Proponents of ranked-choice voting claim it incentivizes candidates to appeal to a broader base, and thus will lead to the election of candidates who better represent the median voter. However, this assumption does not always hold: when voters are divided by partisan, ethnic, geographic or cultural cleavages, some recent theoretical work shows that RCV can actually incentivize candidates to target their narrow bases (Buisseret and Prato 2024).

The sparse empirical evidence on this topic is similarly mixed. At the local level, RCV does not appear to be associated with changes in either the average ideology of candidates in city council elections or the variance in candidate ideology (Vishwanath 2024), which implies no effect on extremism or polarization. But some recent simulations with relatively strong assumptions have shown that winners under RCV tend to be less extreme on average, especially when there are fewer candidates in the race (Acharya et al 2024). Furthermore, in general elections featuring co-partisans (for example, two Democrats or two Republicans), RCV appears to give the more moderate candidate an edge since voters are far more likely to rate the moderate ahead of the extremist when ranking candidates from the other party (Gelman et al 2024).