The "Center Squeeze" Effect
The “center squeeze” effect is a common type of electoral scenario in which approval voting behaves better than plurality voting, delayed runoff, or instant runoff voting.
Say we have the following voter preferences:
% of voters Their ranking
35% conservative > centrist > liberal
33% liberal > centrist > conservative
22% centrist > liberal > conservative
10% centrist > conservative > liberal
[E.g. the first row says that 35% of the voters prefer the conservative, over the centrist, over the liberal.]
Now whom should we elect? Which candidate best represents the will of these voters? Note that the centrist is preferred to the conservative by a huge 65% majority (the 2nd and 3rd rows). And the centrist is preferred to the liberal by an even bigger 67% majority (the 1st and 3rd rows). It seems clear that the centrist is by far the most broadly appealing candidate.
If we use Instant Runoff Voting, the centrist is eliminated first, and then the liberal defeats the conservative, 55% to 45%. With an ordinary delayed runoff, the conservative and liberal would be pitted head-to-head in a second election, where we would expect the same result (unless a significant number of voters change their minds between the first and second election).
But either way, the centrist doesn’t win. This is called the “center squeeze” effect, because the broadly appealing centrist is squeezed from both sides by candidates who absorb most of the support of the two main sides of the political spectrum. Warren Smith, a Princeton math PhD who has specialized in election theory for over a decade, even argues that IRV favors “extremists”. The diagram below illustrates the effect.
Approval Voting & Centrists
More generally, approval voting tends to elect beats-all winners. This is because many of the liberals would also support the centrist, in order to prevent the conservative from winning. And many of the conservatives would support the centrist in order to prevent the liberal from winning. This behavior has been supported by decades of mathematical analysis and empirical data.
This is yet another reason why we believe approval voting is superior to plurality voting, delayed runoffs, and instant runoff voting. Here’s a more in-depth comparison between approval voting and IRV.