Comments on: An afterthought on Bush vs Gore

By: John Quiggin

John Quiggin — Thu, 28 Oct 2004 10:05:54 +0000

“it’s virtually impossible to have the wrong person win.” That’s why effective ties are rare, and why two ties in a row would be most unlikely

By: Dubious

Dubious — Wed, 27 Oct 2004 16:24:41 +0000

I’d be interested to see how you modelled it, JQ. My instinct would go something like this: Normally, if we thought that each individual vote was an iid sampling from a distribution, we would just model this as a binomial problem with a really large n. In this case, with an electorate that is 50.1% Gore, 49.9% Bush, and with 3 million votes, we would have an expected gap between the two of 6000 votes, with a st.dev. of 866. In this case, it’s virtually impossible to have the wrong person win. But the sorts of problems that people worry about aren’t iid sampling errors. Rather, they worry about errors from a biased instrument. Suppose there’s a 50% chance that A) the instrument correctly counts 100% of Gore votes while only counting 99% of Bush guys and 50% chance of B) the other way around. The lost 1% in each case is just thrown away. Under A), a Bush win is virtually impossible. Under B), we now have final votes which are distributed 49.6% Gore and 49.9% Bush. In this case, a Gore win is virtually impossible. Obviously, all this depends on the particular underlying value of the voters and the distribution of the error rates. Given a less evenly divided electorate, you’d need larger, more uneven, or more probable error rates to generate a “50% chance of picking the wrong guy” result.

By: John Quiggin

John Quiggin — Tue, 26 Oct 2004 23:05:49 +0000

As I’ve pointed out quite a few times, the likelihood of two successive elections as close as that in Florida is minimal. I’ll put up a post with the math on this when I get time.

By: Dubious

Dubious — Tue, 26 Oct 2004 21:59:11 +0000

Thanks for the thoughful reply, James. I suppose what I worry about is something like this: Nov 2 Florida goes Y by 49.5 vs. 49. Media reports various overcrowded polling stations, disproportionately in X party districts. Nov 3: Florida court orders new election Dec 3: New election held (must allow time for logistics, including new mailing of absentee ballots) Election is still close, because underlying voters are closely divided. But whichever party lost now points to irregularities which might have it. Dec 4: Florida court decides… Jan 3: New election held… et cetera. My guess is that if the electorate has no confidence in the process, no close election will satisfy them.

By: James J. Kroeger

James J. Kroeger — Tue, 26 Oct 2004 14:16:10 +0000

Nicholas: “…I don’t care whether there is cheating in close elections.” Thank you for clarifying your position.

By: James J. Kroeger

James J. Kroeger — Tue, 26 Oct 2004 13:54:48 +0000

Dubious: Suppose the second vote were also close, as seems fairly likely if the first one was. Would you then go on to do ‘best out of three’ or would you keep on voting till someone got a large enough majority? Again, the idea is not---as cynics like to suggest---to grant the losers of close elections fresh opportunities to try to garner enough support to win the last election; it is to correct [and preclude future] acts of electoral fraud. Expressed suspicions of illegal activity are generally not sufficient to warrant a fresh election. It would be wise to make an exception to this rule, however, if the electorate is deeply polarized and suspicions have inflamed the losing party into a very agitated state. In such a scenario, it would probably be wise to address those suspicions by scheduling a fresh election and taking extraordinary measures to reassure the disenchanted that the entire voting process is being carried out fairly. Second, more generally, it seems like any voting system is an imperfect measurement device for measuring the preferences of everyone who shows up that day. All measurements of reality are imperfect. Absolutely; a vote is an imperfect measurement of something that can change from day to day. But that reality is not a sufficient reason to ignore the importance of reaffirming the integrity of the democratic process in a country. Daily variations in sentiments would have to be considered secondary to the need to assure the electorate that elections in their country can be counted on to be fair. If 1) your party happens to have benefited from a daily change in sentiments among swing voters on the scheduled voting day, 2) some members of your party were subsequently caught cheating, and 3) your party then lost a fresh election by twenty-five votes due to another daily variations in sentiments, then it could be argued that you should have been more vigilant in enforcing discipline on your party members. By making sentiment variations secondary to fairness issues, political parties are given an incentive to not cheat in the first place.

By: Nicholas Weininger

Nicholas Weininger — Tue, 26 Oct 2004 13:39:05 +0000

James: you’re right, I don’t care whether there is cheating in close elections. There always will be anyway; politicians are all power-lusting bastards, after all. It is not possible to be “unfair” to them, since they are irredeemable scum, and it is not being “unfair” to the voters to give the post to the 49.9% guy instead of the 50.1% guy. The sensible objective here is not to satisfy some nonsensical cosmic ideal of “fairness.” It’s to provide a system that doesn’t go too far wrong (say, no more than the 3-4% error margin of a poll) and is efficient and regular in its operation.

