Hartney said prior survey work by Terry Moe showed unionized teachers were generally satisfied with how unions advocated for them. Still, he wondered whether COVID produced slippage between what everyday teachers wanted and the institutional positions unions took, especially when unions had to represent members with different views on vaccines, reopening, and risk.

Lyon’s response emphasized that reopening decisions were formally made by school boards and districts. Collective bargaining and negotiations have two parties, she said, and the district ultimately decides whether schools open. If district leaders believe teachers will not show up, that is a serious constraint, but it does not make the union the only decision-maker.

Her own limited COVID-specific research came from a survey of state policymakers and advocates across all 50 states. A quarter of respondents said teachers’ unions were definitely key decision-makers in reopening decisions, and about half said they were definitely or probably key decision-makers. Union representatives were even more likely to say unions were key decision-makers: about 70 percent of teacher-union leaders said so, compared with 49 percent among non-union leaders.

Those responses support the idea that unions mattered. But Lyon warned that “key decision-maker” does not mean “solely to blame.” In hindsight, she said, few actors want to take full responsibility for extended closures. She also highlighted the identification problem: it is difficult to separate union effects from district choices, local political culture, and the characteristics of places that tend to have strong or weak unions. Strong-union districts may also be places with different partisan alignments, different public-health preferences, different bureaucracies, and different community expectations.

The disagreement was not over whether unions influenced reopening. Both researchers said they did. The tension was over how directly to assign responsibility, and how much one can infer from the correlation between union strength, political context, and closure duration.

When Church turned to artificial intelligence, Lyon was cautious. She said she had not done specific research on teachers’ unions’ responses to AI or on classroom integration of AI, and therefore did not want to speculate too far beyond her expertise. But she agreed that education is at an inflection point, not only because of AI but because of declining enrollment and the status of teaching itself.

Lyon pointed to her paper with Matthew Kraft, “The Rise and Fall of the Teaching Profession,” which she said shows the profession’s status at a historic low around 2020–21 across multiple measures. That possible low point coincides with enrollment decline and technological change. She said unions will need to adapt.

One possible adaptation, in her view, could come through school choice. If private-school employment expands, teacher organizing need not be limited to traditional public schools. In many states, she noted, private-sector workers have stronger collective-bargaining rights than public-sector workers because of the National Labor Relations Act. She saw no obvious reason teachers in private schools could not unionize and organize, though she placed that in the more creative category of possible union responses.

At the national level, Lyon said the AFT has already made prominent public collaborations with AI companies, while unions are still trying to determine what role national organizations should play in AI advocacy. Locally, adoption will vary widely because decisions will be made across thousands of districts and schools. Some unions may work aggressively with AI tools in pursuit of student achievement, while others may be slower or more resistant.

She also noted that technologies can become partisan-coded in unstable ways. Virtual learning was more Republican-coded before COVID, then became more associated with Democratic reopening positions during the pandemic. AI, she said, seems recently to be trending toward Republican coding, but it is too early to know where that will settle.

Hartney argued that the AI question is downstream of a more basic institutional question: what are schools going to be for? If enrollment is falling, if AI changes some academic functions, and if schools are under pressure from parents and taxpayers, then the justification for public-school staffing and spending depends on how the institution defines its purpose.

One vision treats schools primarily as academic institutions, judged by whether students leave prepared for college, career, and civic life. Another treats schools as broader custodial and community institutions — places where children spend the day while parents work, and where families can access health, dental, mental-health, and other services.

Hartney saw pressure on both the right and left away from academic accountability as the central measure. On the right, he described a “universal unbridled school choice” view: if the parent has chosen, the policy has succeeded, and test outcomes in math or English language arts are no longer the center of the conversation. He did not present that as necessarily wrong, but as a normative shift. On the left, he said, teachers’ unions’ enthusiasm for community schools reflects a similar move away from academics as the principal institutional rationale.

