During the recent debates over public employees’ collective bargaining rights, especially around the Wisconsin protests, I heard a few people argue that Republican governors are intent on destroying public sector unions, at least in part, because union members are more likely to vote – and to vote Democratic.
The latter argument (union members are more likely to vote Democratic) is generally true (also here) – although the union “effect” on candidate/party choice is of course complicated. The former argument (more likely to vote in general) is also valid, but there is some underlying public/private variation that is both interesting and important.
As is almost always the case, isolating the effect of a given factor (in this case, how being a union member affects the likelihood of voting) requires one to compare how this factor “operates” on people who are otherwise similar. For example, in a previous post, I compared public and private sector workers’ earnings. In order to uncover the “effect” of public sector employment on earnings, I used models that controlled for other relevant, measurable factors, such as education and experience. In doing so, I was able to (imperfectly) ensure that I was comparing public and private employees who were similar in terms of skills and qualifications.
The same basic concept applies to voting.
Since different “types” of people are more likely to vote – e.g., older, more educated, higher-income, home-owning individuals – one must attempt to account for these other factors (e.g., using logistic regression), in addition to geographic and institutional conditions (e.g., state), to see the “true effect” of being a union member. If you simply compare union members with non-members, this would be misleading, since the two groups differ, on average, in terms of other characteristics that are known to affect voting behavior. (Side note: See this fascinating paper on the relationship between income and voting.)
The literature on electoral turnout is well-developed, and does include analyses of the “union effect” (for example, see here, here, here and here). Although effect sizes vary between these studies, most find that union members are indeed more likely to vote than their non-union counterparts, even controlling for other influential characteristics and factors. This finding usually holds for both public and private sector employees, but the degree is rather different.
Among private sector workers, being a union member increases one’s likelihood of voting substantially. For example, a paper published last year found that, all else being equal, the predicted probability of voting among private sector union members was around 62 percent, compared with roughly 56 percent among their non-union counterparts. That’s a pretty big difference.
The story among public employees, however, is a little different. The above-mentioned 2010 paper, for instance, found that, holding all other factors constant, the probability of voting was about 64 percent for non-union public employees, and 66 percent for their unionized peers. The difference was significant but the effect size was smaller – 2-3 percentage points – about one-half the size of the union effect in the private sector.
To understand why this difference might arise, think about how unions increase turnout. They do so in many ways, but principally by removing barriers to – or reducing the “costs” of – voting. Unions mobilize – they provide their members with information about when to vote, where to vote, the candidates’ positions on issues of importance to members, and which candidates the unions have endorsed. They encourage members to register ahead of time, and help them get to the polls on election day. But it’s not just time and money – by providing a sense of group action and empowerment, unions also reduce the “psychological costs” of electoral participation, thus helping to overcome apathy and other mental barriers.
Public employees bear fewer of these costs, regardless of whether or not they are unionized. They are, on average, more senior and better educated than average voters, and less likely to be single parents, which may give them more opportunity to get to the polls (the analyses above may or may not control for these and other measurable factors). By virtue of working for some level of government, they are also more likely to think that government and the selection of its leaders are important – and are more likely to be well-informed about local polling places and candidates. In short, their “initial costs” are lower than those of private sector workers, so there is less “room” for unions to exert their influence.
In terms of voter participation, this means that there have been consequences to the long-term decline in private sector union membership. Public sector union density has held fairly constant over the past 30 years (at roughly 35-40 percent), but private sector membership has dropped dramatically, from around 20 percent of all workers in 1980 to about six percent in 2009 (membership data here). As a result, while public sector unionization (and its turnout-increasing power) has at least remained constant (and significant), the influence of organized labor as a whole on voter turnout has declined. The private sector workers who were most “susceptible” to the participation-inducing effects of unions are no longer members, which has limited the degree to which the labor movement en masse is a potential force for getting-out-the-vote on election day.
Bear in mind, though, that union members have a positive influence on turnout among non-members (also here), through mechanisms such as social ties (e.g., family, friends) and public organizing campaigns, and the “union effect” may also persist among former union members. In addition, unions tend to target their efforts in specific states or districts (e.g., for competitive races), and the overall probabilities above mask the fact that turnout effects are much higher in these areas (this is borne out in studies of specific elections, in which the effects of union efforts, especially personal contacts, are rather powerful).
Nevertheless, unless they’re misinformed, governors like Scott Walker aren’t taking aim at public employee unions solely because of their effects on turnout among their members (they are more likely seeking to limit labor’s clout on the donations-to-candidates and on-the-ground organizing sides of the equation). But the reaction to Walker’s attempt to restrict collective bargaining – the protests, which also have sprung up in several other states considering anti-union legislation – certainly threw an interesting monkey wrench into the union/turnout calculus.
Judging by the results of Wisconsin’s spring election, in which turnout almost doubled from the last comparable cycle, allowing a virtually unknown candidate almost to defeat an long-term incumbent, the voters of Wisconsin are energized, and promise to turn out in droves for 2012 (perhaps earlier, for recall elections). So too might pro-labor voters – members and non-members – in other states.
If they do, we might say that the turnout-increasing power of public employee unions has been significantly enhanced not by their existence, but by the efforts to destroy them.