The “quants” have taken quite a hit this year, most notably Nate Silver’s mea culpa. I’m not going to summarize the discussions song and verse, I’ll just refer people to the excellent commentators at Huffington Pollster (political science PhD!) and Monkey Cage, among others. Where they differ from many media outlets is that they almost never trumpet the result of a single poll. The results of a single poll are seldom newsworthy and are much more prone to error.
What is certainly wrong is the kind of muddled, ostrich head in the sand by Virgil and Carl, who have decided that since they stuck their fingers in the air (and by way, were clearly reading newspaper coverage and polls) and did a better job this one year in a few primaries than Nate Silver, that therefore all polls and all quantitative analysis of elections is pure bunk.
That’s hogwash, but what worries me is how many of my friends and colleagues seem to believe this kind of clap trap, and mainly because they are acting like regular old human beings. They remember the polls this year that were off–and therefore newsworthy–without remembering the vast majority that were right on target.
Aaron lists DHM as one of the “losers” in the May primary:
The Portland pollsters not only failed to predict Sen. Bernie Sanders’ win in the Oregon Democratic primary—they missed it by a whopping 28 percentage points.
Yep, that poll result for the Clinton/Sanders race was a real boner, and John Horvick of DHM deserves credit for being up front about the bad estimate.
But just like one poll is not the best way to predict a race, one race within a larger poll is not the best way to evaluate a firm. If you look across all the candidate races that DHM asked about in their May poll, things look a lot different. Mesh focused on the tree–the presidential contest–while ignoring the forest.
In the GOP contests for President, Sec’y of State and Governor’s race, the average “miss” was between 1.7% and 2.4% (all estimates shown below allocate the “don’t knows” proportionally–thus understating any last minute shifts in sentiment). In the mayor’s race, even with a large pool of candidates and a high percentage of “don’t knows,” the average miss was just 1.5%, and Wheeler’s margin was off by just 3%.
Something was going on in the Clinton/Sanders race, but the pattern of other results indicate that it was probably something about that contest, about respondents willingness to provide answers, or volatile sentiments more than it was something about response rates, survey methodology, or firm bias.
More importantly, coverage of the poll points out a weakness in Oregon’s political and media environment–we really have just a single dominant polling firm with only a scattered set of other polls being conducted, mostly by national firms using robo-calls, without many of the detailed questions that a local or regional pollster would ask.
We’d all be better informed, and less likely to focus on a single result, if there were a few more players in the field.
Anyone who wants to spreadsheet used to create these figures, just drop me a line.
Adam Ambrogi and Paul DeGregorio wrote today about “The 5 Principles of Election Integrity.”
The article highlights the challenge for election integrity that is created when we local election officials are chosen by competitive partisan contests:
In most cases, election administrators work hard to be fair and transparent and to promote integrity. But a large percentage of election officials are elected to their offices on a partisan ticket or appointed on partisan basis. This can lead some to believe that these officials will favor one political party over another in their decisions.
This is exactly right–in an era of deep partisan polarization, even ostensibly non-partisan policies can be swept up in partisan competition. Partisan and ideological sorting creates a worrisome feedback loop, where partisans express deep levels of distrust and even anger at any political actor from the opposite party, no matter how anodyne the statement or non-partisan the issue.
I have been collecting data on public perception of election officials using the Cooperative Congressional Election Study for a number of years. In 2010, I asked a series of questions about public attitudes toward local and state election officials, including how they should be elected and whether or not they could be fair in election disputes.
The results are both encouraging and discouraging.
The encouraging result is that local election officials receive high levels of approval when compared to the US Congress, the Supreme Court, and state governors and legislatures. (Higher scores mean a higher level of approval; scores above zero mean that a majority of respondents said they either “strongly approved” or “approved” of the job performance.)
When it comes to how we should choose local election officials, the survey respondents endorse elections, but are more than twice as likely to opt for non-partisan versus partisan contests. This is neither encouraging nor discouraging, but is what I (and I think Ambrogi and DeGregorio) would expect.
