How Should I Vote?

Yesterday’s elections in the UK raised again an old question, which receives surprisingly little public discussion. Should I vote on the basis of my own self-interest (or the interest of my family), or should I vote on moral, or ‘other-regarding’, considerations?On some ethical theories, there isn’t really a problem. If we take the ‘should’ here as a moral one, according to ethical egoists I’m morally required anyway to do what is best for myself, so there’s never a conflict between morality and my own good. An Aristotelian virtue ethicist might argue that my own self-interest consists in my being morally virtuous, so again there’s no conflict. And some consequentialists will claim that, though there could in theory be a gap between self-interest and morality here, in practice there isn’t: because each person knows their own interests best, they should vote on the basis of those, and that will – through the invisible hand – produce the morally best outcome.

But many people, including many philosophers, do think there’s a gap. Party A is going to make me worse off financially: it will increase income- and purchase-tax. But it has committed itself to spending a good deal of that money in some of the most deprived communities, as well as making serious improvements to power stations to cut carbon emissions. No other party has such worthwhile aims.

Well, if the ‘should’ is a moral should, the answer is still clear. Morally, I should vote for party A. But of course that doesn’t really answer the question at all, since if we take the ‘should’ to mean ‘should, from the self-interested point of view’, then I should vote for some other party.

One view many people find natural is that the ‘should’ in the original question is, as Bernard Williams put it, ‘unsubscripted’. So it’s neither moral nor self-interested: it’s just ‘should’. Surprisingly few philosophers have developed this natural view, though there are notable exceptions (Bishop Butler, Henry Sidgwick, and more recently Derek Parfit, for example). What one has to do is to weigh moral considerations against self-interested ones (if there is a conflict), and see where the balance comes down. Let’s imagine I’m pretty well off, the communities to be helped by party A are very badly off, and the predicted tax increases aren’t huge. Then I should vote for party A. But if the tax increases are huge, and the plans to benefit the worst off communities seem rather half-baked, then perhaps I have strongest reason to vote for party B.

The main political parties in the UK, and indeed Europe as a whole, for obvious reasons, put far too much weight on how they will advance the self-interest of their voters. But morally speaking they are not equal. Some parties show greater concern than others for the poor, the stateless, non-human animals, future generations, and other morally important groups. And most voters, I suggest, do not have self-interested reasons of sufficient weight not to vote for them. But of course voting is plausibly seen not as purely expressive, but as an attempt to shape the future, and this raises the question whether or not to vote tactically. And here as well moral and self-interested considerations will both have to be placed in the balance.

  • Facebook
  • Twitter
  • Reddit

15 Responses to How Should I Vote?

  • Anders Sandberg says:

    A related issue is whether one should vote if one does not know what party or policy is good. This might be more relevant on smaller scales like when voting in a group, but the problem seems to carry over to society as a whole. If I know I am ignorant, then I ought not to add noise to the system (assuming others know more). Maybe I ought to seek out somebody I do think have properly considered opinions and copy their vote.

    There is also the issue of rational ignorance: spending the time to learn about the issues might be costly, and as long as party A or B are unlikely to significantly change one’s own situation (if one is self-regarding) or the total situation (if one is other-regarding) it might be better to stay on the couch. Rational ignorance makes more sense for a self-regarding person. To an other-regarding person the personal cost doesn’t count as heavily. However, if the likely outcome is that there is still no clear choice, then the other-regarding person could still be justified to not vote (in order to not be noise); she might still have to learn the issues even if she has little hope of becoming decided.

    It seems to me that it is easier to discern benefits to oneself from proposed policies than benefits to others, so self-regarding voters would be less likely to rationally abstain since they would actually have more accurate perceptions of what option is good. Other-regarding voters would both have to try harder to figure out what option is good, and may then have to abstain from voting anyway. It seems that there would be a bias towards self-regarding voters voting among a population of moral voters, especially when the issues are complex.

    • Roger Crisp says:

      Thx Anders. This seems right to me. But of course on my view voters shouldn’t be entirely self-regarding, and because what is at stake is so significant there is a strong case for finding out enough about the options so that one can make an informed choice.

