In 2004 San Francisco began using Instant-Runoff Voting for local elections. Early analyses revealed that after its inaugural use the rate at which voters ranked candidates dropped dramatically, and disqualifying errors on the ballots were more common in neighborhoods where more African-American and foreign-born citizens reside (Neely and Cook 2008). We extend the inquiry over time, examine other types of elections, and refine the methodology. Are some voters more than others able to exercise the franchise in these elections and, if so, what does that imply for the equality of political voice? We find no evidence that ranking candidates is in decline. Overvotes are more common in IRV than plurality contests, but such errors appear to be a function of complexity in general and not IRV per se. While they occur disproportionately in precincts with more African-American residents, and are often more likely where Latino, elderly, foreign-born, and less wealthy citizens reside, the pattern of overvoting is similar in both IRV and non-IRV contests. We discuss what these results imply for San Francisco and other jurisdictions using or considering similar election systems.