By: James J. Kroeger

James J. Kroeger — Tue, 26 Oct 2004 13:03:53 +0000

Nicholas: The calamity of 2000 was not that the election was decided contrary to the Holy Will of the People; Bush and Gore had essentially indistinguishable degrees of approval from said Holy Will. The calamity was that we spent so much time wrangling over a question that might just as fairly, and much more cheaply, have been solved by a coin flip. I’m not sure if you realize what you are actually saying. Implicit within your argument is the assumption that an unfair election result matters little if correcting the unfairness would not significantly affect the total amount of satisfied voters. If I agree with your logic, then why should I not embrace the idea of cheating in order to win an election that is expected to be close, since it really doesn’t matter which party ends up winning it? I would argue quite the opposite: that the Fairness Variable becomes a matter of transcendent importance when elections are close. I would also argue that making it harder to challenge the results of a close election result would actually encourage election participants to cheat more. If on the other hand it appeared certain that any discovered acts of cheating would result in a fresh election, why bother?

By: jet

jet — Tue, 26 Oct 2004 12:42:46 +0000

This thread is ridiculous. Either several of you, or I, have taken a distinct vacation from reality. If you think Gore, suing to only recount in “his” counties and to not recount in Bush’s counties, is fair, then perhaps you are open to debate that Hitler’s election was fair. Bush was fighting to recount all counties or no counties. And if you think Bush was wrong for that, you are an extreme partisan willing to cheat to win. Because as was proved, whichever way a county leaned, that is the way the recount favored. No wonder people on the right have a tough time taking the left seriously as this is the level of debate on an academic site. Crying because Gore didn’t get to recount in every single possibile way until he won is boring and wins no one to your side.

By: Brett Bellmore

Brett Bellmore — Tue, 26 Oct 2004 10:43:47 +0000

James, “The stopping rule: cease challenges when the collective efforts of both parties to produce an “unassailable” election result are widely perceived to be successful. Such a result will never be achieved, of course, if efforts to manipulate the outcome are unending.” That’s not a recipe for legitmate outcomes, that’s a recipe for giving the edge to whoever is more unreasoningly adamant about “assailing” any victory by their opponent. It boils down to, “Keep holding new elections until the side that lost the latest one gives up.”

By: Tom Doyle

Tom Doyle — Tue, 26 Oct 2004 10:14:19 +0000

Administration and Cost of Elections Project Vote Counting-Guiding Principles To establish and maintain public confidence in the electoral process, vote counting systems and procedures should incorporate the fundamental principles of vote counting in a democratic election. These fundamental principles are: transparency; security; professionalism; accuracy; secrecy; timeliness; accountability ; equity Transparency For the counting process to be open and transparent, representatives of political parties and candidates should be allowed to witness and/or participate in the process, and permitted to copy the statement of the results of the counting process. National and international electoral observers should also be allowed to witness the process and permitted to copy the statement of the results of the counting process. In some countries, ordinary citizens are encouraged to watch the counting process. Manual counting is by its nature more transparent than computerized counting. If vote counting is computerized, new mechanisms for ensuring transparency, such as external audits, need to be introduced. Security The security of the ballots and the ballot boxes, from the time voting begins to the completion of the count, is fundamental to the integrity of the counting process. Polling and counting officials, representatives of political parties and candidates, and national and international electoral observers should carefully watch the ballots and the boxes at all times, and accompany them if they are moved from one location to another. [...] Professionalism Polling and counting officials must act in a professional manner. They should be thoroughly trained in the counting process, as distinct from the voting process; thorough in their procedures; and committed to treating electoral materials with care and respect. Once a person accepts work as a counting official, he or she must be non-partisan throughout the entire process. [...] Accuracy Accuracy is directly related to the integrity of the count, and of the elections themselves. Later discovery of errors and correction of mistakes can lead to accusations of manipulation or fraud. The accuracy of the count will depend on clear procedures and manuals, adequate staff training, and their commitment to the process. Clear audit trails of ballots and ballot boxes, as well as checking and rechecking mechanisms, will contribute to the accuracy of the results. Whilst mechanical voting or computerization may enhance accuracy, this must be balanced against the resulting apparent loss of transparency. [...] Accountability Clear responsibility and accountability for each stage of the counting process are important. At the national level, the electoral management body should be accountable. At the electoral district level, it may be a senior election officer or commission official. At polling stations, specified poll workers may be responsible for polling and counting. Clearly defined complaints and appeals processes are also important. Counting rules, including criteria for rejecting ballots, should be clear, known in advance, and understood by everyone involved in the election, including election officials, the general public, political parties, candidates, non-governmental organizations, and national and international electoral observers. Clear audit trails are essential in ensuring accountability. Equity Equity generally means that the rules are the same for all participants in the electoral process, and that they accept these rules. The proper training and non-partisanship of counting officials and polling officials, and the presence of political party representatives, national and international electoral observers will help to ensure that counting is conducted in a fair and correct manner.