It’s encouraging that a plurality of 39% do think that state officials would be fair, but more than half the survey thought they would not be fair or didn’t know.
And among those who answered “no” or “don’t know,” just over one-third thought that election officials would favor their own party.
I wondered if there were partisan differences underlying this second question, and it turns out that there are. As we would expect in an era of party polarization, partisans are worried that elections officials would favor candidates of the other party. But what jumps out to me are the totals in the third column. Democrats and Independents mostly assume that election officials will decide disputes in favor of their own party; Republicans choose this option only slightly less often than “favoring Democrats.”
The good news, I suppose, is that almost 40% of respondents think that election officials will be fair in the case of an election dispute. But among the other 60%, they assume that partisan self-interest is the way disputes are resolved.
As long as partisanship invades election administration via competitive elections, and as long as elections continue to be close (and disputed), perceptions of fairness and integrity of election administration will remain at risk in some quarters.
I’ve experienced a long day of education–or should I say reeducation–at the hands of my friends in the elections community.
Business reply mail allows a sender (“the mailer”) to distributed preprinted First-Class mail to customers. The return postage is paid by the mailer, not the customer. Postage is collected on a per-piece fee only on pieces that are returned.
One colleague in the elections community says this cost is approximately twice the normal first class postage. The rate for BRM is .476 + .011 – .476 + .066 per piece depending on volume (and some account fees).
I have another good friend who is rapidly sawing away, because he tells me that business reply mail does dramatically increase response rates for registration mailers and change of address mailers, so he believes that prepaid postage will increase turnout and increase by-mail return rates. I’m trying to get more information on this…
The cost of requiring postage on all ballots will be the annual account fees plus either .487 or .542 per piece of returned mail. The current by-mail return rate is approximately 40%. There is almost no incentive to switch to actual first class postage until the rate of by-mail return exceeds 90%, and that’s not likely.
What’s interesting is that legislation that would require pre-paid postage might actually create fairly strong incentives to encourage citizens to drop off their ballots.
That may be an unintended consequence of the legislation, though not necessarily a bad one. One common criticism of voting by mail is that ballots leave the hands of government officials; drop boxes eliminate that on the back end.
I leave this to those with much more expertise in direct mailing to point out the flaws in this logic. I’ve learned more about business reply mail in the last two days than I ever wanted to know! (Memories of my days learning about signature verification software …)
More corrections in the next posting.
There are two details about business reply mail that I was not aware of. First, there is no cost to the sender for business reply mail that is not returned. Second, there is twice the postage cost for business reply that is returned.
As Dean Logan once told me, running elections in some jurisdictions is like running a direct mail operation, and some of the details are very specific. I apologize for mistakes in the original version of this posting.
Don Palmer, senior adviser at the Bipartisan Policy Center and previously chief elections officer in Virginia and in Florida wonders about the costs of ballots that are not returned (and I presume also not mailed).
Don uses 140 characters to pose a devilishly complex question. It is easy to figure out the cost of adding postage–just take the number of ballot envelopes (in Oregon, this is roughly equal to the number of registered voters, the only adjustments are for the small number of individuals who have provisions in place not to have their ballot mailed because of court protection orders and similar) a
nd multiply by the cost of postage.
There are no costs at all for unreturned business mail. Thanks to Tammy Patrick at BPC for pointing me to this link explaining business reply mail http://pe.usps.com/mpdesign/mpdfr_brm_intro.asp. From that page:
The Postal Service collects the applicable postage, plus a per-piece fee, only on pieces that customers actually send back to the mailer. This allows mailers to save postage costs on large volumes of distributed reply pieces when a response is not assured.
The rest of this is just an academic exercise. I’m leaving most intact because it identifies all the various categories of voters that may or may not respond to postage paid return envelopes.
