This is part one in an ongoing series on voting in America. This article focuses on First past the Post voting; later articles will focus on gerrymandering, disenfranchisement, the electoral college, and voter participation. Stay tuned.
In some countries, most notably the United States, Canada, the UK, elected representatives are chosen in a winner take all voting process called “First past the Post” (FPTP). When ballots are counted for a given race, the candidate at the end of the night with the most ballots in their box wins. This is a simple enough idea, but in recent years many constituencies have been pushing for a reformation. Most notably, the current Labour party of Canada ran on a platform of voting reform which swept them to victory in 2015. What are the problems with FPTP, and is there really a more fair way of holding elections?
Elections in the US are weird. Presidential elections are not won based on number of votes, but rather on winning individual states’ electoral votes. This system should not be confused with First past the Post; indeed, the Electoral College tends to favor candidates who receive fewer votes but who are broadly popular in less populous states. Elections are run by the individual states. While the winner of the presidential race might not be the candidate with the most overall votes country wide, he has to get the most votes in enough states to capture the majority of electoral votes.
To see FPTP in action, let’s examine the famous results of the presidential race in Florida in the 2000 election cycle including the top four candidates.
Under the First past the Post system, George Bush, the winner of the most votes was declared winner of the State of Florida.
Critics of FPTP say that this result does not correctly represent the will of the people. One reason is vote splitting. In the example above, “conservative” candidates (Bush and Buchanan) represent 49.14% of the total votes, while “liberal” candidates (Gore and Nader) represent 50.48%. This implies that the state of Florida would prefer a liberal candidate overall, but because the liberal vote was split more than the conservative vote, a conservative won.
This leads to the second problem many people see with FPTP: it encourages tactical voting. It is likely that many voters would actually prefer Nader or Buchanan to Bush or Gore, but because those candidates were expected to receive a low vote share, many voters decided to vote for one of the two candidates who were likely to win.
There are certainly other critiques of First past the Post, but for now, I am only going to discuss these two issues. The question for today is, are vote splitting and tactical voting easily solvable problems like some anti-FPTP activists claim, or would any solution create a bigger problem?
Vote splitting comes in two forms. First, as in the Florida example, there are so called “spoiler candidates,” people who eek out just enough votes to deny a victory to an ideologically similar (enough) candidate. Second, there are cases where two or more viable parties compete for the same pool of voters allowing a third party, while less popular than the first two combined, to win because their vote is not split. An imperfect example can be seen in the 2006 Canadian federal elections where the conservative party, led by Stephen Harper, won control of parliament by winning roughly 40% of seat races compared with 60% split between three center-left or left wing parties. Based on these results, one would expect Canada to have a left leaning government, but instead a conservative government was formed due to vote splitting.
To maintain clarity and continuity, I won’t be discussing election results on a country wide scale due to differing systems of governance (parliamentary vs federalist, etc), but rather on a race-specific basis. So, to illustrate vote splitting, imagine the Canadian result is played out in a single race. Candidate A receives 40%; B, 30%; C and D 15% each. B, C, and D are from different parties, but all are much more ideologically similar than any of them are to A. The spoiler situation will be as follows: A receives 49%; B, 48%; and C 3%. Again, B and C are of different parties but much more ideologically similar than either are to A.
The first way to solve this problem that has been proposed is a mandate that some candidate get 50% of the vote to win. One way to do this is call a runoff election between the top two candidates if nobody gets above 50% the first time. This would solve the spoiler candidate issue; the spoiler (Candidate C) would be eliminated and a new election would happen between A and B. Presumably B would win C’s voters, or at least they would stay home, and win a majority of votes.
This would also deal with the Contrived Canadian example I laid out above. C and D would be eliminated and their voters would presumably go to B. In these contrived cases, the result we would expect occurs. However, this system also delivers some strange results under different circumstances. If candidate B was, although ideologically similar to C and D, considered by everyone outside his 30% base of support to be eccentric and unfit for office, the results of the runoff might be that A would get 70% and B would get 30%. It might be that C or D would have won a runoff against A because they are better compromise candidates than B. In other words, if B’s voters would be fine with either C or D, but C and D’s voters would not be fine with B, maybe C or D should be chosen for the runoff instead of B.
Maybe a system of ranked choice could solve this problem. In the situation above, I can imagine ranked voting played out like this: A is ranked first by 40%, third by 30% and fourth by 30%. B is ranked first by 30% and fourth by 70%. C is ranked first by 16%, second by 60%, third by 25%. D is ranked first by 14%, second by 40%, third by 45%. In this case, it looks like the runoff should be between A and C, the consensus choice. Or maybe even between C and D, the only two candidates who over 50% consider to be either first or second. If we assign points to ranking, (let’s say three points for a first place choice, two for second, and one for third) and 100 voters, the point values look like this for the race:
Indeed, on that scale, the runoff should be between C and D. That is a weird result though, because it totally excludes A, the candidate who differs most. Intuitively, it seems like the runoff should offer a more significant choice. And also intuitively, it seems like the runoff should not be between the candidates who were fewest people’s first choice.
