CD5 candidate Christine Jones and the Jones for Congress campaign sent a letter to Arizona Secretary of State Michele Reagan and Arizona Attorney General Mark Brnovich on Monday calling for an investigation into voting anomalies. The detailed letter cites anomalies they identified that occurred during the tabulation of early ballots by the Maricopa County Recorder’s Office following the Primary Election.
A mere 16 votes separate Jones from the race leader, Andy Biggs.
The Maricopa County Recorder’s Office has been shrouded in controversy since the Presidential Preference Election earlier this year in which voters were disenfranchised.
According to the attorneys, “The tabulation of early ballots revealed a consistent statistical trend between the 4 candidates until the final batches of ballots were tabulated on Friday, September 2, 2016 (8,451 total). At that time, Andy Biggs pulled ahead of Mrs. Jones in a manner that differed from the tabulation trend and his vote surge therefore appears to have been caused by some external event.”
The lawyers referred to that as the “Friday Tabulation Anomaly.”
The attorneys claim that “there is an unexplained drop in ballots tabulated from certain Chandler precincts during the “Friday Tabulation Anomaly.” They allege that “an entire region of Chandler appears to be missing.”
The attorneys argue that the “number of overvotes in the CD5 election at 577 is unusually large and requires a further inquiry into how this happened to be sure that the overvotes were not caused by a machine error.”
Attorneys advise the Secretary of State and the AG’s Office that “investigating these anomalies as well as other potential irregularity with the CD5 tabulation is within your authority pursuant to your respective duties during the recount.”
“Such investigations also comports with your respective obligations under ARS 16-452 to assure the maximum degree of correctness, impartiality, uniformity, and efficiency on the procedures for early voting and election day voting and of producing, distributing, collecting, counting, tabulating, and storing ballots,” reads the letter.
The attorneys suggest that the officials may have to hire an expert in research methodology and statistical analysis to review the tabulation pattern.
The letter reads in part:
1. Friday Tabulation Anomaly
Over the course of the evening of August 30, 2016, following the closing of the polls, Maricopa County made several updates to the unofficial election results. As is normally the case, early ballots that were mailed in and accepted by the County were the first to be posted. This grouping of ballots is typically the largest percentage of total ballots cast in any given race as it is a collection of ballots received by the County for the period in which ballots are accepted prior to the date of the election. The linear sequencing of these ballots is not quantifiable since the county posts all the ballots at once. What is clear, however, is that Ms. Jones succeeded in winning the majority of these ballots by 1.88 percentage points over Andy Biggs.
The County next tabulates votes cast by voters at their polling places on Election Day. Because these voters participated on Election Day, it is likely that their behavior is influenced by events and messaging that occurs during the final 72 hours of a campaign. This may explain any deviation from the early ballot results previously tabulated. In the case of CD 5, it became clear following the early and Election Day postings on the evening of August 30, 2016, that Mr. Biggs had gained vote share and ultimately led among the late voting electorate. He led Ms. Jones by 1.22 percentage points among all ballots cast on Election Day.
After all early ballots received prior to Election Day, as well as Election Day precinct ballots are processed, Maricopa County next begins to tabulate early ballot “drop-offs.” These are early ballots that are dropped off by voters at polling places throughout the County or that the County receives by mail on Election Day. It takes a number of days to process these ballots as they make up a considerable portion of the electorate and must be processed and verified. It is standard practice for the results of these ballots to be posted over the course of the days following Election Day.
There is no way to know when these ballots were filled out by the voter. Because of their close proximity to Election Day, however, they usually behave similarly to Election Day voters. Although less common, there are some instances in which they will correlate more with early ballot voters. In the case of CD 5, Thursday’s results posting suggested that the drop-offs would correlate with Election Day voters, as Mr. Biggs maintained his position with 32% of these votes and Ms. Jones trailed with 29%, which is a difference of 3.03 percentage points. 1
Friday’s ballot drop-off proved to be something else entirely. Justin Olson, who had previously remained in the low twenties in percentage points, posted a 25.33% vote total of the 7,551 ballots. His increase was almost exclusively taken from Ms. Jones’ performance and she sustained her largest percentage drop of the election, creating a gap between her and Mr. Biggs of 7.16 percentage points. As noted, this Friday Tabulation Anomaly is inconsistent with the tabulation trend and may have been caused by some external event.
