Women’s suffrage ushered in a sea change in American politics that affected policies aside from taxes and the size of government. For example, states that granted suffrage were much more likely to pass Prohibition, for the temperance movement was largely dominated by middle-class women.16 Although the “gender gap” is commonly thought to have arisen only in the 1960s, female voting dramatically changed American politics from the very beginning.
Suppressing Voter Turnout: The Poll Tax, Secret Ballots, and Literacy Tests
Most Americans today would agree that secret ballots were a great boon to democracy, while the poll tax and literacy tests—which were often designed to suppress the African American vote in certain states—were terrible injustices. These suppositions are doubtlessly true, but studies show a surprising effect of these measures on voter turnout: while the poll tax significantly reduced voting participation rates, secret ballots also lowered turnout. In contrast, literacy tests actually had little effect on voting rates during most of the years in which they were used.
The poll tax—a fee that must be paid before a citizen can either register or vote—was originally meant to ensure that those voting on government expenditures would also contribute something—even if just a symbolic amount—to the state coffers.17 Yet, the poll tax became the clearest-cut example of a policy that discourages voting.18 Sixteen states implemented a poll tax since 1870, although by 1963 only five still retained such fees, which were finally banned from federal elections in 1964 and from state elections in 1965.19 While the poll tax was only used in a few northern and western states, including Massachusetts, Nevada, and Pennsylvania, it was implemented in all eleven southern states, primarily as a means to discourage African Americans from voting. Indeed, many southern whites viewed the practice as “the main solution to the suffrage problem of disenfranchising African Americans without disenfranchising too many whites.”20 Although poll taxes were relatively small (usually just a dollar or two) and changed little over time, they effectively reduced voter participation rates by around 10 percentage points.21
Poll taxes were in effect for so long that voting rates were still depressed long after their elimination. When the taxes were abolished, voter participation rates rose a mere 4 percentage points in the following election. It took over two decades before voting rates returned to their pre-poll tax levels. For Alabama, Arkansas, Mississippi, Texas, and Virginia, turnout rates were still reduced by around a percentage point even into the early 1980s.22 Here we see the same delay in fully exercising voting rights that was evident after women gained the vote. Among African Americans, many would-be voters had lost interest in political issues while the poll tax prevented them from voting. After the tax was rescinded, it took African Americans a generation to recapture their previous level of political involvement.
One would not normally associate a manifestly discriminatory practice like the poll tax with a progressive reform like the introduction of secret ballots. But there are many aspects of secret ballots that defy conventional wisdom. Secret ballots were not a fixture of early American democracy. They were first introduced in Kentucky in 1882, with South Carolina representing the last state to adopt them in 1950. Secret ballots were implemented to encourage voting by making the system fairer. They certainly succeeded in this among many voters, who could finally vote without fear of offending or angering anyone by their choice.23
But overall, secret ballots had the opposite effect than intended—voting participation fell by an average of 4 to 5 percentage points when secret voting was introduced.24 This can partly be explained by a more nefarious purpose for which secret ballots were used. As historians have noted, southern Democrats used the secret ballot “to depress the turnout of illiterate voters, and thus keep Republicans and Populists from power.”25 This was accomplished by switching the type of ballot simultaneously with the introduction of secret voting. Before the secret ballot, voting was as easy as selecting a colored card that represented a political party. But afterward, party names were written out on the ballots, thus making it more difficult for illiterate citizens to vote.26 However, this only explains a small part of the drop in voting caused by the secret ballot—only a few percentage points of eligible voters were functionally illiterate, and many of them didn’t vote even before the secret ballot’s introduction.27
Much more significantly, secret ballots lowered voting rates in a beneficial way—they helped to reduce vote buying. With secret ballots, the practice of paying people to vote for a certain candidate became less useful, since it was difficult to verify for whom a person voted.28 And when people stopped getting paid for voting, they voted less often.
Literacy tests arrived later than secret ballots. Nineteen states eventually instituted them, with nine of them adopting the requirement after 1900. Before casting a ballot, prospective voters were usually required to read some common document such as the U.S. Constitution.29 Although literacy tests are commonly associated with the South, a majority of states with literacy tests—eleven of them—were in other regions: Arizona, California, Connecticut, Delaware, Maine, Massachusetts, New Hampshire, New York, Oregon, Washington, and Wyoming.
For decades, literacy tests had no discernable effect because illiterate voters and poor African Americans were already discouraged from voting by the poll tax.30 From the 1930s to the early 1950s, however, southern states began phasing out the poll tax. In need of a new method to keep African Americans from the polls, southern state officials began subjectively administering literacy tests to achieve this goal. Literacy tests reduced voting in southern states that had them by an estimated 6 percentage points in 1948.31 These tests helped to keep Southern Democrats in office until the tests were eliminated in all nineteen states by the 1965 Voting Rights Act. Despite the exploitation of literacy tests for these ends in the mid-twentieth century, however, it was the timeworn poll taxes and secret ballots—not literacy tests—that historically had the biggest effect on voting results in the south.