For the purpose of this exercise, let’s call the total number of ballot envelopes N, the cost of postage P,
and the “waste” as W.
We know the state will be on the hook for N*P, estimated at $1.2 million per election. (To make the calculations easier to follow, I’ll just assume P=1 for the rest of this posting.)
But how much of this expense is “wasted” (W)? NONE.
I think Don is asking: how many postage paid envelopes are either not returned or are returned into drop boxes? This calculation isn’t simple. Let’s assume that counties will not reduce the number of drop boxes (thus making it harder to drop off ballots) in response to this change.
Let’s start with the 2012 election as our baseline. There were 2,199,360 registered voters in the 2012 election, and 1,541,782, or 82.8% turned out to vote.
This gives us our first piece of baseline number of non-voters in Oregon
Note that this assumes that voter turnout is unaffected by the inclusion of the stamp. I’m not comfortable making any assumptions about turnout, as I detailed in the previous post, but if you agree with advocates that this will help younger voters, lower income voters, and rural voters, then you’d need to make some adjustment to allow for increase turnout
(hence lower waste).
Our second piece of
W non-mail returns comes from the 59% of voters who currently drop off their ballots. If everyone who currently drops off ballots continues to do so, this adds an enormous amount of waste: 1,066,939. (This figures comes straight from the Secretary of State’s report on methods of ballot return.)
That’s a pretty big number, but we have to subtract ballots that are sent by mail in the future because of free postage but were dropped off in the past. Here’s where thing get fun–and hard. Predicting how future behavior will change based on changes to the rules is the essence of modern political science, but is also an exercise fraught with uncertainty.
We reported yesterday the reasons given for dropping off a ballot.
I think it’s fair to assume that the 18.8% who cited “cost” or “no stamp” would mail under the new system. That reduces
W drop offs by .188 * 1,066,939=200,584.
I also think it’s fair to assume that those who cited security concerns (17.9%) or said “I was too late to mail” (12.8%) would not change their future behavior. These total 327,550 (.307*1,066,939) voters who will continue to drop off their ballot.
This leaves us with two sets of voters–those who cited “Convenience” as a reason to drop off their ballot (40.2%) and those who cited “Habit” (6%). I’ll assume both of these groups will change their behavior and use the mail–this increases the mail flow by
reduces W by 46.2%*1,066,939=492,925.
We’re nearly done.
Our current estimate of waste is made of up those who don’t vote (657,578) plus those who have security concerns or who complete their ballots too late to mail (327,550).
The total is 985,128 ballots with business reply envelopes
paid postage that are either not returned or continue to be dropped off. (Mathematically minded readers will notice that there are 5% of the drop off ballots that are unaccounted for–these are individuals who gave no reason for dropping off, or gave a reason that we could not code.)
Just one more complication. Things are never easy!
Automatic voter registration (AVR) in Oregon (known in the state as “Oregon Motor Voter”) throws another wrench into the works.
AVR is projected to add as many as 275,000 citizens to the registration rolls, but it remains very unclear how many of these new registrants will actually vote. This will increase the number of “wasted” ballot materials
s and postage paid ballot envelopes, but by how many?
I think it would be generous to assume that half of these new registrants will vote, and if the same 30.7% vote late or don’t trust the postal system, that will result in another 42,212 wasted ballots.
The main sources of uncertainty in my estimate, for those advocating for and against this legislation:
- The number of individuals who turn out overall may increase as a result of
pre-paidbusiness reply postage. Non-voters are the largest source of waste in any universal ballot delivery system.
- My estimate of turnout among the newly registered (AVR) citizens may also be low. I used a very generous estimate of 50% even though academic research estimates this value to be as low as 25%.
- Campaigns and vote mobilization organizations respond to changes in the rules and the laws, and it may be that these organizations will react to pre-paid postage by encouraging more citizens to vote their ballots early (reducing the “too late” numbers) and will be able to convince voters that the postal service is trustworthy (that’s going to be a much more difficult lift).
