The union label on the ballot box: how school employees help choose their bosses.

Moe, Terry M.

Fifteen thousand strong, school boards are among the most numerous of this country's governmental institutions. Within the framework laid down by state and federal law, they are responsible for much of what happens on the ground in American public education. They build schools, select textbooks, design curricula, recruit teachers, award diplomas, set rules for discipline, and oversee a vast array of operations, plans, and policies that shape the education experiences of most American children.

From their origins in the 19th century until the present day, school boards have been regarded as shining examples of local democracy, the keystone that links public education to ordinary citizens. But this is one of the enduring myths of American folklore. The reality is that, while some 96 percent of school boards are elected (according to data collected by Frederick Hess of the American Enterprise Institute), these elections are usually low-turnout, low-interest affairs in which the vast majority of ordinary citizens play no role at all. Special interests, well organized and largely unchecked by the public, often have ample opportunity to engineer outcomes in their own favor.

This is not a good thing for children or schools, but there is nothing surprising about it. Americans are apathetic about almost all aspects of politics; they're just more apathetic about school-board politics. School-board elections are often held at odd times, when no other offices--particularly major ones, like president or governor--are being voted on. Moreover, roughly two-thirds of registered voters are not parents of school-age children and so have only weak incentives to pay attention or participate. To make matters worse, the vast majority of these elections, about 89 percent (according to Hess), are nonpartisan; and without party labels to guide them, most voters have no information about the various candidates running for multiple board seats, and so are confused and even more uninterested than they would normally be.

Who Cares?

But apathy stops at the schoolhouse door. One group of local citizens--teachers and other employees of the school district--has an intense interest in everything the district does: how much money it spends, how the money is allocated, how hiring and firing are handled, what work rules are adopted, how the curriculum is determined, which schools are to be opened and closed, and much more. The livelihoods of these people are fully invested in the schools, and they have a far greater material stake in the system than do any other members of the community.

As individuals, then, district employees have strong incentives to get involved in school-board politics and to take action in trying to elect candidates who will promote their occupational interests. The things they want are simple and straightforward--and have nothing to do, at least directly or intentionally, with quality education. They want job security. They want higher wages and fringe benefits. They want better retirement packages. They want work rules that restrict managerial control. They want bigger budgets and higher taxes.

School employees have the additional advantage of being well organized. Unlike parents and other citizens, who are typically atomized and ineffectual as political forces, most school employees are represented by unions. Many of these employee unions get engaged in school affairs. But among them, the teacher unions are almost always the most active and powerful, and they generally take the lead in championing the cause of employee interests in politics.

In school-board elections, the incentives of the teacher unions are strong and clear. If they can wield clout at the polls, they can determine who sits on local school boards--and in so doing, they can literally choose the very "management" they will be bargaining with. (Private sector unions, which square off against independent management teams, can only dream of such a thing.) These same elected board members, moreover, will make decisions on a gamut of policy issues, from budgets to curriculum to student discipline, that teachers have a stake in and can benefit from enormously. Under the circumstances, it would be irrational for the unions not to get actively involved in school-board elections.

They have the resources, moreover, to do just that. While unions are nominally collective bargaining organizations, they can readily turn their organizations toward political ends. They also have guaranteed sources of money (member dues) for financing campaigns, paid staff to coordinate political activities, and activist members to do the invaluable trench-work of campaigning. For these and related reasons, the unions have major advantages over other groups, which can often translate into electoral power.

These advantages also apply in urban settings, where the unions have lots of potential competitors: business, community, ethnic, and religious groups that could (and sometimes do) get involved in school-board elections. Even when these groups are well organized for political action and flush with resources--which is usually not the case--they almost always have social and political agendas that reflect a wide spectrum of public issues, not just education, and they divvy up their resources accordingly. The teacher unions, by contrast, have a vested interest in public education--and only public education--and that is where they focus all their resources and attention.

This doesn't mean that the unions always prevail over other constituencies, nor that they are a dominating political force in all districts. Later in this article, in fact, I'll discuss several basic conditions that place limits on union influence. Still, in the normal course of events, teacher unions tend to have important advantages relative to other groups in both incentives and resources--so, that over the long haul, they often (but not always) succeed in getting their favored candidates into office. As a result, there is good reason to be concerned that the local governance of schools tends to be more responsive to the interests of teachers (and other school employees) than a focused concern for quality education--and the interests of children--would warrant.

