Do Election Laws Affect Voter Turnout?

For the last 30 years North Carolina, like most states, has been making it easier to register and vote. In the early 1980s the only way you could register was to go before an official of the local board of elections in person. Since then registration has been expanded to driver license offices and other government agencies, and now is available online. Once you register your name no longer is removed if you fail to vote in several consecutive elections, as used to be the case. Absentee voting previously was available only by mail and was reserved for voters who could not get to their precinct on election day; in recent years “early voting” polling places have been open for nearly three weeks before election day for anyone who wants to cast a ballot in advance. Other changes also have made registration and voting more convenient. In 2013 the General Assembly backtracked on some of those voting changes. The most significant actions were the shortening by one week of the time for early voting and the elimination of same-day registration and voting. Before the new rules take full effect with the 2014 election it may be a good time to ask what difference those election law changes of the last several decades made in whether people voted. Did any of the expanded opportunities to register and vote added since 1980 appreciably affect voter turnout or not? The answer is not clear cut. In tracking voter turnout with election law changes since 1980 it is hard to discern that any particular election law change has made an appreciable difference in how many people vote. Yes, voting in North Carolina in the last two presidential elections was notably higher than before, but other presidential elections have not brought as many people to the polls, and turnout in non-presidential elections is still as low as ever. It may be that the 2008 and 2012 presidential year jumps are simply anomalies, with unusually high interest, and not indicative of any trend. While this review of North Carolina data is not very sophisticated, it is consistent with recent political science research, discussed below, which suggests that early voting may actually decrease voter turnout. The same research shows, on the other hand, that same-day registration and voting can increase turnout somewhat. When done together, early voting and same-day registration neutralize each other. And, as one might expect, factors other than election laws have more to do with how many people vote. The data Data about voter turnout and election law changes in North Carolina from 1980 through 2012 is found on this chart:

(Click  here to enlarge)

The chart shows the percentage of the state’s voting aging population that voted in each general election in November of even-numbered years since 1980. Voting age population is used rather than registered voters because the law on purging the voter rolls changed in the mid-1990s. The data on which the chart is based is here. On the chart each election is marked to indicate whether it included a race for president or the United States Senate, the most prominent statewide elections. A list of all elections may be found here. The election law changes Along the bottom of the chart are notes indicating significant changes in the laws governing registration and voting. The changes are: 1983 — For the first time voter registration became available at some driver license offices and public libraries, rather than voters having to go before board of elections officials. 1987 — Political parties were given the option of allowing unaffiliated voters to vote in their party primaries, which both Democrats and Republicans have done since then. 1992 — The deadline for registering to vote was extended by one week to allow registration up to the 16^th weekday before the election rather than the 21^st weekday. 1995 — Registration by mail became available; all driver license offices started registering voters; and a voter’s registration was no longer purged for failing to vote for two presidential election cycles. 2001 — No-excuse, one-stop absentee voting at early voting sites, popularly known as “early voting,” became part of all elections. 2005 — Online registration became available; all voters were given access to provisional ballots when there are questions on election day about their eligibility; the statewide computerized registration system was established. 2007 — Citizens could register to vote and cast a ballot at the same time at early voting sites during the early voting period. Some observations What do the data show about turnout over the 32 years? The chart is a busy document and offers an opportunity for lots of interpretation. Here are a few observations about the data:

• Not surprisingly, turnout is always higher in presidential election years than any other time. The lowest turnout for a presidential election year was in 1988 when 43.7 percent of the voting age population voted in George Bush’s lopsided win over Michael Dukakis. By contrast, the highest turnout in a non-presidential year was two years later when 41.2 percent voted in the first of the high profile U.S. Senate contests between Jessie Helms and Harvey Gantt. The highest turnout for a presidential election year, 62.9 percent in 2008, is over 20 points higher than that highest non-presidential-year Senate race.