That being said, I think an estimate of approximately one million ballots that are not returned in a presidential year is within the right ballpark. In non-presidential years, the numbers will be much higher–15% higher in midterms and 60% higher in special elections.
And let’s be clear–if the focus is solely on “waste”, universal ballot delivery incurs many more costs by mailing ballots to everyone. Oregon officials can tell you how much it costs to prepare and deliver a ballot to every registered voter, but I have been schooled by my friends and now know that
and I suspect that 30 cent (total guess!) business class postage doesn’t increase the per-ballot costs at all by that much for those ballots that are unreturned.
Advocates will assert that it is in these low turnout elections, such as specials, that there is substantially more room for increases in turnout due to
pre-paid business reply postage. The problem is that these elections already are among the lowest turnout in the state even with universal mail delivery.
Senate Bill 1586, which will require Oregon election officials to provide postage paid envelopes for the return of vote by mail ballots, was reported out of the Senate Rules Committee and is making its way to the floor will be discussed in a public hearing tomorrow, Feb. 24.
This is an interesting bill. At first blush, it seems like a reasonable accommodation to make voting easier. But as we’ve learned in many cases, making voting easier often does not increase turnout. Adam Berinsky summarizes the situation nicely:
The problem, I believe, is that when we talk about the “costs” of voting, we have been thinking about the wrong kinds of costs—the direct costs of registering to vote and casting a ballot. Most politicians and scholars have focused reform efforts on these tangible barriers to voting, making it easier for all citizens to vote, regardless of their personal circumstances. But, as I have argued elsewhere, the more significant costs of participation are the cognitive costs of becoming involved with and informed about the political world. Studies of voting from the last 60 years make this point clear. Political interest and engagement, after all, determine to a large extent who votes and who does not.
The specifics of adding a pre-paid envelope can be a lot more complicated and the impact more difficult to predict than they might appear at first blush.
A large proportion of Oregon voters do not return their ballots by mail–they drop them off at county offices or at drop boxes. 56% of Oregon voters in the 2014 Survey of the Performance of American Elections report that they returned their ballots in person, and these data align closely to state figures. The percentage returning by means other than mail have been creeping upwards in recent years, according to the Oregon Division of Elections.
This had led Phil Keisling, the granddaddy of vote by mail in Oregon, to call the system “Universal Vote By Mail”, not voting by mail, because what is key is universal ballot delivery, not return by the postal service. The point here is that prepaid postage will impact a shrinking portion of the current voting population.
In order to understand the reasons for the choice to hand drop better, I coded the 117 SPAE survey responses into four categories: convenience (“close by”, “on my commute”, “easy”), cost (“save money”, “I didn’t have a stamp”, “cheaper”), too late to mail (“last minute”, “make sure it gets there on time”), habit (“Always do it this way”, “In Oregon we can drop it off”), and security (“don’t trust the post office”, “mail theft”, “make sure it’s counted”). I was very generous in coding cost so that even if the reason started with another rationale, any mention of costs, dollars, expenses, “didn’t have a stamp”, etc was coded as cost.
Most voters said they chose to return their ballots by mail because it was more convenient or because they had security concerns. Just under 20%, or 11.8% of all voters (20% of the 59% who dropped their ballots) said they returned their ballots by mail due to cost or inability to get a stamp . 11.8% is not minimal, but these 11.8% also found it possible to return the ballot by other means. Unfortunately, we have no information about those who chose not to vote at all because they couldn’t buy a stamp.
The Oregon Bus Project is advocating for this legislation, and they claim that young people today don’t know where to buy stamps or can’t because it’s too inconvenient:
Because post offices have been closing, many Oregonians must go out of their way to purchase a stamp in order to cast a ballot. Millennials often go to school, work multiple jobs, and are not available to go to the post office during business hours, leaving them with few options.
