Past Election Data Predicted High Independent Voter Turnout In March Election
On March 22, when Maricopa County voters had to wait in line for hours to vote in the presidential preference election at just 60 voting centers, Maricopa County Recorder’s Office staff said they were surprised by the number of independents who turned out to vote since only members of the Democratic, Republican or Green parties were allowed to participate.
“The problem is these independent voters are showing up, and they are not eligible to vote in this election,” Elizabeth Bartholomew of the Maricopa County Recorder’s Office told KJZZ on election day. “So they’re taking time; they’re waiting in line; the poll worker has to explain to them why they’re not eligible to vote.”
But while county election officials were seemingly caught off guard by the number of independents at the polls this year, a KJZZ analysis found that even more independents tried to vote in 2008’s presidential preference election— the year county officials used as a model to predict turnout.
According to records provided by the Recorder’s Office, 20,043 independent or unaffiliated voters in 2008 in Maricopa County cast provisional ballots that were ultimately rejected. That is slightly more than the 18,322 provisional ballots cast by independents this past presidential preference election.
In the aftermath of the March 22 election debacle, county election officials were criticized for reducing the number of polling sites from 400 sites in 2008 to 60 larger vote centers that voters could visit regardless of residence. The last time both major parties had a competitive presidential primary was in 2008.
Recorder Helen Purcell took responsibility for the troubles at the polls, but also insisted her plan to have just 60 vote centers was based on historical turnout data.
During a legislative hearing days after the election, county officials explained they had projected months in advance there would be 310,000 eligible voters who were registered to a party and not on the Permanent Early Voter List. They looked at 2008 in-person voter turnout to predict 23 percent of those voters would actually show up to vote, so they were expecting only 71,300 eligible voters on March 22.
But there were problems with all of those estimates. There was far more interest in this year’s presidential preference election than election officials predicted, since the race was more competitive than expected and candidates visited the state the weekend before the election and encouraged voters to go to the polls.
The actual universe of party-affiliated, in-person voters wound up being more than 341,000. Their turnout rate was close to 26 percent.
While the plan to have 60 voting centers did involve a buffer — election officials prepared for an extra 300 people at each location — the model did not include hard predictions for unregistered individuals and independents who might also get in line even though they were not eligible to participate.
In actuality, on election day more than 108,000 people went to the polls and tried to cast ballots, of them some 88,000 were eligible to participate.
There were more than 20,000 provisional ballots cast that were rejected, including 18,322 provisional ballots cast unsuccessfully by independent voters.
“Did we factor in that some of them would probably go to the polls?” Purcell said of the independents who turned out. “That is what we always have to do. But we thought not that great number.”
Purcell acknowledged that in actuality the numbers were similar this year compared with 2008. She said she believed the number of independents would drop this year because of voter education efforts.
Both the Recorder's Office and the Secretary of State’s office used social media and events to spread the word that voters had to register as Democrat, Republican or Green in order to vote on March 22. The Citizens Clean Elections Commission spent $200,000 on media to educate voters about that message.
Tom Collins, the commission’s executive director, said it is unknown if the number of independents could have been even higher without that outreach.
A complicating factor is that a portion of the independents who got in line at the polls mistakenly believed they were affiliated with a party or were not aware their registration had been switched after making updates with the Motor Vehicle Division, which is an issue that various government agencies are investigating.
The Recorder’s Office maintains another factor that should not be overlooked - far more voters voted in person in 2008 than in 2016, because early voting by mail was not yet as popular as it is now.
Purcell acknowledged that one reason for adopting the 60-vote-center plan was financial, since the state legislature has not yet reimbursed the counties for the cost of the presidential preference election.
When it comes to the calculations, Purcell stands by her decision to base the turnout model on the potential eligible voters, rather than on independents and unregistered voters who are likely to show up despite being ineligible.
“They are not eligible to vote in the election, so they are certainly not in the plan to put together how you conduct that election,” Purcell said.
Purcell said a lesson from this election is that in the future more polling locations should be added to accommodate all voters who show up.
“If that means that allows people who are not eligible to vote in an election to go to the polls, then that is what we have to do,” Purcell said.
Still, she insisted such an effort would be atypical. “When anybody is looking at an election, you are looking at the people who are qualified to vote in that election,” Purcell said.
But Rice University political scientist Bob Stein said best practices dictate that Purcell should have done a statistical model to predict everyone likely to show up to the 60 vote centers, whether or not they were eligible to have their ballots counted.
Stein, who is an expert in the adoption of vote centers, said that analysis should have also included projections of provisional ballots, which take longer to process.
“I don’t think they did the due diligence necessary for estimating what the turnout was going to be,” Stein said. “What the turnout of Republican, Democratic and independent voters.”
Stein said it is not a good idea to switch from traditional polling sites to vote centers during a high-turnout election, or a difficult-to-predict election like a presidential preference contest. He said it is also best to only reduce the number of polling locations by no more than 50 percent from one election to the next.
This is the first presidential election year that Arizona election changes do not need to be precleared by the U.S. Department of Justice. That’s because a U.S. Supreme Court decision three years ago struck down a portion of the Voting Rights Act, ending that requirement.
But Stein said Purcell would have benefited from that review.
“A preclearance report is very detailed and would have made her realize very quickly she was in trouble,” Stein said.
Jim Barton, a lawyer for the Arizona Democratic Party agreed.
He said if there were more than 20,000 independents in 2008, Purcell’s team should have known 60 vote centers wouldn’t be enough this time.
“Frankly, the numbers that we are looking at right now, now cause more questions than answers,” Barton said. “Because now it is really hard to understand how they could have thought this would have worked out.”
In fact, not everything went very smoothly in the 2008 presidential preference election, either. Even with 400 polling sites, news accounts from that election documented confusion over who could vote and long lines.
Stetson University College of Law professor Ciara Torres-Spelliscy, who specializes in election law, said election officials should have paid closer attention to what happened in the past.
“2008 was a teachable moment for Maricopa County, and Maricopa County did not learn their lesson,” Torres-Spelliscy said.
2008 Presidential Preference Election Provisional Ballots
- Total provisional ballots: 51,367
- Counted: 24,433
- No party affiliation: 20,043
- Were not registered: 2,142
- Attended wrong polling place: 2,518 (not a factor in 2016)
- Other: 2,501
2016 Presidential Preference Election Provisional Ballots
- Total provisional ballots: 24,639
- Counted: 4,631
- Rejected: 20,008
- No party affiliation: 18,322
- Were not registered: 1,162
- Other: 524