Boston voters benefit from HAVA

One of the requirements in HAVA was that voting machines start warning voters when they over- and under-voted ballots. I have just been shown a small example of why this was a good requirement.

Boston had a very interesting (and important) mayoral election this year. I previously blogged about the layout of the polling places where I did some observation in the preliminary (“primary”) election held in September, to choose two candidates who later faced off against each other in the November general election.

One of the things I noticed in my observing was that it seemed that quite a few voters had to spoil their ballot, because they had over-voted for mayor.  I asked the Elections Department if they could give me the number of spoiled ballots in each precinct in the election — to see if more spoiled ballots led to more lines at the polls.  (That’s the topic of another posting.)  I’ve just received the spreadsheet with the data.  Here are the results:

Out of 113,319 ballots cast overall in the preliminary election, 3,597 were spoiled, presumably for over-voting the mayor’s race.  That’s 3.2% of all ballots.  In contrast, the final election returns reported that only 421 of the ballots contained no vote (“blank”) for mayor.  That’s a residual vote rate of 0.4%.  If the spoiled ballots had been voted instead, the residual vote rate would have been a substantial 3.6%.  As a consequence of the over-vote protection, nearly 3600 Bostonians had their preferences for mayor reflected in the mayor’s race.

You might ask, “why so many over-votes?”  The reason is simple:  the purpose of the election was to choose two candidate who would survive to the general election.  The radio stations during the day were urging people to go to the polls on election day to “vote for the two people who will face-off in the general election.”  It seems that some fraction of voters took their voting instructions from the talkers on the radio, not from the instructions on the ballot.

The over-vote protection does introduction confusion and delay into voting, but this is yet more evidence (however small) that it works to ensure that every vote does count.