This claim stretches credulity. Stamps can be purchased at any grocery and many convenience stores, most of which (at least in Portland metro) are open 24 hours. It may be harder for rural voters to purchase stamps, but I suspect that rural voters are much more used to using the postal service on a regular basis.
Even if the state does require paid postage, there’s another wrinkle–business mail is handled differently than first class mail, and this could require other changes to Oregon’s vote by mail system. A study conducted by the California Voting Foundation showed that prepaid postage slowed the processing of ballots because of the way business mail has to be processed and billed (see Section IV.6 of this report http://calvoter.org/issues/votereng/votebymail/study/findings.html#h10) (boldface added)
All three counties have postal accounts to cover additional postage costs (though they don’t advertise it). The “postage due” costs were relatively minimal in all counties, typically amounting to a few hundred dollars in a major election. While some have suggested providing postage-paid envelopes to all VBM voters (and not just those overseas or living in an all vote-by-mail precinct as current law provides), doing so can actually delay VBM ballot processing since postage paid mail is typically sent business class, not first class. In addition, the cost must be debited from the account holder before the mail piece can be delivered. Ensuring postage-paid mail is debited from the correct account adds extra time to ballot processing and can further delay the return of voted ballots.
Moving to pre-paid mail may have to be accompanied by changes in the timeline for vote by mail ballots–when they are mailed out and when they can be returned (possibly including a provision that would only require ballots to be postmarked by election day). There’s nothing wrong with this, but it’s a consequence of the change that needs to be addressed.
Requiring pre-paid postage may be a very reasonable improvement to Oregon’s vote by mail system. By lowering the barrier to returning the ballot, it could help a small number of citizens who currently find it too expensive or inconvenient to purchase a stamp and can’t drop off their ballot. All indications, however, are that the impact will be quite low. What we do know for certain is that the change would cost the state approximately $1.2 million per election.
I’ve often heard advocates argue that “you cannot put a price on the right to vote,” and in principle this is true, but in practice, everything has a cost, and we have to evaluate those costs relative to other ways we can improve elections.
Let me close by referencing another reform that I am in the midst of examining–ballot tracking. Ballot tracking was recommended by the Presidential Commission on Election Administration as a way to address public concerns about the security of the vote by mail ballots, and has been adopted by a number of jurisdictions that handle large numbers of vote by mail ballots. Ballot tracking systems have been put in place by a number of jurisdictions that use voting by mail. Ballot Trace, for example, used by Denver County, has won national awards.
But the question needs to be asked: does ballot tracking really improve public faith and confidence in the system? Does anyone really track their ballots? Indications thus far are that only a tiny percentage of voters actually use the ballot tracking systems. Is this a solution in search of a problem? Before we spend millions of dollars on free postage paid envelopes or ballot tracking systems, we should have a good answer to these questions.
As the South Carolina results roll in, many of the commentators on CNN seem flummoxed by the fact that Donald Trump criticized the war in Iraq and still won in South Carolina, “a state with a high number of conservatives and veterans.”
But do conservatives and veterans think the war in Iraq was a mistake or not? CNN seems to think they do.
The reality is that public views of the war are very mixed–Democrats overwhelmingly think the war was a mistake, Independents (who are able to vote in the SC primary), and Republicans continue to support the war.
What about veterans? The assumption on CNN seems to be that veterans would support the war, but in fact they do not. Veterans, whether measured as “I served in the military” or “I am a member of a military family” are almost evenly divided on the wisdom of invading Iraq.
There are many reasons that Trump won SC, but his criticism of the Iraq War should not surprise anyone. Among Independents, Democrats, and even members of the military, the war remains very unpopular.
Here’s a nice followup to the article I just posted. It describes a 10 year followup to a Democracy Fellows program at Wake Forest University.
Our analysis of both the quantitative and qualitative data revealed that there continue to be significant differences between the Democracy Fellows and the class cohort. Although both groups dislike the degree of political polarization they encounter in their daily lives, the fellows continue to be more engaged in the political process than does the class cohort.
