Wednesday, 12 November 2014

Mandatory voting remains a bad idea

Andrew Coyne thinks voting should be mandatory in Canada*; Dylan Matthews makes a similar argument for America.

The best argument for compulsory voting, Coyne's, is that it could shift politicians away from campaigns based on getting out the vote: if everyone has to vote, then GOTV matters less, as does playing to the party's base.

While that's one notch in favour of compulsory voting, there are a few considerations against it that might give us pause.

Blood donation is almost entirely a public good. Were we to implement a mandatory blood donation regime, we would have to couple it with a tick-box allowing "donors" to indicate that their blood probably shouldn't be used or at least should undergo heavier screening: a Hepatitis B patient's blood probably shouldn't enter into general circulation.

But nobody seems to worry much about the quality of a compelled vote. They should.

Jason Brennan makes a pretty convincing argument that while none of us have a positive duty to vote, those choosing to vote have a duty to vote well: to weigh seriously whether the policies offered up by the different parties would achieve the ends that the voter wishes and that the trade-offs are worthwhile, in the voter's estimation. While few voters meet that duty, things are even worse among current non-voters: non-voters, on average, aren't made up of rational calculators and conscientious objectors like me; most non-voters instead have little political knowledge and little intention of acquiring any. And forcing them to vote does not encourage them to acquire more information.

I'll offer instead a compromise position: make voting compulsory, but also implement a simple quiz at the ballot box. Voters would need to be able to match parties with their main supported policies. They'd also be quizzed on a few basics, like which parties formed the prior government, whether crime rates increased or decreased over the prior administration, whether income inequality has been increasing or decreasing, whether most climate scientists agree that the planet has been warming, and order-of-magnitude level quizzes on the composition of the budget. The quiz questions would vary election to election but would hit on the baseline matters necessary for understanding the policy issues at stake in that year's election. Everyone's vote counts for at least 1, but we could award bonus votes for voting well. Everyone would have the opportunity to vote well: Elections NZ would put up the 50-or-so quiz questions and their answers; each voter might get a random draw of 10 at the ballot box. Those compelled to vote but unable to vote well would have less opportunity to do harm to their own interests with their vote. I won't go all the way to nudge-based voting, but we could at least avoid some harm with this version.

Think of the incentive effects. GOTV campaigns would be focused on informing supporters about basic facts so that the party's supporters' votes might count for more.

I could support compulsory voting if it came with this kind of mechanism. Or we could keep things voluntary.

* And see his prior similar argument here.


  1. IMO any argument for voter testing has to address why such a scheme would not be used to disenfranchise certain populations. Think voter literacy tests etc. in the South. Or is this out of the straw man playbook?

  2. There were two very large problems in the Southern US voter tests. The whole point of those were to disenfranchise black voters. And so the quiz questions were culturally loaded, deliberately, and the folks running the election booths would coach whites through the quiz while not helping blacks. Massive differences in educational opportunities also meant that any literacy standard necessarily had racial disenfranchisement effects. And, that system banned those failing from voting at all.

    Set it instead where there's a set of factual questions about the current state of the world and about which parties propose which policies, make the questions and answers well known well in advance, and give a subset of questions on random-draw when the voter enters the booth, and simply provide a small bonus for doing well on the quiz.

  3. I am sure that been made to vote, when I lived in Australia, has some sort of effect on preference formation. Maybe Jon Elster was right.

    I can't bring myself not to vote in New Zealand even though I know my vote doesn't count that much.

    However, in New Zealand, because of the vagaries of MMP, your party vote can be quite decisive in determining the last seat in Parliament. The last seat in Parliament is always determined by a handful of votes.

    A number of parties appear to be just on the cusp of another list MP, and that is before you start doing all the calculations as to who is the last MP elected.

    The two most understudied topics in public choice are the economics of MMP and the incentive to vote, and the Tasmanian legislative Council.

    For the uninitiated, The Tasmanian legislative Council has always been dominated by independents. The Liberal party does not contest seats. The Australian Labor Party is likely to win two seats out of 15.

  4. Eric, the Tasmanian legislative Council is also the ultimate juxtaposition of compulsory voting and rational ignorance in voting.

    It is compulsory to vote in elections were practically every candidate is an independent. In consequence, a son or daughter of the retiring member has the the inside running.

    After that, being a local mayor or sporting hero gives you a good chance of getting in the Tasmanian legislative Council as does being a radio or TV personality.

    The only Legislative Council member I am met was when he was a candidate. He gave himself no chance of being elected, so naturally he won the election relatively easily on preferences.

    Furthermore, under the optional preferential voting system, you must rank at least three candidates: one, two and three. Therefore, you must know three complete strangers, vote for them, or cast an informal vote to avoid the fine

  5. Eric, Australia used a language test to run the White Australia policy.

    The immigration officer could administer a language test at his discretion in any European language.

    They are trying to keep an East European Communist out in the late 30s, who was also a polyglot. He failed the test in Scottish Gaelic.

    He won an appeal because the High Court of Australia ruled that Scottish Gaelic was not a European language.

  6. I'd of course add to your climate question a question on whether any of the currently proposed solutions had any likelihood of materially impacting the climate in many decades, and a requirement to rank the major global problems facing the world in terms of likely deaths. Anyone incorrectly placing global warming as the greatest moral challenge of our time would of course not be allowed to vote.

