The law professor Scott Shapiro has a fun series of tweets where he pretends to be carefully making a decision about how to vote in the 2020 presidential election. A recent tweet reads:
The joke is that the arguments against Trump are so strong (and Shapiro’s leanings so clear) that such careful evaluation is useless.
But there is a lurking question that I don’t think political science has addressed as well as it should. That is, how should citizens vote? Not how do they vote - we have lots of analysis of this - but what rule they should use in voting. We could call this the normative analysis of voting (as opposed to empirical analysis). In Trump’s case, not much analysis is needed, though I’ll show at the end that some commonly used voting rules lead to surprising conclusions. In any case, if voters determine a lot of what happens in democracies, then it is pretty important that we have a good ethical theory of how people should cast their ballots.
It’s worth taking a quick stop, however, to think about what voting should achieve. We might posit two main goals. The democratic point of view is that voting should lead to a government that is close to the people, a government that does what people want. The problem is that people sometimes want bad things, things contrary to their own interests because they don’t understand the effects of policies.
A second goal then is that voting should lead to a government that acts in the best interests of its citizens and pursues the common good. Yes, this would require some definition of best interests, but we could posit such widely held values like maintaining sustainable and equitable economic growth.
These two goals are sometimes referred to as responsiveness and (substantive) representation, but it’s easier to think of them as doing what voters want and doing what voters need. The voting rules I describe below sometimes favor one of these effects over the other.
So, how exactly should citizens vote to bring about either one of these goals - getting what they want or getting what they need?
Voting in Space
High school civics classes give us one model of how good citizens should approach elections. The standard civics class advice tells people to study the platforms of the parties and choose the one that is closest to them. Various election calculators help in this process - they match your answers to a series of policy questions with each party/candidate’s answers and find the one that is closest. Political scientists call this spatial voting (because you are voting for the party closest in policy space) or prospective voting (because you are looking at how parties will behave in the future). I am leaving out nuances in how exactly to do this - the nuances lead to a variety of forms of spatial voting like directional voting.
This approach comes fairly close to our basic idea of democracy, what Achen and Bartels call the folk theory of democracy. Spatial voting seems best able to produce a government that will do what people want and so is democratic. But the spatial conception falls short in a number of respects. In the first place, it is hard - most voters aren’t very familiar with party positions or platforms and may even have trouble placing them on a left-right scale. Further, people often prefer policies that are not very good for them or for society as a whole - eg, raising taxes and lowering spending at the same time. Thomas Frank’s What’s the Matter with Kansas gets at the way that poor people may vote against their economic self-interest (though his empirics may not be on the mark). And it is unclear whether parties or candidates can fulfill their programs whether for reasons of institutional gridlock, their privileging of their own self-interest, or in Trump’s case basic competence.
Voting to Punish
An alternative conception focuses on assessing the performance of incumbents. This is the famous “Are you better off today than four years ago?” that political scientists call retrospective voting. Here voters reward parties or candidates that have governed well or produced good outcomes and punish those who have not. The sanctions here are powerful in that they can both remove bad politicians and encourage politicians to act well today lest they be judged negatively in the future. Again, there are nuances in how to carry this out, for example, voters might keep a running tally of the government’s performance in their head.
But as with prospective voting, there are difficulties for voters. Are they capable of distinguishing between outcomes due to the behavior of the politician in question or to forces beyond their control? Was the pandemic or the global recession the government’s fault? Did their response help things or make them worse? Were governing politicians actually capable of doing something given an opposition trying to obstruct them or coalition partners altering their plans? Even more seriously, can voters assess whether outcomes are good or bad. Democrats and Republicans can look at one and the same economy and come to very different conclusions about whether things are going well. And how high should one set standards for being “better off”? Set the standards too high and no politician can achieve them and they will just give up; set them too low and politicians won’t try hard enough.
Voting for Character
For all these reasons one might give up on policy and outcomes and focus instead on the politicians themselves. Commonly referred to as selection of types, voters might try to choose the best person whether this means honest, principled, competent, or experienced. Here the view is that governing is more technocratic and less ideological and zero-sum. Voters seem to like this conception - James Fearon points out how voters dislike office-seeking, support term limits, and prefer consistent and principled politicians - but are they very good at it? Can they make good character judgments? Trump is another case in point here. And does character truly matter?
Descriptive representation might be seen as a subtype of this sort of selection. Here the idea is that members of one’s group are better able to understand one’s interests or more motivated to push for them. Thus, women might vote for women and those of a particular ethnicity for a member of their ethnic group in order to assure themselves better representation. This is certainly an easier rule to follow (unless you are faced with two female candidates), but would it necessarily lead to the best representatives? There are worries that by limiting choices, we will end up with less skilled candidates, but there are counters to that as well. Anzia and Berry find that female candidates are more skilled than men due to what they call the Jackie/Jill Robinson effect - women face so many barriers that only the very best women succeed.
Voting Strategically
Most of these forms of voting would be leavened with a strategic imperative: don’t vote for candidates who don’t have a chance to win representation. This applies most dramatically to a first-past-the-post system like the US. The idea is that if you have a preference among the top two candidates, then you will hurt the one you prefer by voting for a third party. In the famous formulation, you will be throwing your vote away. This strategic imperative applies mainly to places where there are two main parties (strategy is more complicated in proportional representation systems). But if you care about the outcome of the election, you should make sure to vote for a candidate or party that actually has a chance.
I’ll stop here so as not to become tedious, but one could come up with other rules, not to mention variations on these principles. In his Ethics of Voting, Brennan controversially argues that citizens should only vote if they are epistemically justified that the policy or candidate they choose would promote the common good. Epistemic justification and the common good are obviously doing a lot of work here. (NB: Brennan’s focus is on whether one should vote at all, not how to choose whom to vote for.)
Trump or Not?
So what should we tell citizens about how to vote? Consider the current presidential election. There will certainly be citizens who feel closer to Trump in policy space whether because of judges, walls, taxes, or guns. In the civics class view, they should vote for him.
Should we then point them to his record over the past four years? Until the pandemic, his economic record, typically the part political scientists (and voters) focus on, was not bad. How much credit or blame does he get for that? The same goes for the pandemic. It is still unclear how much responsibility Trump bears, especially given performance in places like Sweden, not to mention federal constraints on his power.
Probably many supporters of the Democrats will ask voters to consider Trump’s character and probably Trump is exceptional here. Would those same supporters ask voters to judge Romney and McCain in the same way? Both were principled and upstanding and proved as much during Trump’s term. Does this mean that we should switch among these standards depending on the circumstances of the time - something like, if the candidate is a truly terrible person, then vote against them, otherwise consider policy and performance. That doesn’t seem right, but it might be true.
As you can tell, I’m not certain of the correct answer. I am surprised though that political science hasn’t provided a better answer. Or maybe I’m missing something.
Of course, there is one thing we can say pretty confidently: If you care about this election, don’t vote for a third party!
PS: Things can get complicated pretty quickly with some of these rules. Consider the fact that politicians respond strategically to any rule that voters select. A recent paper by Ashworth, Bueno de Mesquita, and Friedenberg shows, for example, that increases in voter information or rationality can sometimes make things worse for them because of the way that politicians respond.