David Becker, Director of the Election Initiatives at the Pew Charitable Trusts, issued a clarion call in the Stanford Social Innovation Review for a “new approach to reversing the downward spiral of low turnout.” The article is part of a series on “The Role of Philanthropy and Nonprofits in Increasing US Voter Turnout” sponsored by the William and Flora Hewlett Foundation.
There is much to be commended in Becker’s article. He calls for a comprehensive survey to understand non-voters, in contrast to the typical academic surveys that focus on understanding voters. The survey will feed into field experimental studies that will identify methods and techniques to motivate non-voters to vote, and perhaps more importantly, move tangential (presidential only) voters into habitual voters.
Now comes the academic rain on Becker’s parade.
First, there is little evidence of a “downward spiral” in voter turnout, if by downward spiral, Becker means a self-reinforcing, vicious circle of a low turnout election, followed by political dissatisfaction (perhaps stimulated by a polarized legislature), followed by even lower turnout.
Kelly Born’s article that launched off the series, drawn from the US Elections Project, shows that voter turnout in Presidential and Midterm elections has been mostly unchanged for nearly a century. There have been bumps up and down, but focusing (as Becker does) on low turnout in 2014 because it coincidentally is the lowest turnout in a federal general election since 1942 is a convenient choice of endpoints while ignoring 70 years of data in between.
What is quite apparent is that turnout in the United States was high in the 19th century, declined during the Progressive era, in part as a consequence of reforms intended to weaken the role of parties in structuring our political system, and in part because two large waves of newly eligible voters entered the system, immigrants and women.
If we focus in more closely on the post-war era, there is an apparent decline in participation after 1968–volumes have been written about the impact of the 1960s on American politics–but also a substantial increase in turnout from 1996-2008.
In short, there is little evidence of a downward spiral in turnout. Instead, as Adam Berinsky points out,
“…the only way to both increase turnout and eliminate socioeconomic biases in the voting population is to increase the engagement of the broader public with the political world. Political information and interest, not the high tangible costs of the act of voting, are the real barriers to a truly democratic voting public.”
This doesn’t mean that detailed surveys focusing on non-voting will not be valuable–they surely will be. And a toolbox of approaches for non-partisan voter mobilization groups would be an invaluable contribution to the field.
But we need to take seriously the political and structural barriers to substantially increasing participation in the United States. For instance, the real gains will be made among younger voters (18-29)(call for proposals here), less well-educated voters, and Hispanics (not African Americans, unless participation among this group declines substantially with Obama no longer on the ticket). (See demographic comparisons here.)
And we need to turn our attention to other elections. In primary elections, for example, turnout levels are abysmal and primaries arguably have a much larger impact on political polarization. State and local elections rank even lower (25% turnout is high for a municipal election). If we are really going to engage citizens with their political system, perhaps engaging them with the neighborhoods, towns, municipalities, and states would yield much higher gains.
Hat tip to Doug Chapin who reported that the North Carolina State Board of Elections has ordered the Watauga County Board of Elections to establish an early voting location on the Appalachian State University campus. (Full disclosure: I have been serving as the inaugural Daniel German Endowed Visiting Professor of Political Science at ASU from 2014-2016.)
Along with Doug, I followed this controversy during the 2014 general election, when I was on campus. I took photographs at the time of the Watauga County Administration Building, the single location proposed by the County to conduct early voting, and the Plemmons Student Union, the proposed alternative location. Unfortunately, these photos were lost along with my camera on a recent trip, so I will have to describe them from memory.
The problems with the County building are many–parking is quite limited, the room designated for early voting is not very large, and the alcove that would shelter waiting voters from the weather (no small consideration in the mountains) is quite small. Contrast this with the Student Union: there is a four story parking deck just across a sidewalk, the room used for early voting is a large ballroom in a much larger building that has good ADA access, restroom facilities, etc.