  7. I have difficulty with compelling folk to vote. I choose to vote, and in fact enjoy taking part in the process of electing our govt, even though I know intellectually that there is an extremely low chance that my vote makes any difference to the final outcome. But compelling people who have no interest in politics to vote will likely result in an increase in spoiled or nonsense votes, and would thereby waste the time of a lot of hard working electoral officials who would have to separate the wheat from the chaff. I also maintain that choosing to not vote is a valid form of protest for some people, and they should be free to exercise that option if they so desire.

  8. I expect that that puts it quickly into the too political and too hard basket. Having good non-partisan sound questions is the hardest part of the whole scheme.

  9. This is a joke, right? I can't see the sense in creating a larger barrier to entry to vote (the quiz) or in posing clearly condescending questions at the voters that you're asking to turn up (the contents of the quiz and the assumption that they need it).

    You've got a civic obligation to do other things as a member of society, like not committing crimes or turning up to jury duty - I don't see anything sacred in the compulsion to vote issue in itself. The big question is whether the voting block in a voluntary system forms a biased sample of the whole; and the answer is generally that it is. So pick your evil, either a biased sample and outcomes that aren't generally what the people would want, or you're asking people to turn up who otherwise wouldn't care and then they are potentially voting ignorantly or randomly. (Or perhaps just as many biases are generated by politicians who pander to ignorant voters under a compulsory system.)

    I'm quite certain that only a politician who never, ever wanted to have any representation amongst the so called 'ignorant' would ever propose the quiz. Maybe you could pitch the idea to CY Leung in Hong Kong.

  10. MMP doesn't make it more likely to change outcomes.

    Maybe your likelihood of pushing the quota just over the line for the last list seat is more likely than making or breaking an electorate tie. Even if you've done that, your vote still doesn't do anything unless it:
    1) Pushes a party above the 5% threshold that otherwise didn't have an electorate; or,
    2) Causes the making or breaking of a tie between alternative governing coalitions; or,
    3) Substantially affects the balance of power within a governing coalition.

    Option 1's odds are next to nil.
    Option 2 can be possible; depends on the election. High chances that your MP is inframarginal even if you've broken a tie.
    Option 3 can also be possible, but hard to say what the effect sizes would be.

  11. 1. There is no civic obligation to vote, or at least none under any moral theory to which I subscribe.
    2. I agree with Jason Brennan that those who have no information have a duty to not vote.
    3. There is pretty ample evidence that voters generally know very little, and decent evidence that non-voters generally know even less. Hit Ilya Somin's work, Bryan Caplan's work, or my work on New Zealand data. The problem isn't just random voting. Were it that, it wouldn't really much matter. Instead, ignorance is biased.
    4. I agree that the policy is unlikely to be adopted by anyone. But I do expect that compulsory voting could worsen outcomes without it; that makes it an argument against moving from the status quo to compulsory voting.

  12. There has to be absolutely no way that any quiz system could be used in this way - the whole thing is a non-starter if this is a potential outcome. I can't see it as being a potential outcome in New Zealand; if I did, I'd never have mentioned it.

  13. Eric, you should look at the full MMP calculations. They are available on the election commission website if you click a link at the bottom.

    The last seat is always very close. As I recall, Kiwiblog said that the last seat in Parliament in 2008 was decided by 40 votes.

    Several small parties are always surprisingly few votes away from another list MP.

  14. found it at

    Spot 120. Labour are in 121.

    National’s quotient is 9160.0 while Labour’s quotient is 9159.5.

    If Labour had 40 more voters turn up (that is less than one voter per seat), then Damien O’Connor would be back in Parliament and Aaron Gilmore would have missed out.

    Or if 22 people who voted National had voted Labour, then Labour would have 44 seats and National 57.

  15. And so instead of a 63 seat National-Act coalition, with 6 more in support from Maori and UF, for 69/122 victory for the National coalition, there'd have been a 62 seat National-Act coalition, with 6 more in support from Maori and UF, for a 68/122 victory for the National coalition.

    And still no single vote made a difference: you'd have needed 40 non-voters to turn out *and vote in Labour's direction*.

    That's a great argument for Labour, and others, to run strong GOTV campaigns; it's not an argument for rational people to vote.

  16. Have not seen these calculations done for any election other than 2008.

    Normally, the argument is your vote has one chance in 1 million of making a difference to who gets elected.

    Under MMP in the 2008 election, the chance of being a pivotal voter was one chance in 22.

  17. That isn't right. If 21 net other people didn't flip with you from National to Labour, then you had zero chance of changing the outcome.

    You could run some simulations on the ex ante likelihood that, in a tight enough race, you might make or break a tie. Results will depend on the parameters you put in for each voter's underlying likelihood of choosing parties other than the one that they did choose.

  18. True, but it still seems that the last seat is close. That is different from usual your vote is one of countless votes.

  19. Close calls aren't the same as pivotal votes though. Note Mulligan's work in the US suggested 1/89,000 votes cast in Congressional elections turned out to be pivotal.

    Plenty more were close calls. Should we buy lotto tickets where close calls that don't pay out give false hope?

  20. True, but I could have been somebody, I could have been a contender, is better than a statistical oblivion

  21. Embrace the Total Perspective Vortex, voting edition.