Voter Fraud
Every honest person wants national elections to be fair and accurate. In the United States, we have adopted all sorts of regulations to encourage voting and to ensure the integrity of the electoral process, including ID requirements, absentee ballots, pre-election voting, voting by mail, and registration procedures and deadlines. Some of these measures are controversial, with critics contending that they actually discourage voter turnout or increase voter fraud. These critics have a point: from a basic economics perspective, if we make voting more “costly” by wasting voters’ time in fulfilling onerous regulations, we can expect that fewer people will vote. However, a comprehensive study of voting regulations leads to a different conclusion; while a few regulations do indeed raise the likelihood of fraud, most of them increase turnout by instilling confidence in the voting system.
Poll taxes indisputably lowered voter turnout—in many places, that was their main purpose. But the effects of regulations meant to decrease fraud are less clear. These are geared toward a legitimate purpose, for electoral fraud remains a serious concern in the United States. Cities such as Philadelphia, for example, have become so notorious for voting fraud that the old mantra “vote early and often” seems to have become the city motto.32 In one particularly egregious case in that city, Democratic poll watcher Fani Papanikolau—who was actually a New Jersey resident—voted over and over again by assuming the identity of dead people. Papanikolau was ultimately indicted on over one hundred counts of election fraud and forgery.33 Similar cases have been found in Atlanta,34 while voter registration rolls in St. Louis are so out of date that people’s names were used to vote up to ten years after their death.35
While Florida is best known for allegations of voter disenfranchisement in the 2000 presidential election, the state has long suffered from serious electoral fraud. In the 1997 Miami mayoral election, thousands of fraudulent absentee ballots changed the election’s outcome.36 The following year, out of 8.1 million registered voters, Florid
a officials found that 17,702 were deceased, 47,000 were registered to vote at multiple locations, and 50,483 were convicted felons with no right to vote.37 These problems are also found in other states, some of which have jurisdictions that show more registered voters than there are people. 38 The Wall Street Journal’s John Fund, the author of a book on electoral fraud, has even speculated that the close 2006 Senate elections in both Montana and Virginia—and thus control of the Senate—were likely determined by voter fraud.39
This kind of fraud is enabled by certain voting regulations meant to increase turnout. For example, absentee ballots make voting more convenient for those outside of their voting district, but they are also “notorious” sources of voter fraud.40 The Election Assistance Commission reported in 2006 that “there is virtually universal agreement that absentee ballot fraud is the biggest problem.”41 Consider one outrageous example: in 1998, former Pennsylvania Democratic congressman Austin Murphy was convicted of electoral fraud after forging the names of nursing home residents on absentee ballots. Nursing home administrators were even paid a bounty for access to their patients.42
In addition to enabling forgery, absentee voting makes it easier to buy votes because absentee ballots, like ballots before the advent of secret voting, can be shown to others before being submitted. With 25 percent of voters using absentee ballots in the 2006 general election, these drawbacks are hard to ignore.43
Similarly, while allowing prospective voters to register by mail makes it easier to vote, it also increases fraud. In October 2006 in Missouri, the left-wing Association of Community Organizations for Reform Now (ACORN) turned in at least 1,500 apparently fraudulent voter registration cards in St. Louis and another 3,000 more in Kansas City. Other problematic registrations appeared throughout the state, including forged signatures and the registration of ineligible teenagers and deceased individuals.44
As a result of these problems with voter fraud, bipartisan support has developed for stricter voter registration rules and for voter identification requirements.45 Nevertheless, these requirements are opposed by some Democratic politicians who are concerned that anti-fraud rules will discourage legitimate voters from casting a ballot. Senator John Kerry (D-Mass) even denounced photo identification requirements as “a new Jim Crow era poll tax.”46 Republicans, for their part, tend to support stricter anti-fraud rules, claiming they will increase voting rates by instilling confidence in the voting system. The clash between the two sides has resulted in vigorous court battles in such states as Arizona, Georgia, Indiana, and Missouri.47
How can we ascertain whether stricter anti-fraud regulations increase or decrease voting turnout? One place to start is to study the impact such rules have had in other countries. A good example is Mexico, which implemented some unusually stringent anti-fraud regulations in 1991. To vote there, a person must present a voter ID card that includes a photograph as well as a thumbprint. The cards themselves are virtually counterfeit-proof, using holographic images, imbedded security codes, and a magnetic strip containing still more security information. As an extra precaution, voters’ fingers are dipped in indelible ink to prevent people from voting multiple times.
