By Peter Miller
The confirmation of Ryan Zinke as secretary of the Department of the Interior has created an exceptional event in Montana: the need for a special election to determine his replacement in Congress. According to the secretary of state’s website, a special election has not been held to fill a congressional seat since at least 1912.
How this special election, scheduled for May 25, will be conducted is proving to be controversial. Senate Bill 305, which would give county election officials the discretion to run the election entirely by mail, passed the state Senate only after the Republican caucus split on the question, with the Democratic members voting uniformly in favor of the bill.
In the floor debate, Sen, Jon Sesso, D-Butte, asserted that “there are absolutely no verifiable facts” that voting by mail advantages one party over another. There are, in fact, a number of academic, peer-reviewed studies on reforms like voting by mail and turnout generally that may be helpful as this bill moves to the state House for further consideration.
Two questions, in particular, were salient in the discussion of this bill in the Senate: will the reform change the balance of power between the parties? And, second, will the reform increase turnout in the election? As someone who has done research in this area, I thought it might be helpful to summarize the academic research on these questions.
The first question that people may be thinking about is whether the reform itself will determine the result of the election and upend the balance of power between the Democratic and Republican parties in the state. In a letter to state Republicans, party chairman (and state representative) Jeff Essmann, of Billings, wrote, “this bill could be the death of our effort to make Montana a reliably Republican state.”
Is there any truth to this claim? Social science research on voting behavior tends to find that a strong predictor of voting is being contacted by a political party or candidate in the course of the campaign. In an early study of the mobilizing effects of absentee voting, J. Eric Oliver found the combination of easy access to absentee ballots and a devoted effort by parties to mobilize their supporters increases partisan turnout.
In particular, he writes, “the Republican party will gain an advantage in low turnout elections by using absentee voting to mobilize its partisans.” That being said, the results Oliver reports may be a consequence of the data from 1992 used in his analysis. A more recent study of voting behavior in the 2008 election found, in contrast, “[t]he supposed Republican advantage with by-mail voting is not evident in our results.”
How can we reconcile these differing effects? The emerging conclusion among scholars seems to be that voting reforms themselves do not have a determining effect on electoral outcomes (presidential primaries may be an exception to this observation according to this paper). Instead, these reforms define the context of interparty competition to persuade and mobilize supporters. The party that is better organized tends to win more of the vote cast prior to Election Day, but this does not, on its own, assure victory.
Hillary Clinton, for example, won a larger share of the the early vote in North Carolina in 2016 than Donald Trump, but lost the state overall (the same result occurred in the Obama-Romney contest in 2012). One of the central values of democratic politics is that elections are pluralist and, usually, not foregone conclusions; there is always a degree of uncertainty about who will win an election.
In a notable lecture in 1909, William James defined the essential character of pluralism as a rotation, “not to lock out any interest forever. No matter what doors it closes, it must leave other doors open for the interests which it neglects.” A defeated party in one election has a strong incentive to try again in the next election while a winning party must be sure to maintain support among the electorate in the next election. As an empirical matter, the places that have adopted voting by mail (e.g. Colorado, Oregon and Washington) are not single-party states nowadays.
What will turnout be in the May election? It is hard to predict with any degree of certainty, given that we do not have set of special elections to compare with this contest. Since Montana became a single congressional district in 1992, average turnout for the House race has been about 70 percent in presidential years and about 58 percent in midterm years. Typically off-cycle special elections have lower turnout than midterm elections. Turnout in Oregon special elections since the adoption of voting by mail is about 21 points lower than in midterm general elections; the gap in turnout in Washington between special elections and midterm general elections is about 8 points. On this basis, perhaps 40-50 percent could be a reasonable “back of the envelope” estimate for turnout in the special election in May.
Will turnout itself determine the winner? We might think that high turnout in an election will mean the Democratic candidate will win. There is little evidence to support this claim (see, for example, this paper and chapter 7 in this book). In fact, one argument for the Democratic dominance in the House between 1955 and 1995 is that the Democratic party did disproportionately well in low-turnout races.
