These findings, however, do not tell the entire story. The debate has overwhelmingly focused on the presidential race, which is understandable in light of the worldwide publicity generated by the 2000 Florida vote. But if we look at races lower down the ballot than the presidential race, such as those for Congress and state legislatures, we find that Votomatic punch card machines actually perform much better than electronic or lever machines and are about as reliable as optical scans.
People naturally cast fewer votes for races lower down the ballot because they don’t know or care as much about those positions as they do about the presidency. But the drop-off rate varies systematically according to the type of voting machine. And, most interestingly, this drop-off is much less evident for punch cards than for other types of machines. The rate of non-voted ballots for Ohio Senate races on Votomatic punch card machines was just under 10 percent, compared to 18 percent for electronic and lever machines. Looking across all of Ohio’s races, switching from Votomatic punch cards to electronic or lever machines—as so many precincts have done nationwide—would have resulted in about 200 more non-voted ballots in the average Ohio ward of 1,696 voters.61
Why do punch cards do so well at the bottom of the ballot? The answer lies in what I call “voter fatigue.” People will vote in fewer races when it takes them more time or effort to vote. Recent research indicates that electronic machines encounter problems with “the willingness of voters to navigate through multiple ballot screens before casting a vote (and) delays caused by the use of the review feature when coupled with extended ballots.”62 These extra screens and reviews mean it takes up to 20 percent longer to use electronic voting machines than punch cards. Consequently, voters on electronic machines get tired faster, and some don’t make it to the end of the ballot. The ability of elderly voters to master the technology of electronic voting has also been questioned, as “older adults consistently perform more poorly than younger adults in performing computer-based tasks.”63 Whatever their other faults, punch cards are relatively quick and simple to use.
Contrary to popular belief, Votomatic punch card machines were also the only voting method for which African Americans had consistently lower rates of non-voted ballots than whites had.64
With all the debate over voting machines and non-voted ballots, one would think that the machines must be too complicated for many people to figure out. But education is not systematically related to the rate of non-voted ballots. The pattern appears completely random, with the rate high for those with less than a ninth grade education, low for those with some high school, high for high school graduates, low for college graduates, and high again for those with post-graduate degrees. Nor does income level consistently correspond to the rate of non-voted ballots although, quite interestingly, the richest voters are fifteen times more likely to cast a non-voted ballot than the poorest ones.65
Ironically, the switch to electronic machines not only failed to rectify the problems they were supposed to solve, but they have sparked new concerns over potential vote fraud. These range from the lack of a “paper trail” on most machines to the possibility of manipulating their programming.66 Similar to the concerns over punch cards, these new objections are largely misguided—paper records are not necessarily superior to electronic voting machines, which keep three separate “read only” memories that are unchangeable. Likewise, tampering would be near-impossible, since electronic voting machines are stand-alone units that are not connected to the Internet or any other network. 67 Tampering with them would be akin to someone trying to hack into your own personal computer while it isn’t online.
Most electronic voting machines transfer election results to a compact disk or some other “read only” format. These CDs are then taken to a central location where they are read into a computer. In the twenty-plus years that these machines have been used in counties across the nation, there has never been a verified case of tampering.68
When computer scientists warn of possible tampering with voting machines, they are not referring to hacking attempts, but to someone physically breaking open the lock on each individual machine and reprogramming each computer. Even if someone could break into a machine and overcome the supposedly tamper-proof seals without being noticed, going through one computer at a time hardly seems like a realistic way to steal an election. Besides, accuracy checks conducted on the machines before, during, and after the votes would detect any reprogramming to transfer votes from one candidate to another.
Contrary to the various conspiracy theories, many people who registered non-votes for some races were most likely conflicted over whom to vote for and simply decided not to support anyone. Undoubtedly, the promotion of these unfounded accusations of selective disenfranchisement offer a short-term political benefit for certain groups. But such claims risk poisoning the political debate for years to come.
The 2000 Florida Vote
Florida’s 2000 presidential election was a battleground for not just punch card voting, but for explosive charges of systematic discrimination against African American voters. Reverend Jesse Jackson was not alone in charging “a clear pattern of suppressing the votes of African Americans.”69 Mary Francis Berry, then chairwoman of the U.S. Civil Rights Commission, called for a criminal investigation of the vote. But how does one determine empirically whether there was systematic discrimination against African Americans during the Florida election and whether such discrimination cost Al Gore the election?
Data are available for every Florida precinct on vote totals for each candidate, the number of non-voted ballots, and on voters’ race and party affiliation. At first glance, the numbers seem to confirm the disturbing claims that African American had higher rates of non-voted ballots than other groups. But critics are wrong in charging that this aberration cost Gore the election, for the group most “victimized” by non-voted ballots, in fact, was African American Republicans.
