New Math: Forecasting Turnout Gets More Complex as Elections Do

[Image by Sidney Harris via sciencecartoonsplus]

After almost a week, controversy still lingers over the long lines that emerged during Arizona’s recent presidential primary. In Maricopa County (Phoenix), some voters waited 3 hours or more to cast ballots, and almost immediately afterwards there were heated debates over why – and who should get the blame.

It probably won’t surprise regular readers to learn that I don’t put much stock in the notion that the problem was a deliberate effort to inconvenience voters, but I still think the situation offers a lesson and/or cautionary tale for election officials across the nation.

The key factor in the long lines in Arizona – especially in Maricopa – was the sharp reduction in the number of polling places, which ran headlong into a pattern of high and late-breaking interest in the primary. USA Today had more the day after the election:

Maricopa County shifted to 60 polling places — down from the 200-plus in use during the 2012 presidential primary — to save money and to reflect the reduced demand for in-person voting as the number of voters who mail in their ballots continues to rise. In 2008, there were 400 polling places.

In contrast to four years ago, voters also could go to any of the 60 county polling sites. Four years ago, voters had to go to a designated polling place for their vote to count.

“It saves a lot of money,” said Elizabeth Bartholomew, communications manager for the Maricopa County Recorder’s Office, which conducted the election. “And there are fewer people who vote.” …

The county identified the 60 sites used as polling places by looking at areas with high mail-in ballot returns, and tried to get several sites in every major city. Phoenix had 12. Mesa had four. Cities on the edges of the county, such as Wickenburg and Gila Bend, had one site each.

But county officials appeared to miscalculate the level of enthusiasm for primary races that had not yet been sewn up.

Moreover, officials were caught off guard by the number of independents who showed up to cast ballots despite not being eligible to do so in the closed primary:

Complicating the confusion at the polls were registered Independents showing up to vote, leading to a surge in provisional ballots.

“These are Independent voters,” said Osborne. “They refuse to not vote, and federal law requires they be given a ballot.”

But if the voters casting the provisional ballot are determined not to be registered with one of the three parties, their vote will not count.

In hindsight, the causes of the problem are clear: lots of people in too few polling places with too many voters ineligible to vote. That created long, slow-moving lines and endless frustration on both sides of the check-in table.

I have no doubt that there will be a longer post-mortem on the Arizona lines, but I want to flag an issue that should be of interest to anyone who thinks “that can’t happen here.”

No issue is more critical to a successful and smooth-running election than a solid forecast of turnout. That number drives planning for polling locations, resource allocation (ballots, pollworkers etc.) and a host of other considerations that contribute to an uneventful and uncontroversial Election Day. The challenge, however, is that as elections (like Arizona’s) become more complex – with increased choice for voters – it’s no longer enough how many voters will cast ballots but also where, when and how.

Moreover, the Arizona experience suggests that part of the forecast should be identifying the likely number of people – not just eligible voters – who will show up, which can be affected by election laws and rules outside the control of the election office. Those people may not be eligible to cast ballots (or if eligible might be confused as to where), but they will still require time and attention to address on Election Day.

Of course, some potential problems can be alleviated by outreach and education in advance of Election Day, but given what we know about how people pay attention to politics and elections it’s not realistic to expect that everyone will get the message.

We are already seeing tremendous strides in many communities in addressing how to manage the flow of people in polling places, but as elections become more complex the next-level skill – the “new math”, if you will – is to have a solid sense of how many people will show up to vote PLUS an idea of where and when PLUS how many will require extra attention due to ineligibility, confusion and other issues. This is specially important as communities seek to manage costs by reducing the number of polling places; if the forecast goes wrong (as appears to have happened in Maricopa) then Election Day will, too.

This is easier said than done, of course, but it’s a critical skill that will be important in communities nationwide as voter interest – and fervor – increases between now and November.

Stay tuned …