The Surprising and Symbolic Power of Unions

Unions may be powerful in shaping opinion, even if their political influence is on the wane.

It may seem like a long time ago now, but just a few weeks back, the political news was all about Scott Walker beating a recall bid in Wisconsin. At the time, it was the result in Wisconsin and the power of unions (particularly for public workers), not health care, that many commentators proclaimed to be a defining issue of the campaign.

Political scientists often warn that the media tends to over-hype the importance of single events in Presidential campaigns. After all, campaigns are long and complicated, voters are predictable, and public opinion moves slowly. Despite the framing of the Wisconsin recall as a battle over public sector unions, it is easy to see why opinions about unions could actually have been unimportant to the election: after all, it is likely that Walker's supporters are Republicans and union members are Democrats and, therefore, the choice of voters would just be a matter of reliable ol' partisanship, not opinions about a single issue such as public sector unions. Moreover, despite the ridiculous proclamations that the Wisconsin recall was a bellwether for November and, therefore, liberalism in general, Walker's victory was not hard to foresee: Republicans tend to have an advantage in special elections where their partisans more reliably turn out to vote. But this advantage is largely minimized in Presidential elections when Democrats close the gap in turnout.

Yet, despite all these reasons why unions might not matter - a person's attitude towards unions actually has a surprisingly powerful effect on who a person says they will vote for in November. In fact, the effect is so strong, that Republicans might be well-advised to keep the issue alive in the campaign.

On a YouGov survey last month, I asked 1000 American adults about their feelings about public sector unions. A person's opinion about unions has powerful, perhaps surprisingly powerful, relationship with a person's stated vote choice between Barack Obama and Mitt Romney. I say the effect is surprising because opinions about unions appear to be more important than opinions about any other single group.

I asked people to say how favorable they felt towards unions on a "thermometer" ranging from "very cold or unfavorable" (0) to "very warm or favorable" (100). Moving across the range where most voters are found (22 to 80), counts for a 26 percentage point increase in the probability of voting for Obama - or going the other direction, a 26 point decrease (this is controlling for ideology, race, and partisanship). In public opinion, 26 percentage points is a massive effect. To give a sense of the size, consider that a Republican is 45 percentage points less likely to vote for Obama than a Democrat (controlling for the same factors), which means that in terms of your opinion about the President, disliking public sector unions takes you more than halfway from being a Democrat to a Republican.

I also asked people about their feelings towards other groups: Blacks, Latinos, whites, Mormons, Muslims, Gays - and not a single one is as predictive of Presidential vote as is feelings towards unions. This is a wide range of groups -- sort of an equal-opportunity for bigotry -- yet, feelings towards public sector unions beats them all. The graph below is the change in the probability of voting for Obama associated with moving the same distance on a feeling thermometer (22 to 80) for each group. Notice that warmth towards gays, the second most predictive group, does not even have half the effect of warmth towards public sector unions. (Improved warmth towards whites or Mormons makes a person less likely to vote for Obama.)

What explains the power of this association between attitudes about unions and Presidential vote choice?

Political scientists often make a distinction between (and argue about the relative importance of) rational and symbolic considerations. These are the considerations that a person might bring to bear when evaluating political objects, such as candidates and public policies. A rational consideration towards a candidate might be one that a person changes depending on current circumstances -- for example, a person needs a tax break, so they vote for the candidate that promises a tax break (all else equal). A symbolic consideration is more of a predisposition that is acquired early in life and doesn't change. Not only do symbolic considerations tend not to change based on current circumstances, but they tend to overwhelm other considerations. For example, most people have symbolic predispositions towards political parties: so, even if you need a tax-break, if it is offered by a Republican candidate and you are symbolically disposed not to like Republicans, you will find reason not to like that tax break. In the United States, attitudes that are usually identified as symbolic attitudes include partisanship and racial attitudes.

It seems that attitudes towards public sector unions may be of a similar symbolic nature. The power of the union variable does not change in the face of controls, including whether or not a person is a member of a union or has a family-member that is a member. This means that the power of opinions about unions in politics is not just caused by people that happen to be members of unions or know people that are members of unions. In other words, the effect of this attitude seems to be independent of obvious situational factors and other factors, such as whether a person is a Democrat or Republican. These results hint that attitudes about unions are stable and exert a powerful, independent effect on attitudes towards politicians - much like attitudes about race.

Cautions against emphasizing the importance of any single issue still stand, but if there is any persuasion to be had in a Presidential campaign, the campaigns may want to take a good look at what can be done with the surprisingly powerful attitudes about unions.

(For those that care, the union effect remains consistent even in regression models that include the four-item Racial Resentment scale (Henry & Sears, 2002), the grand-daddy of all symbolic attitudes. This, along with intuition and previous scholarship, makes me suspicious that the feeling thermometer for some groups, especially whites and blacks, fails to capture true attitudes because of social desirability bias in responses for Blacks and that a "warmth" measure may not capture existing in-group affect among whites.)

Please read our community rules before posting.
comments powered by Disqus


Ryan D. Enos

Ryan D. Enos is an Assistant Professor of Government at Harvard University and a faculty associate of the Harvard Institute for Quantitative Social Science and the Center for American Political Studies.  He specializes in American politics with an emphasis on the politics of race & identity and voting behavior. Before entering academia, he was a teacher at Paul Robeson High School in Chicago, IL.