Furthermore, Mexican voters cannot register by mail; they must go in person to a registration office to apply for a voter ID card, then return three months later to get it. Absentee ballots were banned due to their misuse during the 1988 presidential election. Although they were reintroduced for the 2006 election, their use is closely regulated, with voters required to request a ballot at least six months prior to the election.48
How have these measures affected voting rates? Voter turnout averaged 68 percent in the three presidential elections held since the reforms were adopted, compared to a 59 percent average rate in the three elections before the reforms.49 Clearly, more citizens were encouraged to vote by the prospect of clean elections.50
We cannot expect to find this large of a change when American voting regulations are strengthened because there is nowhere near the same level of corruption as in Mexico. Additionally, in the United States, photo IDs have only been required in a few state general elections beginning in 2006, making it too early to evaluate their effectiveness. However, we do have data on the effect of other regulations. These include anti-fraud measures such as non-photo ID requirements and registration deadlines, as well as rules meant to increase voter turnout such as allowing provisional ballots, “no excuse” absentee ballots, registration by mail, and pre-election day voting.51 Surprisingly, none of these regulations appear to have any effect on voter participation rates.52
Yet, these “non-results” may be misleading. Consider just the anti-fraud measures. On the one hand, these regulations may reduce the total number of votes cast either by usefully eliminating fraudulent votes or by detrimentally discouraging voters by complicating the voting process. On the other hand, anti-fraud regulations may increase voting rates by raising confidence in the voting system. A non-result may simply be the result of all these effects occurring at the same time.
How can we disentangle the different possibilities? One solution is to study the effect of the two kinds of regulations—those that make voting more “costly” by making regulations more strict, and “easy” regulations that make voting less costly by simplifying voting procedures—on voting rates in counties with low rates of fraud. We can then compare these results with the regulations’ effect on voting rates in high-fraud counties known as “hot spots.”
The American Center for Voting Rights provides the only comprehensive national list of voter fraud hot spots.53 Its 2005 report listed six major hot spots: Cuyahoga County, Ohio; St. Clair County, Illinois; St. Louis County, Missouri; Philadelphia, Pennsylvania; King County, Washington; and Milwaukee County, Wisconsin. Analyzing the effect of voting regulations in these counties, we find that “costly” regulations increased voting rates, while “easy” regulations reduced them. The changes were small, never exceeding a few percentage points, but neither type of regulation had any discernable effect outside the fraud hot spots. This result strongly indicates that “costly” regulations encourage voting by instilling confidence in the voting system, while “easy” regulations lower turnout by increasing the perception of a high likelihood of fraud.54
There is only one regulation that impacts turnout outside of fraud hot spots: pre-election day voting. Allowing voting before election day results in a 1.5 to 5 percentage point drop in voter participation. This would surprise some analysts who would expect pre-election day voting to increase voting rates either by making voting easier or by allowing the casting of more fraudulent votes.55 Yet, the result is quite consistent with the general connection between voting turnout and fraud outlined above: by loosening voting regulations, pre-election day voting increases fraud and thereby discourages voters from participating in an election.
Voting Machines
1 million black votes didn’t count in the 2000 presidential election. It’s not too hard to get your vote lost—if some politicians want it to be lost.
—San Francisco Chronicle, June 20, 2004
Ballot fraud by individuals is clearly a problem in the American electoral system. But many people believe there is a far more extensive crisis in American voting. There is a corrosive perception that the voting system selectively prevents certain people, especially African Americans, from voting through the widespread manipulation of voting machines. Democratic operatives repeatedly asserted that punch card machines disenfranchised large numbers of Florida voters in the 2000 presidential election. Since then, litigants with complaints about punch card voting machines tried unsuccessfully to derail California’s 2003 special gubernatorial election, which was eventually won by Arnold Schwarzenegger. Other states, such as Ohio, faced similar litigation from the American Civil Liberties Union during the 2004 election.56
Polls show that a high percentage of Americans believe that systematic disenfranchisement is occurring. A 2006 poll found that only 30 percent of African Am
ericans and 45 percent of Democrats feel confident that their votes will be counted. Among all voters, the number is still low at 60 percent.57 The question of disenfranchisement, it seems, has become an increasingly pressing political issue.
Concerns over punch card ballots have led to billions of dollars being spent to develop new voting methods, reversing a decades-long increase in the prevalence of punch cards. The number of voters using punch cards plummeted from 34 million in 2000 to just over 4 million in 2006. Optical scan ballots, used by 69.5 million people, are now the most frequent method for voting. Electronic machines are a close second, with 66.6 million users.58 All these efforts to eliminate punch cards, however, were counter-productive because punch cards perform better in many ways than the machines that are replacing them.
Some legitimate problems with voting machines do exist. Perhaps the most famous of these is the occurrence of “non-voted” ballots—punch cards on which a vote is recorded either for more than one candidate in a single race or for none at all. During the 2000 presidential election in Florida, representatives from both parties—as well as their lawyers—spent hours discussing ballots that lacked a recorded vote for the presidential race. Some argued that non-voted ballots stemmed from problems in using punch card machines, while others maintained that they largely reflected a voter’s choice simply not to vote for the top race on the ballot.
Nationally, the non-voted ballot rate for typical presidential elections is about 3 percent for punch cards, 2.9 percent for electronic voting machines (DREs), 2.1 percent for optical scans, and 1.9 percent for paper ballots and lever machines.59 In a study of the 2004 presidential elections in Ohio, I arrived at a similar finding. Votomatic punch cards, used in 69 of Ohio’s 88 counties, averaged a 2.4 percent rate of non-voted ballots, compared to 2 percent for optical scans, 1.5 percent for levers, and 1.1 percent for electronic machines.60
Freedomnomics: Why the Free Market Works and Other Half-Baked Theories Don't Page 17