There has been quite a bit of media coverage this week about the long lines that occurred at the polls in Maricopa County, Arizona during the recent primary elections. The questions raised include whether the lines were caused by having too few voting locations and confusion over new voter identification requirements. However, the bigger issue was likely a lack of staffing at the polling locations and lack of care in selecting where to locate the polls.
In our book Evaluating Elections: A Handbook of Methods and Standards, Michael Alvarez, Lonna Atkeson, and I discuss how simple data analysis can help local election offices avoid making simple election mistakes. A simple review of the historical data for Maricopa County suggests that having only 60 polling locations would require very large and very well staffed polling places.
In planning for the 2016 election, the election office could look at the 2012 primary election data as a baseline for what to expect in 2016. In 2012, 80% of all votes cast in the primary elections were cast during in-person early voting or by absentee ballot. Maricopa County clearly did not need to have the same number of polling places as would exist in a state like South Carolina, where the lack of early voting and no-excuse absentee voting means that most voters go to the polls in a primary election.
However, even with 80% convenience voting in 2012, there were still 63,852 voters who voted in person in the primary elections that year. The County had every reason to think that turnout would be higher than this in 2016, based on news accounts prior to the election With 60 polling places, and assuming the same turnout in 2012, the County had to plan that each polling place could accommodate, on average, at least 1,064 actual voters.
Handling this many voters requires having a very large number of voting booths and staff to check in voters. Presidential primaries have short ballots but they also have complex eligibility rules. The polls have to be set up so that voter check-in ensures that voters show the correct identification, registered Republicans and Democrats are given the appropriate ballot for the appropriate primary, and non-affiliated voters are screened out (Arizona has closed primaries). The polls also needed to be set up so that an election official can explain to voters with a registration problem the reason why they cannot vote in the primary. Some voters will think they are registered with a party but are not and some unaffiliated voters will not understand why they cannot vote in the primary. Addressing these problems on the fly, with voters waiting in line, can dramatically slow the voting process.
The other big issue with having a smaller number of polling places would be where to locate them and how to staff them. The County would want to locate polling places in centrally located, high traffic areas with a lot of space and a lot of parking. Using its voter file, the Country could also know which parts of the County had lower than expected convenience voting among its high propensity, party-affiliated voting population and located more polls in those areas.
Using the right data, Maricopa County could have had an election that went off without a hitch, even with only 60 polling places but it would have required high quality data analysis and operations management. This analysis would have ensured that the polls had the correct staffing, polling place locations and layout, and consideration to how many voters have to be accommodated during high demand periods. When these analyses are not done, lines are inevitable in elections.