On the So-Called Voter “Enthusiasm Gap”
For as long as I can remember, every election season there is talk of the so-called “enthusiasm gap”, which always favors Republicans. The concept (I would call it more of a theory) arises largely from the fact that polls of registered voters prior to elections overestimate the extent to which voters are likely to vote Democratic and underestimate the extent to which they are likely to vote Republican. Presumably this is due to a relative lack of enthusiasm on the part of Democratic voters – so the theory goes.
As a result of this evidence of polling inaccuracy, the concept of “likely voters” vs. all registered voters was developed. The polling of only “likely voters” (as measured by various voter information collected during the polling process) is supposed to be a better predictor of election results – and indeed it has been shown to correlate better with official election results than polling of all registered voters (in fact, correlation with election results is how the statistical modeling used to determine who is a “likely” voter was developed).
There are many different ways in which voter “enthusiasm gap” is purportedly demonstrated. One is simply to ask voters how enthusiastic they are about voting. Another is to compare pre-election polls with exit polls. If one political party shows a smaller percentage of voters showing up for an election as measured in exit polls, compared to what pre-election polls predicted, that is taken as evidence of a voter “enthusiasm gap” for that Party. The characteristics of voters who were exit-polled are analyzed in connection with voters who were pre-election polled, to determine what characteristics predict “likely voters”. Another method is to compare the pre-election polls to the official vote count. As with the comparison with exit polls, if a political Party fares worse in official vote count compared to what pre-election polls of all registered voters predicted, that is taken as evidence of an enthusiasm gap – presumably meaning that relatively many voters from that political Party did not show up to vote.
Nate Silver notes in a recent column of his that in each of the six U.S. Presidential elections since 1988, “likely voters” voted more heavily Republican than all registered voters. The difference was 0.7% in 1988, 1.1% in 1992, 2.5% in 1996, 1.6% in 2000, 1.2% in 2004, and 2.1% in 2008. Also of note is the fact that in 5 of 6 of those elections the “likely voter” polls ended up being closer to the official results than the total registered voter polls.
In another column about the “enthusiasm gap”, Silver points out that what the “enthusiasm gap” really measures is voter “turnout”. I put turnout in quotes for the same reason that I put “enthusiasm gap” in quotes. Official voter “turnout” really is not quite a measure of turnout – that is, who showed up to vote. Voter “turnout” is defined as the percent of registered voters who actually register an official vote. But the fact of the matter is that every Election Day in the United States hundreds of thousands or millions of voters who show up to vote do not actually register a vote.
THE MANY REASONS FOR LOW VOTER “TOURNOUT” HAVING NOTHING TO DO WITH LACK OF ENTHUSIASM
It is very instructive to consider the many reasons why so many registered voters either don’t show up at the polls or do show up at the polls to vote but don’t register an official vote – because the answer to why this happens illustrates numerous problems with our election system. All of these barriers to voting work against the Democratic Party:
Old broken down voting machines
The problem with old voting machines failing to register a vote was perhaps most notoriously illustrated by the Presidential election in Florida in 2000. The initial vote count on Election Day in Florida gave George W. Bush a 1,784 vote lead over Al Gore. That immediately precipitated a machine recount, which reduced Bush’s lead to 327. Why? It was mostly because hanging pieces of cardboard remaining on the punch cards at the spot where the voter intended to vote had blocked the voting machine from registering a vote for President. When the ballots were run through the machine a second time, many of those hanging pieces of cardboard became dislodged and thereby allowed the vote to be registered. This process favored Al Gore because the old faulty voting machines tended to be concentrated in low income precincts which contained high concentrations of Democratic voters. Later, visual examination of every paper ballot in the state by staff of the Miami Herald several months following the election demonstrated thousands of additional undervotes – that is, ballots that the voting machine failed to recognize as a vote – that provided evidence of voter attempt to vote for a Presidential candidate (for example, tears or indentations in the cardboard). The counting of these ballots as votes would have given Al Gore an additional 1,314 net votes, which would have been enough to win Florida and the Presidential election. But for various reasons, including the stopping of the vote count by the U.S. Supreme Court, most of those votes were never officially counted, and Bush therefore “won” the election.