Although union power in school-board elections would seem to have vast consequences for public education, it is a subject that is rarely studied. Over the past several years, I have been engaged in a project that tries to do something about that, and I am now in the process of writing articles that present the findings. The findings offer basic, much-needed evidence on what the unions actually do in school-board politics, how successful they are, and what strengths--as well as weaknesses--are most important for an accurate, balanced understanding of their roles in education and its politics.

Here I want to present the results of one of these studies, which focuses on a particularly interesting way that the teacher unions can attempt to influence election outcomes. As I suggested earlier, the unions have many means of influence at their disposal: they can contribute money to candidates, they can unleash their activists to make phone calls and distribute literature, they can pay for advertisements, and so on. But another weapon in their arsenal is the voting power of teachers themselves. If teachers vote at higher rates than ordinary voters, if their allies in other unions do the same, and if ordinary voters turn out at their usual low rates, then employee-favored candidates clearly ought to have a systematic advantage. It may not be enough, all by itself, to win the election for them. But when combined with the other union weapons, it may contribute to a winning union strategy.

More specifically, this study of teacher turnout brings evidence to bear on three central questions. First, do teachers and other district employees vote at higher rates than other citizens? Second, are they turning out for reasons that are essentially public spirited, or are they turning out to promote their own occupational self-interest? And third, are the turnout differentials (if any) great enough to be of any consequence in boosting the unions' chances of victory?

A Study of Teacher Turnout

As part of the larger project, I gathered data on the names and zip codes of school district employees in a stratified sample of 70 California school districts, all of them unionized, and I matched these names to county voter files to get each employee's voting history. In the study I'm describing here, I restrict my attention to nine of these districts, all located in Los Angeles and Orange counties. These nine are analytically useful because they are clustered in close proximity to one another, and teachers who don't live in the district where they work often show up as residents of one of the other districts. Being able to compare these two types of teachers--those who live and work in a district, and those who live in one district but work in another--is quite helpful in understanding the basics of teacher turnout, as well as its connection to power.

If this were an analysis of national or state elections, we could go directly to an investigation of turnout, the presumption being that turnout is a measure of electoral clout. Yet when we look specifically at teachers in school-board elections, turnout is a second-order issue. The first-order issue is whether teachers live in the districts where they work, because if they don't, they aren't even eligible to vote. Obviously, this has a lot to do with whether turnout can translate into power. The data show that the percentage of teachers who live in their own districts varies a great deal--from 8 percent to 55 percent in this sample--and tends to increase with the affluence of the district. Even in the more affluent ones, however, a strikingly large percentage of teachers in this sample do not live where they work and thus cannot vote. Other district employees are much more likely to live where they work, regardless of the district's affluence. This enhances their value to the teacher unions as political allies.

It is unclear how representative these findings are of districts generally. Living outside the district is most common when multiple districts are packed into an urban area, as they are in Los Angeles County and Orange County. In districts that are suburban, rural, or geographically spread out, far fewer employees may live outside their own districts. Still, some degree of nonresidency is probably a fact of life in most districts. And to the extent it is, teachers and their allies should have a harder time translating their own turnout into power.

The Turnout Gap

Two types of elections are most relevant to turnout: school-board elections and bond elections, both of which are nonpartisan, meaning that candidates do not carry a party label. For school-board elections, I focus on those that are held during odd years, when there are no general elections for federal and state offices. These elections offer the best opportunity for studying how teachers and other district employees act on their job-related incentives, because little else is being voted on. For bond elections I focus on those that are not held at the same time as general elections or school-board elections.

The data show that turnout among the local population is downright abysmal, even in the more affluent districts. In the off-year school-board elections for which I have data, 1997 and 1999, the median turnout of registered voters is 9 percent, as can be seen in Figure 1a. This percentage would be even lower, obviously, if the denominator were the voting-age population as a whole, for many people in the electorate--about a quarter--are not even registered. For bond elections (1998-2000), the turnout is 23 percent (see Figure 1b). In both cases, low turnout gives the unions an opportunity to mobilize support and tip the scale toward candidates they favor.

Do teachers vote at high rates compared with average citizens? The answer is clearly yes, as Figures 1a and 1b illustrate. Indeed, if we compute the turnout gap between teachers and average citizens in each district, the median gap over all districts and elections (both school-board and bond) was 36.5 percent, which is a huge number given the very low turnout overall. In 1997, for instance, only 7 percent of registered voters in the Charter Oak school district voted in their school-board election, but 46 percent of the teachers who live there did. In Claremont, 18 percent of registered voters went to the polls, but 57 percent of the teachers who live there did. Similar figures can be recited for every district, and the conclusion is the same whether we look at board elections in 1997, board elections in 1999, or bond elections. Teachers who live in their districts were from two to seven times more likely to vote than other citizens were.