His union-specific interpretation was blunt: if enrollment is declining, academic outcomes are weak, or AI reduces the centrality of some instructional functions, then maintaining current staffing and funding levels requires another justification. Community schools provide one. They frame schools as “bedrock community institutions” whose purpose extends beyond academic learning.

Hartney did not think teachers’ unions are going away. Wisconsin, he said, is the only state that really “took the feet out from under” teachers’ unions, and even there they continue to exist operationally. The more important question is what kind of institutions they will become.

He connected that question to broader concerns about institutional decline and civic life. Since Robert Putnam’s “Bowling Alone,” Hartney said, Americans have become familiar with the decline in people joining and participating in civic organizations. Janus adds another layer for unions: teachers, police officers, firefighters, and other public employees can no longer be compelled to support a union as a condition of employment.

Hartney said one can make a strong case for that right. But he also raised a possible unintended consequence. If conservative or moderate teachers exit the union, the median active participant may move ideologically leftward. He compared the resulting union membership to a primary electorate rather than a general-election electorate: more ideologically committed, less representative of the broader workforce.

That could help explain the rise of more progressive leadership in some large urban local unions, he said, citing Chicago as an example while noting some of those trends predated Janus. Because labor law makes unions relatively durable institutions — decertification is possible but difficult — the institution can persist even as the active membership changes. As progressives invest in those institutional vehicles, unions may increasingly take up issues beyond wages, benefits, and classroom conditions.

Hartney listed Israel-Palestine, institutional racism, unaffordable housing, and other broad political issues as examples of what some local unions are now discussing. He contrasted that with the early NEA and AFT focus on raising teacher pay and making the profession attractive. Whatever one thinks of the politics, he said, that is a very different institutional posture.

Lyon agreed that non-education activity deserves more attention, but she framed it somewhat differently. Because civic organizations representing working people have declined, teachers’ unions may be among the few remaining local institutions with organized capacity and an interest in economic policy. That makes their role in broader local politics important to study. She said researchers and the public should pay attention to how teachers’ unions shape “the whole ecology of education” and how they influence decisions not narrowly tied to schools, such as minimum-wage policy, even though teachers themselves are not minimum-wage workers.

Hartney’s concern was that single-issue interest groups across American politics are being absorbed into the general left-right divide. He compared teachers’ unions to the NRA and ACLU: institutions that once focused more narrowly on guns or speech, but became part of broader ideological coalitions. His worry is strategic as much as normative. If teachers’ unions lean too far into progressive causes and stop endorsing Republicans they could work with, then when Republicans take power the right’s response becomes not negotiation but abolition. The left’s response becomes more resistance. The institution loses cross-partisan room to maneuver.

Hartney closed by arguing that the right often fails to grapple seriously with the need to make teaching attractive. He endorsed Lyon and Kraft’s finding that the profession has declined in status and said teachers need to be paid more. But he immediately tied that to the fiscal tradeoffs unions, districts, and voters face. Education dollars are already committed to pensions, increasingly expensive health benefits, and class-size policies. “You can’t do everything under the sun,” he said.

His example came from school-board elections in California. Hartney said that in research with a co-author, he examined endorsement decisions by the state teachers’ union in school-board races. They found that whether the union endorsed incumbents for another term was related to pay increases given to very senior teachers — the teachers most active in the union — and not to starting teacher salaries. He presented that not as a complete solution but as evidence of an internal accountability problem: union bargaining priorities may reflect the most active members, not necessarily the needs of newer teachers or the profession as a whole.

For Hartney, the remedy begins with transparency. Public information should be available not only to taxpayers but also to union members, so teachers can ask their leaders why they are bargaining for one thing rather than another without needing to spend 12 hours a week in union meetings. Voters should also have better information in school-board elections: the state of the district budget, student achievement, labor agreements, and the implications of candidate endorsements.

He also pointed to election timing. School-board elections are often held at odd times of the year, which can depress participation and magnify organized interests. He described reforms such as better disclosure and election-timing changes as “nudge-like” improvements to public-education democracy — not a wholesale redesign, but changes that could make the system work better for both teachers and the public.