African American Republicans who voted in Florida were an incredible fifty-four to sixty-six times more likely than the average African American to have had a ballot declared invalid. Another way of phrasing this is that for every two additional African American Republicans in the average precinct, there was one additional ballot that did not register a vote for president.70
In 2000, there were 22,270 registered African American Republicans voters in Florida—or about one for every twenty registered African American Democrats. In a state where the 2000 presidential race was decided by fewer than 1,000 votes, this is no trivial number. Since we can assume that these voters, as Republicans, would vote mostly for the Republican candidate, the overall numbers show that George W. Bush was hurt more by the loss of African American votes than was Al Gore.
These results are indeed disturbing. They indicate that, if there was a conspiracy to disenfranchise some group of Florida voters, that effort was directed at Republicans, not at African Americans as a race. This conclusion conforms with another fact that the data reveal: among white voters, Republicans were much more likely than Democrats to cast non-voted ballots. Additionally, I found that the overall rate of non-voted ballots was 14 percent higher when the county election supervisor was a Democrat, and 31 percent higher when the supervisor was an African American Democrat.71
So where does all this leave us? It’s hard to believe that there was some nefarious plot to disenfranchise Republicans overall or African American Republicans in particular. More likely, a significant proportion of these Republicans, particular those living in heavily Democratic areas, simply did not like the choices for president and therefore declined to vote for that race. Although tampering cannot be definitely ruled out, we can dismiss claims that the poor and African Americans overall were disenfranchised, since figures show that income and race were only one-third as important in explaining spoiled ballots as were the methods and machines used in voting.72
Early Media Calls
What happens when exit polls leak out on election day? Would you still vote if you kn
ew the election was already decided and your candidate had lost?
As noted earlier, people are usually most interested in voting for the top races on the ballot—for president, governor, or U.S. Senate. Close contests in these top races typically increase voter turnout, while an expected blowout keeps more voters at home. It is widely assumed that an expected landslide primarily discourages those who support the losing candidate, since no one wants to be on the losing side. This is borne out by post-election surveys in which greater proportions of respondents claim to have voted for the winning candidate than is indicated by the actual vote count.73
The media has made three early calls in presidential elections: 1980, 1996, and 2000. During the landslide Republican victory in 1980, NBC named Reagan the winner at 8:15 p.m., well before voting ended on the West Coast. President Jimmy Carter “dispirited” many fellow Democrats when he conceded the election at 9:45 p.m., an hour and fifteen minutes before voting had ended in .California.74 Many Democratic candidates on the West Coast blamed their own losses on Carter’s early concession, which allegedly discouraged Democrats from going to the polls.75
Carter’s defenders countered that Reagan’s landslide ensured that these Democratic candidates would have lost anyway. His supporters also cast doubt on surveys showing that Democratic voters were discouraged by the early concession, arguing that fellow Democrats wanted to blame the party’s losses on a tactical mistake by Carter.
When Bob Dole challenged Bill Clinton for the presidency in 1996, it was the Republicans’ turn to suffer from an early election call. Citing exit polls, the networks called the election for Clinton before the polls on the West Coast had closed. Republicans blamed the media for discouraging their West Coast supporters from voting, while Democrats insisted that Republican candidates would have lost anyway.
What impact do early media calls actually have on voting? It is hard to estimate their impact in the 1980 and 1996 elections because they affected all the races in the Western states—it is difficult to determine whether it was the early election call or something unique to those states that caused the drop off in voting. Fortunately, the 2000 election was different: polling had ended in most of Florida, but was still ongoing in the state’s ten western Panhandle counties, when the media declared that the Democrats had won the presidential race as well as the Florida Senate races.76
Florida’s polls were open until 8:00 p.m. on election day. But Florida’s heavily Republican Panhandle counties are on central, not eastern, time, so they stayed open for an additional hour. Yet, beginning at 8:00 p.m. EST, all the major networks (ABC, CBS, CNN, FOX, MSNBC, and NBC) incorrectly and repeatedly announced that the polls were closed in the entire state. CBS national news alone made eighteen statements indicating that the polls had closed.