Why do teachers turn out at such high rates? The answer may well be that they have an occupational self-interest other citizens don't have. But this claim needs to be tested, for there is clearly a plausible alternative: that teachers are not only better educated and more middle class than the average citizen, but also more public spirited, more committed to public education, and thus more likely to vote in school-board elections regardless of their personal stakes. Can the evidence show that occupational self-interest, and not these other possibilities, accounts for the turnout gap?

The data offer a revealing test. Many teachers in the sample live in one school district but work in another. These teachers are presumably just as middle class, public spirited, and committed to education as other teachers are; but because they don't work in the district where they live, they do not have an occupational stake in their local school-board elections. Will these teachers vote at the same high levels as teachers who do have such an occupational stake?

Whether we look at the 1997 elections, the 1999 elections, or the various bond elections, the answer is the same: in every case that allows a comparison, the teachers who live in a district but don't work there vote at lower rates than the teachers who both live and work there. The size of the difference is almost always substantial (and statistically significant). In Claremont, to take a rather typical example, 57 percent of the teachers who both live and work there voted in the 1997 election, but only 23 percent of the teachers who live but don't work there voted.

A corollary issue is whether teachers who live in a district where they don't work vote at higher rates than ordinary citizens do. Here the answer is less clear, and the low numbers advise caution. Statistical significance aside, these teachers turned out at higher rates than ordinary citizens in 12 of 18 elections, but in 5 they actually turned out at lower rates. Of the cases when they turned out at higher rates, moreover, only six are statistically significant. Across all the school-board and bond elections, the median difference in turnout rates between these teachers and ordinary citizens is just 7 percent, which could be simply due to social class.

Taken together, these findings contradict the idea that the teachers who live and work in a district turn out at high rates because they are public spirited, committed to education, or socially advantaged; they bolster the notion that self-interest is in fact mainly responsible. A plausible addendum, however--although I do not have the data to explore it--is that teacher turnout is getting a double boost from self-interest: one because the teachers themselves have an occupational stake in voting and another because their unions have a self-interest in mobilizing them. It seems likely that both are at work, and that the turnout differential is not solely due to the incentives of individual teachers.

Valuable Allies

Now consider the other district employees. This is a heterogeneous group that includes administrators, nurses, and librarians, as well as janitors, secretaries, cafeteria workers, and bus drivers. The low-paid members of this group, however, far outnumber the high-paid members, and some 40 percent are Hispanic. On class grounds alone, therefore, we would expect these employees to vote at much lower rates than teachers. In more affluent districts (and perhaps others), they should also vote at lower rates than ordinary citizens.

These class-based expectations are quite wrong. In every district with available data, and for all three sets of elections, other district employees who live and work in their districts vote at substantially higher rates than ordinary citizens do--rates that, on average, are just a shade lower than those of teachers who live and work in the district. The median difference in turnout rates between them and the teachers who live in their own districts is just 4 percent, which is stunningly small given the underlying differences in social class. Clearly, something other than class is at work here. And that something is probably that these other employees, just like teachers, approach elections with their own self-interest in mind, and their unions mobilize them on those grounds.

This interpretation is bolstered by the fact that, when we look at other employees who live in a district but don't work there, and thus do not have an occupational stake in the elections, their turnout proves to be decidedly lower on average than that of other employees who both work and live there. The former turn out at lower rates in all of the 16 cases for which there are data, and 13 of these are statistically significant. For all elections, the median difference in voting rates between the two groups is 20 percent, and it is not uncommon for the gap to be much larger.

As was true for teachers, the other employees who live but don't work in the district tend to look pretty much like ordinary citizens in their turnout rates. The median difference is 8 percent, which is virtually the same advantage we found for teachers. In this case, though, social class obviously does not explain the turnout gap. And because this is so, it is reasonable to suspect that it doesn't explain the differential between teachers and ordinary citizens either. Some other common factor probably accounts for both differentials.

