By Brandon Quester, Arizona Center for Investigative Reporting

Our commitment to accountability journalism continues today with an analysis of rejected ballots from last year’s presidential election. We compared the number and rate of rejected ballots from 2012 to previous federal election cycles in our state. While rejected ballots help to document continued problems in Arizona’s election system, we also analyzed current legislation to see if our state legislators are proposing bills that will address the problems at hand.

How we did the analysis


AZCIR collected county-level rejected ballot data from federal elections in 2008, 2010 and 2012. Each county is required to collect and submit to the Arizona Secretary of State summary data on the type of ballots cast, rejected and the reasons for rejection. This information is compiled by the Secretary of State’s office and then submitted to the U.S. Elections Assistance Commission, which was created in 2002 by the Help America Vote Act to monitor nationwide election performance.

AZCIR then analyzed the data by calculating ballot rejection rates, or the total number of rejected ballots divided by the total number of ballots cast. This percentage calculation was done for total rejected ballots, by county, and by type of ballot cast. We then compared rates from each election cycle.

We used rates of rejection instead of the difference of raw numbers because Arizona’s electorate is disparate in the number of ballots cast in each county. For example, Maricopa County accounted for roughly 60 percent of all ballots cast. By using rates of rejection, AZCIR was able to compare counties’ raw numbers as a percent of total ballots cast in each county.

Still, comparing 2010 data to 2012 isn’t entirely fair. The 2010 election cycle was an off-year election, which included candidates for Arizona’s governor and Congress. It wasn’t a presidential election. These off-year election cycles tend to have lower voter turnout, which makes it difficult to compare with presidential elections.


While the 2010 data still documented the same issues we raise in our story, a comparison between presidential elections in 2008 and 2012 was a more accurate and relevant analysis.

Why didn’t the analysis include a demographic breakdown of each county?

Simply put, it’s not fair to correlate county-level summary data to demographic breakdowns from the 2010 Census. That type of analysis can’t determine if specific demographic groups cast the ballots we analyzed. The information exists, but the counties weren’t willing to – or couldn’t because of technology limitations – share that information.

In fact, AZCIR spent nearly two months trying to collect precinct-level rejected ballot data, which would have allowed us to do a fair comparison with demographic information from the 2010 Census.

AZCIR sent public records requests to almost half of Arizona’s 15 counties but did not receive precinct-level rejected ballot data. Some counties collect the data to monitor elections and provided the information without an official records request. Others don’t or can’t collect precinct-level data for rejected ballots.

Because counties aren’t required to gather that information and the technology used to gather it won’t allow elections officials to run a report detailing said data, according to county recorders, most counties could not fulfill the public records request.

The lack of consistent data collection between counties also hindered our joint analysis with the Arizona Capitol Times for our Mapping the Vote project. A commentary by Brandon Quester of AZCIR and Evan Wyloge of the Arizona Capitol Times discussed those issues. The commentary is relevant for AZCIR’s analysis of rejected ballots, too.

What’s the benefit of summary-level rejected ballot data?

Summary-level rejected ballot data can document at a county level where there is a disconnect between voters and election administrators. Sure, precinct-level data could detail specific areas by precinct, but summary data still provides an overall indication of election performance.

Rejected ballots are absentee, early voting or provisional ballots in which an individual went through the voting process but later had his or her ballot thrown out after review by elections officials.

These ballots can be thrown out because of reasons like a voter not being registered in time for the election to voting in the wrong precinct to not signing the ballot.

The reasons for rejection highlight education shortfalls between voters who don’t know the rules and regulations about voting or election administrators who haven’t done a good enough job educating voters.

But how these numbers relate to the rest of the nation also is important. Unfortunately, nationwide data has only recently been collected and analyzed for comparison. The Elections Performance Index, created by Pew Charitable Trusts and released in February, is the first-of-its-kind analysis that compares states on a series of indicators like the number of absentee and provisional ballots cast and rejection rates among each type of ballot, among others.

Nationwide data for the 2012 election is not yet available. AZCIR was instead able to use Arizona’s 2012 election data and compare that to previous elections here. That comparison, combined with Pew’s Election Performance Index that compared all U.S. states in 2008 and 2010, provided context for Arizona’s 2012 statistics.

Creative Commons License

Brandon Quester is the co-founder, executive director and editor of the Arizona Center for Investigative Reporting.