Even for the western Panhandle voters who knew that their polls were actually still open, they still had to consider the media claims that the races’ outcomes had already been determined. The Florida Senate race was called for the Democrats fifty-eight minutes before the polls closed in the western Panhandle, while the presidential race was called for Al Gore twelve minutes before the polls closed. Even before the presidential race was called, the media offered numerous reports indicating that Gore appeared likely to win.77
After the election, surveys showed that the early media call and the perception of Democratic victories had discouraged voters in the western Panhandle. Two-thirds of voters in that region voted Republican, so even if the early call discouraged voters from both parties at the same rate, more Republicans ended up not voting. According to Democratic strategist Bob Beckel, George W. Bush suffered a net loss of up to 8,000 votes due to the early media call. Another survey of western Panhandle voters by John McLaughlin & Associates, a Republican polling company, estimated 10,000 lost votes for Bush. But both surveys suffer from the same problem afflicting the polls done after Carter’s defeat: respondents may have let their political interests color their answers. In this case, disgruntled Republicans may have been more likely to exaggerate the effect of the early call.78
That voting dropped, at least to some degree, after the early call is confirmed in the sworn statements from many election clerks and poll workers. According to a Clerk for Elections in Okaloosa County, for example, “We had over 1300 people turn out with an average of about one hundred voters per hour until the last hour. When the doors were open, there were quite a number of people waiting in line to vote. There was a heavy flow throughout the day....Soon after 7:00 p.m., I noticed that the volume dropped to almost zero.”79
Statistical evidence supports this, too. During the 2000 election, the western Panhandle counties suffered a drop in turnout not only relative to their past turnout rates, but also compared to the rest of Florida. Comparing how each county’s turnout changed during the day, the western Panhandle counties also suffered a drop relative to the rest of the states.80 The evidence indicates that the early call cost Bush a net loss of at least 7,500 votes.81
The early call in Florida probably also affected other races across the country because it made it appear as if Gore had won the election before the polls closed in many other states. But it is much more difficult to estimate the effect outside of Florida, since only that state has data to allow for a comparison between polling stations that remained open after the early call and polling stations that were closed. However, a simple look indicates that if the results for Florida are correct, the early call most likely affected some other races. For example, Democrat Maria Cantwell won a Senate seat in Washington state by just 2,200 votes. If the early call’s effect in Washington was 1/24th of the smallest estimated effect of the early call in Florida, then Cantwell would have lost to Republican Slade Gordon. This outcome alone may have significantly changed U.S. political history, as Gordon’s victory would have meant that the Republicans would not have lost control of the Senate when Vermont’s Jim Jeffords left the Republican Party in 2001.
The presidential race was ultimately decided by Bush’s victory in Florida by just 537 votes. The turmoil that engulfed the nation after the vote would probably have been avoided had there not been the early media call, as Republicans would likely have won Florida by a larger, indisputable margin.
Incentives matter in voting, just as they do in determining so many other kinds of behavior. And early media calls can reduce the incentive to vote. In the 2000 election, this caused Americans a lot of sleepless nights.
Felon Voting
Michael Milken, Martha Stewart, and Leona Helmsley share something in common besides being convicted felons—they are all Democrats.82 While their wealth sets them apart from the typical felon, their party registration is the same as most former convicts. Ever since the razor-thin outcome of the 2000 presidential election, some Democrats have led an effort to restore voting rights to former prisoners. If felon voting had been allowed in 2000, it could easily have tipped the election to Gore who, as previously mentioned, lost Florida by a mere 537 votes. Since that election, twenty states have made it easier for felons to vote,83 leaving only ten states with lifetime voting bans for felons.84 Senators Hillary Rodham Clinton and John Kerry introduced the Count Every Vote Act in 2005, which Clinton claimed was “critical to restoring America’s faith in our voting system.”85 The bill, which never became law, included a measure to restore voting rights to “felons who have repaid their debt to society” by completing their prison terms, parole, or probation.
As discussed in Chapter Two, many convicted criminals face severe penalties in addition to a prison sentence. Many jobs are forbidden to felons, often making it hard for them simply to earn a living. Yet, since the 2000 election, the loss of voting rights has suddenly emerged as the most pressing problem that former convicts supposedly face. Restoring voting rights, we are told, is indicative “in so many ways of citizenship that it is more important than owning a gun or being able to hold [a particular job].”86
Felons themselves, however, have other priorities. In addit
ion to finding a job, felons, who frequently live in poor, high-crime neighborhoods, want to be able to defend themselves. In Virginia, the number one reason felons cite for asking for clemency is the desire to regain their right to own a gun.87 The Assistant for Clemency for the Governor of Virginia for 1994 and 1995 reported that restoring “voting rights was never on the application for clemency.”88
According to academic studies, from 1972 to 1996, on average 80 percent of felons would have voted Democratic. An overwhelming 93 percent ostensibly would have voted for Bill Clinton in 1996. In addition to giving the Democrats the White House in 2000, this “felon vote” would have given Democrats control of the Senate from 1986 to 2004.89
But these studies are problematic. Felons’ voting patterns are assumed to be the same as those of non-felons of the same race, gender, age, and educational status. The estimates do not account for the possibility that there is something fundamentally different about felons that could cause them to vote differently. If two people are of the same race, gender, age, and educational status but one person commits murders or rapes, there might be something quite different between these two people that could affect how they vote.
Public Opinion Strategies surveyed 602 adults in Washington State in May 2005. Of the respondents, 102 were felons who had their voting rights restored, while 500 were non-felons. They were asked about their political preferences, as well as background information about their race, gender, education level, religious habits, employment, age, and county of residence. This survey makes it possible to test the assumption that felons and non-felons are essentially the same.90
Freedomnomics: Why the Free Market Works and Other Half-Baked Theories Don't Page 18