What these teachers and other district employees have in common is that they both take a self-interested approach to elections and they both belong to unions. Because they don't work where they live, they have less incentive to vote and they are not mobilized by the local union (to which they don't belong). But they may also recognize--with reminding by their own unions--that they are all enmeshed in a big collective-action problem, and that they should vote in their home districts to protect one another's jobs and interests. Because voting is not a very costly act, this could easily account for a turnout rate that is 7 to 8 percent above that of ordinary citizens.

This analysis reveals that turnout can be an important resource for teachers and their unions. Teachers turn out at much higher rates than other citizens do, they act on their occupational self-interest, and exactly the same is true of the other district employees. This makes them key political allies and essentially allows the teacher unions to double their voting strength. There is also a downside, however, that weakens their ability to convert these advantages into electoral power. This is the problem of residency. The high turnout rates and the driving force of self-interest are of political value in school-board elections only to the extent that teachers and other employees live in their districts. And many do not.

Slim Margins

Because of the residency problem, turnout is unlikely to be as potent a resource as money or political activism in producing electoral victory. But it can contribute in a positive way to the larger union effort, and in some cases--when elections are close--it can even be pivotal. These cases may be fairly common, in fact, because the margin of victory in school-board elections is often rather small. By my own estimate (based on a separate sample of 245 districts for another study), the median gap between the best-off losing candidate and the worst-off winner is about 3 percent. Thus, in many elections it doesn't take much of a vote swing to change the outcome.

Consider some rough calculations for the Charter Oak school district. In the 1997 election, three candidates competed for two seats. The total number of votes cast (two by each voter) was 3,506, and the margin of victory was 2.54 percent, or 89 votes. Are the turnout differentials in Charter Oak large enough to overcome an 89-vote gap and bring victory to a union-backed candidate? The answer is yes. The district had a total of 350 teachers, only 22 percent living in the district and voting at a rate of 46 percent. Thus there were 35 teacher-voters. The district also had 354 other district employees, 50 percent living in the district and voting at a rate of 41 percent. This means that there were 73 voters among the other employees, and, when the teachers are added in, 108 total votes by school personnel. This figure alone exceeds the 89 votes needed for victory, and it makes no allowance for other sources of pro-union votes (such as relatives, friends, or neighbors). Similar calculations could be carried out for the other districts, showing that the turnout differential alone is often sufficient to overcome the margin of victory, or at least comes close.

As I said, these are indications of what can happen in elections that are close, as many are. Not all elections are this close, of course. And we can't really expect all employees to vote as a bloc (although the prime role of occupational self-interest certainly promotes such an outcome). Yet these sorts of calculations help to show that high employee turnout rates can indeed boost the prospects for union victory, even when considerably diluted by the residency problem. And even when turnout is not pivotal to the outcome, it is clearly a resource that works to the unions' advantage, contributing to their larger effort to control electoral outcomes.

More Evidence

I can't report in detail on the rest of the research project, but the findings to date point to two very general themes.

The first addresses the obvious bottom line. It is one thing for the unions to have a capacity for power through the various resources that they control, but it is quite another for them to put that capacity to effective use--which, in the end, is what really counts. The question we ultimately need to answer is: to what extent are unions successful at getting their favored candidates elected to office?

The answer is that they are quite successful indeed. The most direct evidence comes from a study of 245 California school district elections and the 1,228 candidates who competed in them during the years 1998-2001. A multivariate statistical analysis shows that, for candidates who are not incumbents, teacher union support increases the probability of winning substantially. Indeed, it is roughly equal to, and may well exceed, the impact of incumbency itself.

The comparison with incumbency is instructive. These are low-information, low-interest elections, and because incumbents tend to be well-known, effective campaigners, and relatively well funded, there is every reason to expect the power of incumbency to be considerable. My statistical estimates show that it is. That the estimates for union impact are comparable, then, says a lot about the lofty level at which the unions are playing the political game. They are heavy hitters.

Their total influence, in fact, appears to be even greater over the long haul. When the unions succeed in getting non-incumbents elected to school boards, these people become incumbents the next time around. Then their probability of victory is boosted not just by their union support, but also by the power of incumbency. When the two factors are combined, as they are when union winners run for reelection, the candidates are virtually unbeatable.

Another study, based on interviews with 526 school-board candidates (winners and losers) in 253 California districts, reinforces the study of electoral outcomes (see Figure 2). The interviews suggest that the teacher unions are typically the most powerful participants in school-board elections and that their power is common across districts of all sizes (and not restricted to large urban districts). They also provide evidence that union electoral clout has genuine substantive consequences: the candidates supported by the unions, as well as the candidates who win, are considerably more sympathetic toward collective bargaining than the other candidates.