Another way to do ranked voting is to eliminate the last place finisher and distribute their second place votes to the other candidates. How would this play out? D Gets eliminated and his 14% of voters get their votes distributes to their second choice. Everybody’s second choice was either C or D, so it follows that all D’s voters preferred C as their second choice. Now the numbers are as follows: A: 40%, B and C 30% each. Presumably, in a large election B and C would not actually tie, so one probably has a slight edge. If B is eliminated next, his votes go to C and C wins. If C is eliminated, her votes go to A and A wins. So, the widely unpopular B who has a strong base of support (a populist candidate) can’t win. This seems right. But, the winner will either be the one candidate most different from the majority opinion (A) or someone who was only 15% of people’s first choice. I will discuss these results after dealing with tactical voting.
Tactical voting is the strategy of voting for someone other than your ideal candidate in order to avoid a candidate you really dislike from winning. Let’s say there are three candidates: A, a left wing populist; B, a middle of the road establishment type; and C, a fiscal conservative. They are polling at 40%, 40%, and 20% respectively. If your main priority in voting is lowering the deficit, you should be in C’s camp. However, if A were to win it would be your nightmare. She would raise spending for all these new programs and blow up the deficit. So, seeing as A and B are so close in the polls, you decide to vote for B who isn’t really that bad.
This violates the idea that the candidate you vote for should be your favorite candidate. Ranked voting or a runoff system would deal with this issue. Nobody would get over 50% to start, and then C’s voters would all go to B, allowing him to win with 60%. C’s voters would still vote for their preference first. However, these systems still have issues with tactical voting in different circumstances.
In a system where the top two candidates have a runoff election (a system used in France as well as some US jurisdictions), voters who favor an unpopular candidate with a small base have an incentive to vote tactically. Let’s say there are three candidates in the first round of elections: A, who is a right wing populist; B, who is a conservative moderate; and C, a moderate liberal. They are polling at 35%, 25%, and 40%. A’s voters know their guy will never win a one-on-one election against the moderate liberal candidate, so they might decide to vote for B in round one so as to have a chance that someone they share some beliefs with will win the general.
Tactical voting occurs in ranked systems as well. In the above scenario, if ranked voting is used, B will be eliminated because he has the fewest first choice votes. His voters’ second choice is the moderate liberal who will then win. In this case, some of A’s voters might put B as their first choice so that A is eliminated first and B wins the election. In fact, this seems like the right result from this election. 60% of voters lean conservative, so it would make sense for a moderate conservative to win. Tactical voting seems to be the only way of producing that result here.
A Silver Bullet?
Changing voting systems would solve at least one problem entirely: the “spoiler candidate” who earns just enough votes to deny a victory to an ideologically similar candidate. The Nader scenario would be solved by either ranked voting or a runoff. However, beyond that,it gets murkier. Some problems would be solved, but new ones would be created. Alternative systems produce winners who are less controversial. Ranking favors candidates who can build a coalition of voters who don’t particularly dislike them over candidates who have a devoted following. This can be seen as either a bug or a feature. Some of America’s most popular presidents today were incredibly divisive in their times. Abraham Lincoln and FDR come to mind. Would a compromise candidate have been preferable with the Civil War looming on the horizon?
Advocates for ending First past the Post often only look at half the picture, as do advocates for keeping it. Every system will sometimes deliver a result that seems wrong to someone. Our own beliefs about who should win are tied up in our beliefs about what criteria we should use to measure victory. Those who want big change might think FPTP is ideal because it allows a vocal minority of 35% to surge to victory. Those who want consensus and moderation might prefer ranking or a runoff system. There are good arguments for and against every voting system: if we are to change it, it should not just be because our guy lost in the current system. It should be because we have determined that the system is no longer in line with our values.
Sources and further reading:
“Why Flip a Coin” by H.W. Lewis
Bonus: The idea that there is no perfect voting system actually has some mathematical validity. Arrow’s Impossibility Theorem says that no system can simultaneously fulfill all the criteria we want from a voting system if there are three or more candidates up for vote. There are a few caveats, of course, but they are too technical for me to go into here. But there is one interesting idea that gets brought up in response to Arrow that I want to mention. That is “cardinal voting.” Basically in a cardinal system, your ballot lists all the candidates who are running and asks you to check off all the ones you approve of. The one with the most check marks wins. If we decide we want the ideal consensus candidate, this would be a good way of running elections. It gives you the right answer 100% of the time if that’s what you’re looking for. Of course, it might be a boring bureaucrat, or even someone nobody’s ever heard of.