Posted: October 30, 2007
AVBM: What If We Could Make It Fraud-Proof?
Oregonians are proud of their all-vote-by-mail elections, and election administrators in other states are investigating the benefits of adopting such a system (see this post). The benefits include avoiding the bottlenecks of having a single Election Day, avoiding duplication of costs involved in administering a hybrid in-person/absentee voting system, and the convenience of mail voting. One downside, critics claim, is the risk of fraud, though Oregonians are quick to point out that no serious fraud has ever been shown to have occurred under their system, which has been in place since 1998. Some elections scholars have also theorized that fraud is more difficult to accomplish under AVBM than under traditional absentee voting because, among other reasons, it is impossible to double-vote by casting both an in-person and absentee ballot (see Edward Foley, The Where and When of Voting, 6 Election Law Journal 270, a review of Absentee and Early Voting by John Fortier). I do not know what risk of fraud is present in AVBM systems, but I do think that smart technology and procedures can probably reduce whatever risk of fraud is present and make AVBM a very attractive option.
Cook County, Illinois, presents one example of how smart procedures can help. Cook County uses a traditional absentee voting system that requires voters to provide an “excuse” to obtain an absentee ballot. To reduce the risk of fraud, administrators look at the number of incoming absentee ballot applications from each precinct to determine whether an unusual number of such ballots have been requested. Where the number of requests is high in a precinct, administrators place a call to each and every voter in that precinct and ask the voters whether they have requested ballots, whether anyone has attempted to improperly influence their votes, whether anyone has offered to return the absentee ballots for them, and other questions. Using this system, Cook County succeeded in preventing a number of significant absentee voting schemes, including one involving about 250 votes in 2003.
The problem with this system is that it cannot be applied to AVBM. The Cook system relies upon “spikes” in ballot applications to target investigations, but these spikes do not occur under an AVBM system where everyone receives absentee ballot and ballot applications are unnecessary. Instead, administrators in an AVBM system would have to find some other way to target investigations or, alternatively, perform some kind of untargeted follow-up. An example of an untargeted follow-up is provided by Chicago, which is within Cook County, but has a separate election authority. In Chicago, administrators place phone calls randomly to 5% of individuals who have requested absentee ballots and ask them the same types of questions asked in Cook. While this system has the benefit of potentially catching fraud in precincts where ballot applications have not spiked, it also has a considerable downside: It says only that 5% of voters had a clean experience, and says nothing about the other 95% of them. Under such a system, the chances of an administrator uncovering an absentee fraud scheme might be significantly reduced.
However uncomfortable one feels about such a system, one thing is for certain: The degree of discomfort is inversely proportional to the percentage of absentee voters who receive a phone call. Five percent might not be enough, but what about 50%, 75%, or 100%? As the percentage rises, it begins to look less and less reasonable and more and more contrary to claim that we cannot have faith in our elections results—although even at 100% perhaps it is still more reasonable than contrary.
I do not think that phone banking alone can bring the risk of fraud down to reasonable levels in every jurisdiction (some jurisdictions are more prone to fraud than others), but it can be one piece of a rational system that accomplishes this goal. Another important part of such a system would be educating voters about common absentee fraud techniques and encouraging them to report suspicious activity. To be effective, the encouragement might have to go beyond mere words and include financial incentives, formal public recognition, or other incentives to blow the whistle. It is also probably desirable for fraud-prone jurisdictions to stiffen penalties and put as much bipartisanship and transparency in their administration as possible, although this may be less necessary for jurisdictions without a demonstrated history of fraud. Finally, there are probably ways that technology could be used to further reduce the risk of fraud, although I am not sufficiently apprised of technological matters to say for sure.
Those who are still concerned about fraud under such a system should consider the true cost of their concern. The cost of AVBM may be that occasionally 250 fraudulent ballots are cast, but I’m willing to pay that cost if—if-- the alternative is an unwieldy Frankenstein-type monster that strains resources, drives good administrators into early retirement, and disenfranchises thousands of voters who do not have the luxury of waiting in line for four hours to cast a vote that might not be counted.
All of this is not to say that AVBM is clearly the answer. AVBM can lead to lost votes because it does not provide the type of voter error correction provided by DRE and in-precinct optical scan systems, and it also may skew participation by disproportionately increasing turnout among those groups already most likely to turn out. But it is possible that the fraud concern could be reduced or eliminated by appropriate use of procedures, and AVBM should not reflexively be rejected on the basis of fraud.