Union-backed candidates are more likely to believe, for example, that collective bargaining promotes good teaching, fosters professionalism, and helps to raise academic performance, and they are inclined to take a more positive view of unions and their activities. With board members of this type, and thus with "management" teams they have helped to choose, the unions are in a good position to get board decisions on personnel, policy, and other governmental issues that are responsive to their interests.

This study of candidates, however, also provides evidence for a second important theme about union power: namely, that the unions operate under constraints that limit what they can achieve. Yes, they are powerful, but they don't always dominate, and they can't have everything they want. In particular:

* They sometimes face opposition from other organized groups, especially in large urban districts. When this happens, business groups are the most likely to represent effective opposition.

* Because incumbents have their own bases of power, they can be more difficult for the unions to defeat than other candidates. As a result, the unions sometimes support incumbents who are not as pro-union as the unions would like in order not to alienate an eventual winner.

* Because voting patterns are shaped by the political culture of a district, unions in conservative districts sometimes find themselves supporting candidates who are less pro-union than they would like in order not to lose.

* After election to the school board, the experience of being on the board--and part of "management"--seems to make members somewhat less pro-union over time; as a result, the unions cannot count on gaining complete control of school boards even when they are continually successful in elections.

It would be extreme, then, to say that the unions totally dominate their school boards. But there is still a serious problem. School-board elections are supposed to be the democratic means by which ordinary citizens govern their own schools. The board is supposed to represent "the people." But in many districts it really doesn't. For with unions so powerful, employee interests are given far more weight in personnel and policy decisions than warranted, and school boards are partially captured by their own employees. Democracy threatens to be little more than a charade, serving less as a mechanism of popular control than as a means by which employees promote their own special interests.

Terry M. Moe is professor of political science at Stanford University and a senior fellow at the Hoover Institution. The studies presented here are adapted from an article in the Spring 2006 issue of the Journal of Law, Economics, and Organization and from Besieged: School Boards and the Future of Education Politics, edited by William G. Howell.

Civic Duty, Redefined

Teachers and other district employees are much more likely to vote in
school elections than ordinary voters, especially if they work in the
same district where they live.

(Figure 1a)

Voter Turnout for School-Board Elections in Selected Los Angeles County
School Districts, 1997, 1999

 Percentage Turnout

Registered voters 9

Teachers:
Only live in the district 20
Live and work in the district 46

Other Employees:
Only live in the district 14
Live and work in the district 35

Note: For each category, the percentage represents the median turnout
for such voters over the 14 board elections that occurred in these
districts.
SOURCE: Author's calculations from school-board election data in Charter
Oak Unified, Claremont Unified, Covina Valley Unified, Garvey
Elementary, Montebello Unified, Norwalk-La Marida Unified, and Torrance
Unified districts

(Figure 1b)

Voter Turnout for School-Bond Elections in Selected Los Angeles and
Orange County School Districts, 1998-2000

 Percentage Turnout

Registered voters 23

Teachers:
Only live in the district 18
Live and work in the district 77

Other Employees:
Only live in the district 34
Live and work in the district 75

Note: For each category, the percentage represents the median turnout
for such voters over the five special bond elections that occurred in
these districts.
SOURCE: Author's calculations from school-bond election data in
Claremont Unified (2000), Montebello Unified (1998), Torrance Unified
(1998), Huntington Beach City Elementary (1999), and Santa Ana Unified
(1999) districts

Note: Table made from bar graph.

Looking Out for Number One

Among both incumbents and non-incumbents, school-board candidates with
positive attitudes toward collective bargaining receive more union
support. And, in turn, endorsed candidates are more likely to win
elections.

(Figure 2a)

 Percentage with Positive Attitudes Toward
School-Board Candidates Collective Bargaining
Running for Office Endorsed by teachers union Not endorsed

Incumbent 64 42
Non-incumbent 84 68

(Figure 2b)

School-Board Candidates Percentage Winning Election
Running for Office Endorsed by teachers union Not endorsed

Incumbent 92 49
Non-incumbent 62 22

SOURCE: Author's calculations based on a random sample of 526 school-
board candidates in 253 California school districts (2000-2003), as
reported in "Teachers Unions and School Board Elections," Besieged:
School Boards and the Future of Education Politics (Brookings, 2005),
ed., William G. Howell

Note: Table made from bar graph.