The Friday Tabulation Anomaly is starkly illustrated in the chart and table below, which shows the percentages of tabulated results from election night to the final provisional ballots tabulated on early Saturday morning. We understand that the drop-off ballots are randomly combined throughout the tabulation process and therefore it is not clear why there is such a large deviation in the Friday posting of early ballot drop-offs. This is demonstrated in the chart below where Thursday’s drop-off and Friday’s second drop-off ballots are closely correlated, but Friday’s first drop-off is not in line with other drops.
These statistical anomalies demand a careful review to ensure the integrity of the voting system is intact, especially due to the upcoming General Election.
2. Chandler Ballot Drop Anomaly
Another concerning aspect of the Friday Tabulation Anomaly is the stark drop in ballots from certain Chandler precincts. Although Maricopa County insists that early ballot returns are tabulated in a random fashion, the attached map details the total of all Friday ballot drops and the turnout percent that it contributed to each precinct. See Exhibit A.
While there are certainly precincts in the district in which this drop contributed a sizeable portion of the turnout, a large portion of the Chandler region is substantially underrepresented in the drop. The other attached map shows how this is the only cluster of precincts in the entire County to provide no significant contribution to Friday’s drop, with the exception of very low turnout areas of CD 7. See Exhibit B.
This is concerning because Ms. Jones won Chandler, 34% to 27%. With such a significant gap between Mr. Biggs and Ms. Jones in the Chandler precincts, we believe the Thursday night drop should have been much closer between the two had the ballots simply been factored into that night’s postings.
If the counting of ballots was indeed random, a proportionate share of the results should have come from the Chandler region. The stark lack of ballots from the Chandler region being counted during the Friday ballot drops raises concern because it could indicate that the Maricopa County Recorder did not employ a random tabulation process.
3. Overvote Anomaly
A final inquiry regarding the race is the inordinately high number of overvotes. There were 577 overvotes in CD 5. See Exhibit C. Overvotes are votes that are rejected because the voter votes for more than one candidate in a particular race. High overvotes are common in races in which voters are allowed to select more than one candidate, but they rarely occur in races where the voter is asked to vote for only one candidate. These overvotes can occur if the voting machine scanner picks up stray marks on the ballot. We urge you to visually inspect these ballots to determine if these overvotes were caused by voter or machine error to ensure that the voters’ intent is honored.
* * *
The sanctity of free and honest elections is the cornerstone of our democracy. ARIZ. CONST. art. VII, § 12. Thus, it is imperative that you investigate the anomalous tabulation questions we have identified in this letter with respect to the CD 5 election to ensure that every lawful vote is counted. With only 16 votes separating Ms. Jones and Mr. Biggs, it is easily foreseeable that the final election outcome could change if the identified anomalies require that votes be added or subtracted to the final totals.
As you are no doubt aware, we have already established in Jones v. Reagan, No. CV2016-014708, that certain lawful votes were not counted as a result of systemic election official error. See Exhibit D. That suit, however, was limited to issues related to wrong precinct voting and early ballot affidavit signatures. Id. Moreover, despite Judge Rogers’ finding that such policy led to voter disenfranchisement, the County Recorder refused our request to tabulate all 136 provisional ballots in CD 5, which were cast in the wrong polling location as a result of the County’s policy to not inform these voters that their votes would not be counted.
The serious issues identified in this letter must be addressed by you to ensure that the CD 5 election is conducted in a fair, orderly, and honest manner and to maintain the public’s trust and confidence in the election system.
Finally, we urge you to recount the CD 5 ballots on “an automatic tabulating system” that is different from the system used by Maricopa County. A.RS. § 16-664(A). Aside from a hand count of all ballots, this is the only way to ensure that the tabulation error did not occur as a result of the tabulation equipment itself. We respectfully ask to observe all recount proceedings conducted by both of your offices.
1 Indeed, County Recorder Purcell noted to Ms. Jones’ tabulation observers at the MCTEC tabulation facility on Friday, September 2, before the Friday tabulation results were released, that a differential above 3.0% would be unusual.