  • Dave Frame says:

    Thanks Roger – that bit about the “unscripted” should is really interesting… though I find the bit about some groups being “morally important” a bit worrying. Fortunately real-world politicians don’t tend to say that some groups are morally important… leaving the obvious implication hanging in the air that some of us are “morally unimportant”. [We know who we are... all we have to do is read the Guardian...]

    In the voting calculation there are considerable dynamic effects that have to be weighed against immediate effects. High taxes and generous welfare may decrease inequalities (let’s imagine this is important) but, as evidence from the post-war period shows, these policies reduce incentives to work and reduce the rate at which average welfare increases. So potentially there’s a trade-off between the interests of the poor today and interests of the poor tomorrow.

    Complete speculation… but in my experience the more economically and mathematically sophisticated people are the higher the weight they give to these sorts of dynamic considerations.

    • Roger Crisp says:

      Thx Dave. Yes, I shld have said ‘especially morally important’! Good point about dynamic effects, but we also need to remember the effects on the environment of continued economic growth now which itself will take its toll on future people (thought of course there is also the non-identity problem to sidestep at some point…!).

  • Joel says:

    “leaving the obvious implication hanging in the air that some of us are “morally unimportant”. [We know who we are... all we have to do is read the Guardian...]”
    I must be stupid as there is nothing ‘obvious’ to me here. Please explain.

    “High taxes and generous welfare may decrease inequalities (let’s imagine this is important) but, as evidence from the post-war period shows, these policies reduce incentives to work and reduce the rate at which average welfare increases”
    Please provide your data.

  • Tim Kitchin says:

    Interesting. I do take should as a moral decision, in that I see government as having the levers of law and thus of evolving justice.

    I tend to think on that basis we should vote ‘situation-blind’ as per Rawls, and that genetic self interest is not a mindful act, and nor is enhancement of immediate quality of life, with all it’s unintended social consequences. That said, voting is often an expressive, rather than an elective act, in most systems – as we’ve just seen – and you might in that case follow a ‘do least social harm’ principle (from an anticipated outcome standpoint), in practice, ‘voting your message’, but bearing in mind you are working within a complex semiotic system.

    • Roger Crisp says:

      Thx Tim. Yes, voting certainly is expressive. But there’s a question about whether it should be, if the expression is seen as an end in itself.

      • Tim Kitchin says:

        Thx Roger. Practically to avoid ‘social’ expression being a relevant mode of decision-making you’d have to design perfect democracy, which I’d loosely equate to 1:1 representation, i.e anarchy. To avoid individual expression, you’d have to rewire the human brain. But to take on your question at face value as theory, I think there is another lens on this which is the question of intent vs outcome, similar to the question of voting values or voting value – both these dilemmas cut across the question of individual vs collective question. Given that there cis relatively less knowledge of outcomes in terms of either collective benefit, or even personal benefit given the vagaries of imperfect democracy, Maybe there’s a philosophical case for voting values/intent. This therefore leaves it a question of free choice. On that basis, if you believe that the greatest good will come from voting self-interest and that is part of your inherited or acquired value-set…then you are justified in doing so (assuming no other moral breaches are implied by those values).,.and thus back to anarchy. NB. This is a rubbish argument, I think, but I wanted to log the point about ‘intent’ and ‘outcomes’ as this has widespread applicability…

        • Roger Crisp says:

          Thx Tim. I wldn’t see direct democracy as equivalent to anarchism (the majority presumably will still rule), and one cld still seek social expression through voting on individual issues. Agree on the intention/outcome distinction, though not on justification (which sounds rather like boot-strapping).

          • Tim Kitchin says:

            Interesting point if I’ve understood you, that boot-strapped democracy – partial representation, partial information and partial choice – can be lived with as a learning system based on improved accountability of outcomes…

            I agree that 1:1 democracy can be used as an instrumental system, with contested dialogue to drive policy formation and institutional experimentation to drive a mosaic of governance (which is what I would actually advocate) cf Zadek – The economics of Utopia.