A large research study on this issue examined almost 2 million unrecorded votes for President in the 2000 Presidential election. Although the national rate of unrecorded votes was 1.8%, the rate was almost twice that, 3.4%, in counties with median annual incomes of less than $25,000. As with our educational system, our roads, and so many other things, less money is typically allocated to poor and minority areas for the purpose of voting than to affluent white areas. This means that poor and minority voters typically vote with older machines, which are more likely to malfunction, especially with regard to their ability to punch clean holes in the designated place on ballots.
Insufficient allocation of voting machines
The 2004 Presidential election in Ohio, whose electoral votes were awarded to George W. Bush, is full of examples of how voters go to the polls to vote and yet fail to register a vote for President.
Franklin County, Ohio, had disproportionate numbers of poor and minority voters. Largely because of insufficient numbers of voting machines allocated to many precincts in Franklin County, 74% of voters waited to vote for more than twenty minutes, and sometimes much longer than that. Furthermore, a study that looked at voting machine allocation per voter by precinct partisanship showed that machine allocation was far less adequate in precincts that voted for John Kerry than those that voted for George Bush. In fact, it appears that there were about 30 Kerry precincts where there was less than one machine per 440 registered voters, while there were no Bush precincts in that category. That same study showed that “voter turnout” (those who actually voted) was substantially less in Franklin County precincts that were allocated fewer machines per voter. This was caused by the fact that long voter waiting times resulted in many voters having to leave the polls before voting, because of work or family responsibilities.
An extensive statistical analysis of the situation by Elizabeth Liddle found that approximately 18,500 voters were disenfranchised in Franklin County because of overcrowding and consequent long waits at the polls. Because Franklin County voted mostly for Kerry, and because the Kerry precincts were affected disproportionately, she estimated that this problem cost the Kerry/Edwards ticket approximately 7,000 net votes.
Furthermore, as Bob Fitrakis revealed, all this happened while 68 voting machines were available in Franklin County but held back. The pattern of voting machine allocation appeared to be quite suspicious:
Purging of voter registration rolls
In Cuyahoga County, Ohio, in 2004, documentation by The New York Times and by the Greater Cleveland Voter Registration Coalition indicated large increases in voter registration, which were not reflected in the official voter rolls maintained by Ohio’s Secretary of State. The New York Times figures suggested a discrepancy of more than 110 thousand voters, suggesting massive voter purging. Confirmation of the probable reasons for the discrepancies came from research by Victoria Lovegren, who posted a report at Ohio Vigilance which indicated the purging, apparently illegal, of 165,224 voters from Cuyahoga County alone, for no other rationale than that they hadn't voted recently.
The likelihood of massive voter purging in Ohio in 2004 was further confirmed by research by Mark Crispin Miller in “Fooled Again – How the Right Stole the 2004 Election and Why they’ll Steal the Next One Too (Unless we Stop them)”. In that book, Miller recounts his conversations with Denise Shull, a poll checker in Summit County, Ohio. During the course of her work on Election Day, Shull noted that approximately 10% to 20% of registered Democratic voters on her list were not on the official list of registered voters. Furthermore, these voters were described as ardent Democrats, as long time voters in the area, AND most of them were not voting. A possible reason for their not voting is suggested by an encounter that Shull had with one of these voters as the voter (or more precisely, non-voter) was leaving the polls. This voter was simply told that she couldn’t vote and was given a phone number to call. And even more disturbing, Shull noted three of her fellow Democratic volunteers who described to her very much the same phenomenon occurring at the polling places where they worked that day.
Disappearing votes
One of the reasons for great Democratic optimism in Ohio on Election Day 2004 was many anecdotal reports of large voter turnout (manifested by very long voting lines) in heavily Democratic Cleveland. To verify those anecdotal reports, I looked at data from the national Electronic Incident Reporting System (EIRS), which received tens of thousands of Election Day reports of voting complaints. This database contained 1,509 complaint reports involving long voting lines. Of these, more than a third, 548, came from Ohio. Of those, more than a quarter – 150 cases – came from Cuyahoga County, and of those Cuyahoga County reports that mentioned the name of the city, 46 of 75 reports were from Cleveland. Therefore, Cleveland accounted for about 6% of complaint reports of long voting lines in the whole United States, whereas they accounted for only a little more than a tenth of a percent of votes in the 2004 election.
