Last Friday we released a policy brief on SB 421, the election reform bill that that is currently being debated in the House of Representatives. Our policy brief went as far as we could then in analyzing the impact of the legislation on access to and ease of voting in the state. In this blog post we want to report on more recent analysis that strengthens our conclusion that the genuine reforms in the bill will increase voting by Pennsylvanians, including Pennsylvanians of color and in down-ballot races, more than the loss of straight-ticket voting will reduce them.
The main focus of our earlier brief was on the elimination of straight-ticket voting and we put forward some data that suggests that the drop-off (also known as the roll-off) from top of the ballot races to down-ballot races, such as those for the General Assembly, is likely to increase if straight-ticket voting is eliminated. We also suggested that voters who have low-incomes or are Black and Hispanic are somewhat more likely to drop off disproportionately to their share of the electorate. Others have made similar arguments and also pointed out that eliminating straight-ticket voting might lead to longer lines at the polls, especially in communities that have historically used straight-ticket voting.
We also pointed out that moving the registration date closer to Election Day and the creation of a vote by mail option is likely to increase turnout in Pennsylvania elections. And since the vote by mail option allows people to vote before Election Day, it will reduce lines at the polls.
Many people who study elections agree with both propositions. The difficult question is about the magnitudes of both effects. And estimating them is very difficult.
We provided some estimates of the impact of the elimination of straight-ticket voting because we were able to find some data that allowed for such an analysis.
It’s even harder, however, to estimate the impact of earlier registration and voting by mail. The practices of voting by mail and early voting have not been found in the United States for long, so there are only a limited number of studies of their impact. Even worse, the studies that have been done have shown varied results. A recent literature review conducted by the General Accounting Office found 21 studies of vote by mail. Fifteen of those showed that vote by mail increased turnout, three found no statistically significant effect of introducing vote by mail, and three found that vote by mail reduced turnout. Of the ten studies that found that vote by mail increased turnout, the range of the effect varied from a half a percentage to a 15 percentage point increase in turnout. There is also some evidence that vote by mail and early voting are likely to have more impact on voters who have lower incomes or who are Black or Hispanic. Given the anecdotal reports we have heard from those who engage in political campaigns in the states that have vote by mail, we are inclined to believe that whether vote by mail will have positive effect on turnout or not depends on how the programs are implemented, how much before Election Day voters can send in or submit their ballots, and whether candidates or independent advocates are effective in taking advantage of the opportunities created by vote by mail as well as the early voting that goes along with it.
The evidence about moving registration closer to Election Day is a bit more unequivocal. Almost all of the studies show that it does increase registration. And the three studies that analyze the effect of moving the registration deadline ten days closer to Election Day show an increase in registration by .5 to 3 percentage points.
Given the uncertainty about the impact of vote by mail and a later registration date on the number of voters, it is almost practically impossible to estimate the impact of adopting these policies in Pennsylvania. So we decided to ask a different question: How much would registration and turnout have to increase in order to overcome the loss of votes we predicted due to the elimination of straight-ticket voting? We were not able to answer this question for all elections to both House and Senate. But since we found that the biggest loss of votes was for House elections in presidential election years, we carried out that analysis first. We found that if the combination of changing the registration date and vote by mail increased turnout on average by a bit more than 4%, the gains in down-ballot races would be greater than the losses we projected.
So would a 4% increase in turnout be possible or likely if SB 421 were to come into effect? The answer cannot be certain, but given the review of the evidence we cited above, we think the answer is more likely to be yes than no.
We should also point out that mail-in voting and an extended date to register to vote will increase turnout in the presidential election and other top-of-the-ballot races for all Pennsylvanians including Black, Hispanic, and low-income voters.
We say all this despite our continued opposition to eliminating straight-ticket voting. We understand that there is give and take in any political compromise. But even though the benefits outweigh the costs in this compromise, we regret that eliminating straight-ticket voting is part of the legislation. As we argued in our policy brief, we believe that is a tool that is useful in overcoming the structural barriers to voting of class, race, and ethnicity. It would have been better not to include the elimination of straight-ticket voting in the legislation and it would be better to delay the implementation of doing so. And we intend to encourage the General Assembly to restore straight-ticket voting as soon as possible.
Vote by mail and a later registration deadline should be supported by everyone who believes that we should make it as easy as possible for people to vote. It is a sad commentary on Republican priorities, which has been to make voting harder rather than easier, especially for people of color and those with low incomes, that they will only support them as part of legislation that eliminate straight-ticket voting.
 Erik J. Engstrom and Jason M. Roberts, The Politics of Ballot Choice, Ohio State Law Journal, Vol. 774, pp. 839-865. See also David C. Kimball, Chris Owens, and Matt McLaughlin, Straight Party Ballot Options in State Legislative Elections, paper presented at the second annual conference on State Politics and Policy, Milwaukee, WI, May 24, 2002.
 Our study used a cross-section methodology in which we compared two states that have straight-ticket voting with two states that do not. While this methodology can be useful, it also allows for other factors that might influence undervotes in state legislative races to affect the results. For this reason, it is generally more useful to do a time-series analysis in which changes in the electoral system are studied over time in a single state. Even here, however, difficulties are found since as we found in our own study that the extent of undervoting in each individual state varied from one election to another time despite there being no changes in electoral laws during this time. Indeed, it’s quite possible that the character of individual elections is far more important than any other factor in determining whether people vote in down-ballot races.
We tried to correct for the difficulties of cross-sectional analysis by matching the states as much as we could. We also made a decision to exclude Arizona, a state without straight-ticket voting, from our analysis because the extent of undervoting in the state made it an extreme outlier. Including Arizona, shows a much more dramatic impact of straight-ticket voting on whether people vote in legislative races than seemed justified to us.
 General Accounting Office, Elections: Issues Related to Registering Voters and Administering Elections, GAO-16-630, June 2106. See pp. 85 and 86 on moving the registration date closer to elections; pp. 93-6 for vote by mail and pp. 97 to 99 for early in-person voting.
 SB 421 allows voters to submit ballots 50 days before the election, which is quite early.
 Our methodology is similar to that in the earlier paper. For each competitive House race in the 2012 and 2016 general elections, we calculate how many additional top-of-the-ticket votes would be necessary to make up for the predicted losses if the elimination of straight-ticket voting were to lead to an undervote rate in Pennsylvania House races equal to the average undervote in lower house races in Georgia, Florida, and Ohio. We assume that the predicted undervote rate is not changed by the introduction of vote by mail and a later registration deadline. We then take an average for the competitive House districts in 2012 and in 2016 and the average the results for the two elections. Competitive districts are defined as those in which the Democratic share of the two-party vote is between 45% and 55%.