By: Dubious

Dubious — Tue, 26 Oct 2004 05:57:14 +0000

First, on the revote question… Suppose the second vote were also close, as seems fairly likely if the first one was. Would you then go on to do ‘best out of three’ or would you keep on voting till someone got a large enough majority? Second, more generally, it seems like any voting system is an imperfect measurement device for measuring the preferences of everyone who shows up that day. All measurements of reality are imperfect. It is incredibly important that we try to make them as perfect (and transparent, that’s important too) as we can. But suppose the true state of affairs in Florida was that those who showed up to vote (or all the registered voters, your choice) really were 50% +/- .05 for Gore, 50% +/- .05 for Bush. Every resampling, either by using different methods of counting the already existing votes or by holding a new vote, is going to have a roughly 50% chance of proclaiming the “wrong” guy as victor. In a close election, there’s just no way around that, as far as I can see. Even moving to 2 out of 3, or 3 out of 5, would not significantly affect the chances of proclaiming the wrong guy the victor. A disturbing prospect, that our vote-measuring is fairly likely to be wrong in a truly close election, but I see no way around it. The best practice seems to be to work at continual improvement of our measuring tools and hoping for better luck next time.

By: Nicholas Weininger

Nicholas Weininger — Tue, 26 Oct 2004 04:26:40 +0000

I hesitate to bring up Steven Landsburg on this blog, but he makes an awfully good point on this subject over at Marg. Rev. Money quote: “if you really believe in democracy, and if the election is close, then it doesn’t much matter who wins.” http://www.marginalrevolution.com/marginalrevolution/2004/10/democracy_theor.html I would add that, if you don’t believe in democracy but rather in the preservation of a stable rule of law, it matters even less. The calamity of 2000 was not that the election was decided contrary to the Holy Will of the People; Bush and Gore had essentially indistinguishable degrees of approval from said Holy Will. The calamity was that we spent so much time wrangling over a question that might just as fairly, and much more cheaply, have been solved by a coinflip. A do-over election would worsen, not solve, this problem. The right answer is to simplify procedures so as to make challenges to close election results much harder, and to disabuse people of the delusion that there is somehow an enormous difference in legitimacy between the guy who won 50.1% of the vote and the guy who won 49.9%.

By: John Quiggin

John Quiggin — Tue, 26 Oct 2004 04:26:37 +0000

Fresh elections are held if and only if the court declares the election void. In practice, this happens when the number of votes subject to irresolvable disputes exceeds the winning margin. This was clearly the case in Florida. The Court has the power to unseat the initiall declared winner and seat another candidate but, as far as I know, this has never happened.

By: Tom Doyle

Tom Doyle — Tue, 26 Oct 2004 03:24:09 +0000

John: You wrote: In a situation where there were enough disputed votes to shift the outcome (and no satisfactory way of determining the status of those votes), the Court of Disputed Returns ( would probably order a fresh election. It seems to me that this is a better way of resolving problematic elections than attempts to determine a winner through court proceedings, though I’d be interested in arguments against this view. In your paragraph above you refer to the fresh election as “a better way than …court proceedings…[for] resolving problematic elections.” But the material on the AEC site seems to indicate that a fresh election would only happen following, and as a result of court proceedings, rather than as an alternative to such proceedings. First, someone-a candidate, the AEC, a voter – would have to file a petition to invalidate the election. “Petitions must set out the facts relied on to invalidate the election and if alleging illegal practices, must show how these could have affected the election result. Only illegal practices within the meaning of that term under the Act can invalidate an election.” [...] “A petition must: “(i) set out the facts relied on to invalidate the election; (ii) set out those facts with sufficient particularity to identify the specific matter or matters on which the petitioner relies as justifying the grant of relief (the Court may at any time after the filing of the petition relieve the petitioner from complying with this); (iii) contain a prayer asking for the relief the petitioner claims to be entitled to” Detailed pleadings are required, it would seem. Regarding the forum: “The High Court of Australia sits as the Court of Disputed Returns.” It has a panoply of inquisitory and remedial powers. “…to compel the attendance of witnesses and the production of documents; “…to grant to any party to a petition leave to inspect in the presence of a prescribed officer the rolls and other documents (except ballot papers) used at or in connection with any election and to take, in the presence of the prescribed officer, extracts from those rolls and documents; “…to examine witnesses on oath; “…to punish any contempt of its authority by fine or imprisonment;……to award costs; “…to dismiss or uphold the petition in whole or part; “…to declare that any person who was returned as elected was not duly elected; “…to declare any candidate duly elected who was not returned as elected; “…to declare any election absolutely void; Are fresh elections held when the court declares elections absolutely void? Under any other circumstances?