Instead, the scholarly consensus seems to be that high turnout benefits the out-of-power party. A study of turnout in gubernatorial and senatorial elections, for example, supports this point. In Montana, that might mean the Democratic Party can hope high turnout will deliver the sole congressional seat for the first time since 1996. That claim is complicated, however, because the Democratic candidate for the House has won, on average, only 39 percent of the vote since Pat Williams, the most recent Democratic representative, retired from Congress before the elections in 1996.
The results of social science research on the turnout effects of these voting reforms depend on how you ask the question. One early survey of local elections found that voting by mail dramatically increased turnout. This particular study is also notable for being the only instance I have found that uses data from local elections in Montana, in particular comparing turnout in Ryegate and Lavina, to reach its conclusion. Turnout in local elections is typically quite low compared to other elections and may, therefore, exhibit a higher than usual turnout effect after the adoption of voting by mail.
Turnout was 62 percent in the local election in Helena in 2007, when voting by mail was used as a test case, doubling the average returns since 1979. By the 2009 election, however, turnout in the Helena local election dropped to about 45 percent. This trend is consistent with social science research finding that there is a temporary spike in turnout after early voting reforms are adopted, but that the spike vanishes over two or three electoral cycles.
There is disagreement among scholars on the effect of voting by mail on turnout. One study I collaborated onfound turnout in Oregon is unaffected by the reform. A study from Washington State, however, found the reform increased turnout by about 2 to 4 points. One analysis of results from California in 2000 and 2002 found the reform reduced turnout in those elections. If we look only at special elections, though, a positive effect becomes apparent. My analysis of turnout in Oregon (as well as another study) and the results from California show voting by mail increases turnout in off-cycle elections, by 7 to 10 percentage points.
Who are the people most likely to vote early? Essmann’s letter argued that voting by mail “is designed to increase participation rates of lower propensity voters” (emphasis added). It is an open question, however, whether voting by mail actually increases the participation rates of lower-propensity voters. Indeed, one interpretation of the lack of consistent turnout effects of voting by mail is that the reform only caters to voters who would have cast a ballot at the polls on Election Day regardless of the election reforms in place.
An added complication is that Montana also allows for election-day registration (called same-day registration in some areas). This reform substantially increases turnout, but does not reduce the long-standing bias in favor of those with higher status in the American electorate (see this paper, this paper, this paper, and this book). It may be the case that election-day registration, and not voting by mail, is the mechanism that increases voting among lower-propensity voters. The research on this point is unclear and further complicated by the move in Oregon and other states to move toward automatic registration in the 2016 elections.
I’ll conclude with a point about research design. In the floor debate on SB305, Sen. Fred Thomas., R-Stevensville, argued, “I think Montanans should be voting the same way, however it’s done. Whether this bill becomes law or not, Montanans should be voting the same way.”
Let’s imagine, for the sake of argument, that every county adopts voting by mail for the special election. How would we then know that voting by mail itself increased turnout in the election if there is no control group to compare with the counties that adopted voting by mail?
The 1997 Washington referendum on a new stadium for the Seattle Seahawks gives a nice example of evaluating the effects of voting by mail. In that election, 27 of the 39 counties used voting by mail, allowing comparisons to be made across the counties to see if voting by mail increased turnout. In a post-election report to the Washington State Association of County Auditors, then-Thurston County Auditor (and now former Secretary of State) Sam Reed wrote, “59% of the eligible mail voters participated while 34% of the eligible pollsite voters participated. Since King County was an obvious anomaly, looking at the other 38 counties is more revealing: 56% of the mail voters participated and only 21% of the pollsite voters participated. In other words, mail voting almost tripled the turnout.”
In the event that SB305 becomes law and some counties choose to retain polling places, it will be interesting to replicate this analysis to see if the effects found in Washington are also present in Montana.
Billings native Peter Miller completed his doctorate in political science in 2013. He is currently a Fulbright Scholar at the University of Tampere in Finland. Any views expressed in this post are his alone and are not necessarily reflective of the views of the U.S. government, the government of Finland, the Fulbright Foreign Scholarship Board, or the Fulbright Finland Foundation. This post was prepared without input or review from any of these organizations.