Most importantly, however, the Union sits adjacent to a large traffic circle that serves as a primary hub for the “AppalCart”, the only public transportation system available in the county (the system is a partnership between ASU and the County).
I did find some problems with the Union. I counted nine entrances to the building, and was able to find one tucked away in a corner that lacked the requisite “no canvassing” sign. In addition, because the room for early voting is inside a larger building, it is not clear whether the “no canvassing” boundary starts from the outside of the building (encompassing the whole building) or from the door that actually enters the ballroom. Do students (or others) wearing politically themed t-shirts inside the building, for example, violate the “no canvassing” rule?
The political patterns in the county are very clear, however. The central part of Watauga (City of Boone) is younger, more likely to be unaffiliated, and vote Democratic at a much higher rate. The outer portions of the county are older and more Republican. Some quick graphics based on 2014 registration statistics are shown here. Boone City is in the center of the map. The link to the state statute that gives the SBOE this power is here.
Most observers are aware that early voting is an important part of the elections process, but much less attention has been given to how early voting may alter candidate competition and voter decision making during the presidential primaries.
As Redlawsk, Tolbert, and Donovan point out in Why Iowa, a key feature of the current presidential primary process is that it is a sequential election. They write (p. 144):
…early events “matter” in part because news about outcomes in early states serves as a major source of information about candidate viability in a relatively low-information choice setting
When you add early voting to the mix, things get a lot more interesting, because some voters in later primary contests may not wait to cast their ballots until election day. They may cast the ballots early, based on a different set of signals.
Consider this: if Bernie Sanders wins the New Hampshire primary on February 9th, as some models predict, no-excuse absentee ballots are already in the hands of voters in Georgia, North Carolina, Tennessee, and a number of other large states.
And early in-person voting will have started in a number of states prior to the South Carolina primary–currently identified as Hillary Clinton’s “firewall”.
Early voting has become an important feature of presidential elections. While research has generally focused on whether programs increase turnout, few have considered whether early voting alters the information environment in campaigns. Those who vote early may do so before important information becomes available in the final weeks of a campaign. I speculate that early voting should benefit early front-runners in presidential nomination contests, as voters may cast early votes for these candidates before fully considering their less-known opponents. Examining exit-poll data from the 2008 Democratic primaries between Hillary Clinton and Barack Obama, I find that Clinton indeed benefited from early voting in several early primary states.
Marc Meredith and Neil Maholtra, in their 2011 Election Law Journal article “Convenience Voting can Affect Election Outcomes,” take a different look at this question, taking advantage of the “natural experiment” that is ongoing in California–some precincts (< 250 registered voters) are forced into fully vote by mail elections while slightly larger neighboring precincts use the normal mix of no-excuse and election day voting.
The especially interesting thing about the 2008 California primary is that John Edwards and Rudy Giuliani withdrew from the race only five days before the primary, and Fred Thompson withdrew 15 days before the primary. These three candidates were clearly “not viable” for those who cast ballots after they withdrew. Not surprisingly, Edwards, Giuliani, and Thompson received a lot more support in the vote-by-mail precincts.
Overall, the authors conclude:
… the use of VBM affects the relative performance of candidates remaining in the race and increases the probability of selecting withdrawn candidates. Our findings have implications both for election administration policy and for the study of campaign effects in American elec- tions. Election officials should consider waiting until closer to Election Day to send out mail ballots, or instruct voters to wait until they are ready to make a decision before voting.
Will these same dynamics hold in 2016? I have some strong suspicions that they will, but it will not necessarily benefit the front runner, as Fullmer found in 2008. The volatility of the GOP field makes it much less predictable. Candidate drop outs later in the season, and just before particular primaries, should alter the vote totals of the remaining candidates, but in ways that are not really predictable ahead of time.
All this makes for an interesting intellectual puzzle, but perhaps not more than that. The percentages of voters who cast ballots more than a week before the scheduled primary is not usually that large. I’ll be posting more information on that matter in a few days.