            I’d still argue that the sheer depth of this imperfection (as well as corruptible power, human nature, and collective mindlessness) legitimises a long-term community-bounded values-led choice for many – which may well co-incide with self-interest. fWIW I have considerable sympathy with Proudhon et al on this topic, and hold a mutualism political line, which may be overwhelming my clear philosophical thinking.

  • Michael says:

    Let’s imagine I’m pretty well off, the communities to be helped by party A are very badly off, and the predicted tax increases aren’t huge. Then I should vote for party A. But if the tax increases are huge, and the plans to benefit the worst off communities seem rather half-baked, then perhaps I have strongest reason to vote for party B.

    The deliberation at stake here seems to proceed by zeroing on the moral goods that the vote aims to secure: when two equally moral goods are aimed for by two distinct parties, you should always vote for the party with the best chances of success. So yes it illustrates the idea that some deliberations with moral stakes are not solely determined by such stakes, but I don’t see how it illustrates the distinction between self-interests and moral interests.
    By the way, I don’t think why we need this distinction (self-interests vs. morality) at all when it comes to electing people (as a Swiss citizen I can tell you there is a pretty substantive difference between elections and votations, i.e. the act of accepting or rejecting amendments to one’s legal corpus).
    Electing people is all about saying who you think would best fit a certain role. Sure, it matters whether the candidates satisfy certain moral standards, and sure, it matter what programme they will be trying to implement. But the moral standards seem to become relevant just in case failing to meet those standards indicates a problem for the fulfilment of the role, where such a problem can be defined non-morally, simply as political values that the institutions finds it desirable to sustain (i.e. integrity, fairness, transparency). So no morality is bedrock here.
    Remains the question of the programme. It seems to me that assessing a political programme threatens to pit self-interests against morality just in case one views the role to be filled as means to improve one’s particular condition over that of others. By contrast, someone who believes that electing is about improving everybody’s condition (no Aristotelian required here!) in accordance to the principles of justice encoded in the institutions will see no gap between self-interests and morality here. (Although (i) one might disagree on what those principles amount to and though (ii) it might be sometimes difficult to weigh the moral desirability of the ends against the chances of success of the proposed means). So, it seems to me, in neither case, candidates or programme, do we need to distinguish between morality and self-interests. So what did I miss?

    • Roger Crisp says:

      Yes, Michael, I agree that my cases weren’t fully described. I was hoping readers might fill in the details! I was imagining a case in which party A would indeed cause a decrease in my well-being, but I’m relying on a view of well-being there which some would reject. On moral standards, I was suggesting that these aren’t internal to the particular political system in question. The problem with politics is that most cases are such that any decision will not improve everybody’s position.

  • Tim Kitchin says:

    Hah, Michael that’s a funny pragmatic argument.
    In so far as self-interest is a proxy for greed, it is immoral, rationally, spiritually and humanisticslly.

    Standing at one remove to select an agent of that greed – a party who (you believe) will disproportionately benefit you (and your direct descendants???) does not make it right. If your intent is that the benefits will even out in the end proportionately to your chosen allocative framework (effort, age, responsibility, achievement, talents…) them you can do that. But if you are disingenuous in your intentions, you are morally guilty and should vote otherwise. To out it more simply, in Christian terms, to offer another approach, no man can serve two masters.

  • Michael says:

    Hi Tim,

    To clarify, y point was not that moral ends cannot conflict with one another when assessing the programme of individuals that propose themselves for filling certain political roles, or cannot conflict with self-interests in such circumstances. My point was rather that if moral ends conflict with self-interests, self-interests should win if and only if one submits to a flawed conception of democracy, i.e. one where individuals should try to secure themselves opportunities and advantages over others. In other words, it is always irrational to privilege self-interests over moral ends if there is a conflict between the two, all other things being equal (including the appropriateness of the means to achieve those ends).

    Once this view is jettisoned what remains are conflicts between distinct pairs, with the weighing issue considered by the author of the original post.

    • Michael says:

      For some reason the following passage was not properly parsed by the site engine:

      Once this view is jettisoned what remains are conflicts between distinct ends-means pairs, with the weighing issue considered by the author of the original post.

Authors

Affiliations