• Turnout in presidential election years varied little between 1980 and 2000, regardless of the closeness of the contests or any election law change. The turnout percentages for those six elections, and the results of the presidential voting in North Carolina:
  • 1980 — 43.9 percent (Reagan edges Carter 49.3 to 47.2 percent)
  • 1984 — 48.8 percent (Reagan routs Mondale 61.9 to 37.9 percent)
  • 1988 — 43.7 percent (Bush routs Dukakis 58 to 41.7 percent)
  • 1992 — 50.1 percent (Bush edges Clinton 43.4 to 42.7 percent)
  • 1996 — 44.4 percent (Dole beats Clinton 48.7 to 44 percent)
  • 2000 — 49.6 percent (Bush beats Gore 56 to 43.2 percent).

• Turnout increased in presidential years after 2000. In 2004 turnout moved up to 54.9 percent of voting age population even though the Bush/Kerry contest was not close (56 percent to 43.6). Then in 2008 the turnout jumped to  62.9 percent as Obama defeated McCain by only 14,000 votes in North Carolina (49.7 percent to 49.4), and in 2012 it was still at 61.8 percent as Romney edged Obama (50.4 percent to 48.4).

• Despite the increases in turnout in presidential years from 2004 on, there was no obvious change in non-presidential years. In fact, the turnout for non-presidential years seems not to have changed much since 1980. More people vote in years when there is a Senate race, but turnout for non-presidential years is consistently low. The turnout for the non-presidential years (and the Senate races on the ballot):
  • 1982 — 30.1 percent (no statewide election)
  • 1986 — 33.6 percent (Sanford v. Broyhill)
  • 1990 — 41.2 percent (Helms v. Gantt)
  • 1994 — 28.2 percent (no statewide election)
  • 1998 — 34.1 percent (Edwards v. Faircloth)
  • 2002 — 37.4 percent (Dole v. Bowles)
  • 2006 — 30.4 percent (no statewide election)
  • 2010 — 37.9 percent (Burr v. Marshall)

Possible conclusions? What conclusions can one reasonably draw from the data? The data seems to support these statements:

• The changes in election law since 1980 do not appear to have appreciably affected voter turnout as a whole. It may be that the voting patterns of some categories of voters have been affected, but overall voter turnout  after the changes does not look much different than before. If making registration available in more places made a difference, one would have expected a noticeable uptick in turnout after the 1995 opening of more registration opportunities and the end of purges for not voting, but that did not happen. And if changes in voting law were critical to turnout, one would expect any increase to occur across all elections, not just presidential elections, but that has not happened either.

• Although  one-stop early voting is enormously popular — a majority of votes in the 2012 general election were cast at one-stop sites before election day — the evidence does not seem to indicate that it increases turnout. In the elections since no-excuse one-stop voting became available for all elections there was no noticeable increase in turnout in the 2002 Senate election; a modest increase in the 2004 presidential election year; the same predictably low turnout in the 2006 election without a statewide contest; a significant increase in the 2008 presidential election; no obvious increase from previous non-presidential elections in the 2010 Senate race; and another high turnout in the 2012 presidential election. If early  voting mattered that much by itself, one would have expected a more consistent pattern of higher turnout over those years. It may be that early voting, for the most part, simply makes it easier for people who would have gone to the polls anyway.
• Turnout  for the 2008 and 2012 presidential elections — but not the intervening 2010 Senate election — was noticeably higher than before, and there must be an explanation. Were the increases simply the result of the dynamics of those two particular elections or did changes in the law have some effect? The explanation for those two elections may be Barrack Obama. It may be the closeness of the contests. It may be the presidential campaigns’ newfound savvy with social media and targeting their supporters to get them to the polls. Or it may be something else. One possibility might be same-day registration and voting.
• The last significant election law change to take place before the high-turnout 2008 and 2012 elections was the introduction of same-day registration and  voting during the early voting period. That is, a citizen who had failed to register by the regular registration deadline, 25 days before the election, could apply to register and could cast a ballot at a one-stop site during the early voting period, up until the Saturday before the election. Because election interest peaks near election day, allowing registration so much closer to the actual election day could be expected to bring to the polls some voters who had neglected to register earlier and otherwise would not have been able to vote.
• Same-day  registration and voting seems to have helped with turnout, though not in sufficient numbers to fully explain the high percentages of voters in 2008  and 2012. About 4.4 million North Carolinians voted in the 2008 election  and about 4.5 million in 2012, and in each year about 100,000 of those voters registered and voted at the early voting sites. If those late registrants were subtracted from the number of voters, the turnout in 2008 would have been 61.4 percent of the voting age population rather than 62.9, and in  2012 it would have been 60.5 rather than 61.8. Assuming that none of those late registrants would have found a way to register earlier, it could be argued that same-day registration and voting at early voting sites increased voter turnout by about 1.5 percent in 2008 and 1.3 percent in 2012. (In 2010, the non-presidential year with a Senate election, only about 22,000 of the 2.7 million voters were early voting registrations, possibly meaning an increase of three tenths of a percent in turnout.) It could be that same-day registration and voting at early voting sites — effectively extending the deadline for registration by several weeks — helped fuel the upswings in turnout in 2008 and 2012, but not nearly enough to explain why those two elections had such higher participation than before.