Yet despite the very long voting lines reported all over Cleveland, official voter turnout was not recorded as high. In fact, it was quite low compared to elsewhere in Ohio. According to the Cuyahoga County Board of Elections website, the voter turnout in Cleveland was only 53.14%, compared to 73.41% elsewhere in Cuyahoga County, and about 70% in the rest of Ohio. This finding had been earlier reported by Richard Hayes Phillips, a statistical expert in identifying statistical anomalies, whose findings have been widely publicized.
Unlike the case of Franklin County, in Cleveland there was no evidence of insufficient voting machine allocation being responsible for the long voting lines. Every indication pointed to very high voter turnout in Cleveland. Thus it appeared that whereas official voter turnout in Cleveland was very low, actual voter turnout was very high.
No other explanation comes to mind to explain this very strange anomaly other than the possibility that tens of thousands of Cleveland voters who voted did not have their votes registered in the final count.
Work interfering with the ability to vote
The working poor, composed disproportionately of minorities, are much more likely to utilize early voting dates, when available, than more affluent voters. The reason for this is easy to understand: Voting often is an economic hardship for working people, especially those who need more than one job to keep economically afloat, since it is often difficult for them to get the necessary time off of work. Consequently, in the absence of opportunities for early voting, the working poor are more likely to be unable to vote than more affluent voters, due to work or family obligations. This has nothing to do with lack of enthusiasm for voting. It is simply a result of the fact that some people have rigid schedules that they must keep in order to maintain access to the necessities of life.
This year the elimination of early voting, targeted to Democratic areas, has been particularly troublesome in Ohio, which required a court ruling to somewhat restore early voting rights.
Voter ID requirements
The first state voter ID laws were passed in 2003. As of September, 2011, 30 U.S. states required either photo ID or some other form of ID in order to vote. The problem with restrictive voting laws is that they tend to disenfranchise voters, especially the poor and minorities (which is precisely the point of many of their architects and supporters). Many poor people do not have access to photo IDs. They may not drive a car, and they may have no other reason to have a photo ID.
The state of Pennsylvania passed an especially egregious voter ID law in 2012. According the Pennsylvania Transportation Department, 9.2 percent of registered voters in Pennsylvania lack the required ID to vote in 2012. Unless this law is successfully challenged (it is currently being taken to court), these voters will be turned away from the polls when they attempt to vote on Election Day 2012. A large preponderance of these voters tend to vote Democratic.
Challenging voters at the polls
Current state laws in 39 states allow private citizens to challenge voters at the polls. Brentin Mock explains that:
There are a number of right wing groups in this country that are organizing efforts to use these voter challenge laws to suppress voting activity. Mock describes one of them called “Code Red USA”, which
THE SIGNIFICANCE OF THE SO-CALLED ENTHUSIASM GAP
Thus it is that there are numerous factors that have nothing to do with voter enthusiasm that pose a barrier to voting. All of these factors pose stiffer barriers to Democratic than Republican voters. Some of them prevent voters from coming to the polls; some prevent voters from voting once they arrive at the polls; some prevent a voter’s vote from being counted in the official vote count; and others cause a combination of these things.
Depending upon how a particular pollster measures voter “enthusiasm”, any of these factors has the capability of depressing “enthusiasm” by virtue of the fact that they all depress voter “turnout” as officially measured. When combined together, as they always are, the depression in voter “turnout” can be quite substantial, especially when right wing organizations organize campaigns with that goal in mind.
To the extent that these factors depress voter turnout, attributing it all to a lack of enthusiasm can be very misleading. Worse, it can and probably does have anti-democratic consequences. By attributing low “turnout” of Democratic voters to lack of enthusiasm, the blame is put on the voters rather than the systematic flaws in our election process. Even most Democratic voters believe that the traditionally low “turnout” of Democratic voters is due to lack of enthusiasm. I have never seen a study that attempted to account for systematic barriers to voting in the evaluation of the lack of voter enthusiasm on turnout. If such a study was ever done, I suspect that the enthusiasm of Democratic voters would compare quite favorably with that of Republican voters. But by attributing it all lack of voter enthusiasm the focus is removed from the systematic problems that are in dire need of reform.
= new reply since forum marked as read