In sum, based on this limited analysis, the changes in North Carolina’s registration and voting laws over the last 32 years appear not to have made a significant difference in how many people vote. While it has become  noticeably easier to register to vote, and to stay on the rolls, about the same proportion of citizens go to the polls now as before. It seems more likely that voting depends on factors such as the nature of the election, how much seems to be at stake, the level of media coverage, and how much effort candidates and parties put into identifying and mobilizing their voters. The one state election law change that might be argued to have boosted turnout is the 2007 legislation, repealed in 2013, to extend the  deadline for registering by allowing same-day registration and voting  during the early voting period. With even the most generous assumptions,  however, that change in the law could not be said to account for more than  1.5 percent of turnout. National research This view of North Carolina’s experience is consistent with recently reported political science research on the effects of election law changes. In an article published in the American Journal of Political Science earlier this year researchers from the University of Wisconsin report from a study of all 50 states, perhaps surprisingly, that early voting decreases voter turnout when implemented by itself. They also say that same-day registration and voting increase turnout, and that the combination of early voting and same-day registration and voting pretty much cancel out each other. That is, if a state had only early voting it probably would see a drop in turnout, and if it had only same-day registration and voting it probably would see an increase — and that when a state has both early voting and same-day registration the latter makes up for any decrease in turnout resulting from the former. One of the researchers’ analytical methods suggests that on average early voting decreases turnout by about three to four percent and same-day registration and voting increases it by about the same level. Other analysis produces different numbers but the same general conclusions. Why would early voting decrease turnout? The researchers suggest that early voting mainly benefits people who would vote anyway. They postulate that spreading voting over several weeks, instead of a single election day, reduces the effectiveness of election day as a social event in which election interest peaks. With early voting there is less election day water cooler talk and other buzz that stimulates people to go to the polls. They also see some evidence that when early voting is available political campaigns advertise less right around election day and draw down their efforts to mobilize voters specifically for old-fashioned election day, the Tuesday after the second Monday in November. At the heart of their analysis is a recognition that election law changes do not operate in a vacuum, that any change in registration or voting rules also affects how candidates and political campaigns and the media behave, and that the behaviors of those other actors is far more important in determining who goes to vote. The Wisconsin political scientists’ study is more sophisticated than the analysis reported here for North Carolina, and the questions raised here would benefit from a more thorough study. It would be worthwhile to see, for example, whether the registration and voting changes over the years have affected white and black voters differently. We know that while African Americans make up only about 21 percent of the state’s voting age population they accounted for over a third of the voters using the same-day registration and voting option in both 2008 and 2012. Election changes also affect the major political parties differently. For example, over twice as many Democrats as Republicans used same-day registration in 2008, and nearly twice as many in 2012. We might learn more about the effects of election law changes by exploring these racial and party differences over time. Comments, other explanations? Comments and speculation on the questions raised in this blog post are welcome.