The Iowa Democratic caucus results are rife with potential errors and inconsistencies that could affect the outcome of the election, according to a review by the NBC News Decision Desk.
The apparent mistakes — spotted in at least dozens of the state's 1,711 precincts — call into question the accuracy of the outcome of Iowa's first-in-the-nation caucus, which was held Monday night.
In some individual precincts, it may be possible to fix the errors; in other precincts, it will probably be impossible to determine how voters truly made their choices.
The potential errors and inconsistencies take on importance because of the closeness of the contest between the two front-runners — former South Bend, Indiana, Mayor Pete Buttigieg and Sen. Bernie Sanders, I-Vt.
With all percent of precincts reported, Buttigieg had 26.2 percent support and Sanders had 26.1 percent.
The NBC News Decision Desk said it has reviewed the reporting data provided by the Iowa Democratic Party and identified several different types of potential problems with the count.
"There are definitely some inaccuracies in the data," said John Lapinski, director of the Elections Unit at NBC News, who leads the Decision Desk.
"When you have a tied result, even the smallest sort of inaccuracy could be consequential," he said. "If there was a lot of spread in this race, these errors would be insignificant. But when we are talking about a tied race, everyone wants to know that every number is correct."
"In close races, every number is consequential," Lapinski added.
Results from the contest held Monday were delayed by what organizers said was a problem with a smartphone app. The Iowa Democratic Party said problems with reporting the caucus results were due partly to "coding issues" with the app, which was being used by precincts for the first time.
NBC News has not called a winner in the first-in-the-nation contest.
Here are some examples of the potential errors and inconsistencies:
Numbers don't add up
The Decision Desk said it identified at least 77 precincts, or 4.5 percent, where the total votes for what is known as "reallocated candidate preference" is greater than the total votes for "initial candidate preference" — a difference that makes no sense.
In the Iowa caucus system, the reallocated preference is based on the raw tally taken after the process of realignment. For instance, if a caucusgoer's initial preferred candidate did not receive enough support to meet the precinct location's viability threshold (15 percent in most cases), the caucusgoer is allowed to shift their support — or realign — to another candidate who did attain viability.
Therefore, it doesn't make sense that the number of voters would increase during this process. If anything, it might decline since caucusgoers whose initial preference didn't make viability might choose to simply leave, rather than sticking around to support another candidate.
And yet the number of voters appears to have increased in precinct Des Moines-62 in Polk County. The total of votes recorded for initial preference was 784 — but the total number of votes in the precinct on the reallocated preference vote increased to 841. It is unclear how to explain an increase between rounds because the initial preference is supposed to include the votes of everyone who is participating in the caucus. The initial preference should reflect the total turnout in the caucus.
Some of the discrepancies are a result of zero votes being reported for the initial preference. In the Des Moines-80 precinct, for example, there were zero total reported votes on the initial preference vote and 215 total votes on the reallocated preference vote.
Because the precinct has no votes recorded for candidates on the initial preference round, it is impossible to know whether the reallocation was correctly done in the precinct. Moreover, because the precinct is not contributing any votes to the total candidate votes on the initial preference ballot, the statewide percentages for candidates doing well in this precinct also will be understated.
Other precincts, however, have discrepancies that are more irregular than simply failing to report any data for the initial preference vote.
In precinct WDM-312 in Polk County, there were only 61 total votes reported in the initial preference round, but there were 339 total votes reported in the reallocated preference round.
Looking at the pattern of recorded votes suggests that no initial preference votes were recorded for any of the viable candidates, but only for candidates who had not attained viability — which is not how the process is supposed to work. Because the data was not reported, it is therefore impossible to know what the initial preference votes were for viable candidates in this precinct, nor how voters changed between initial preference and reallocated preference.
Issues with state delegate equivalents
The Decision Desk also said it found allocations of "state delegate equivalents" — the caucuses' most important prize for candidates — that are hard to reconcile with the other data being reported out of the precinct.
Iowa Democratic caucus results are not actual votes cast. The percentages received by candidates, based on returns of the estimated number of state convention delegates won by each candidate through the caucus process, are known as state delegate equivalents, or SDEs.
Because the 41 elected delegates that Iowa is sending to the Democratic National Convention are elected by the state delegates selected by the Iowa caucus results, issues with the SDEs are potentially more consequential in terms of the overall Democratic nomination for president.
Under the rules, candidates should not receive SDEs if they have less than 15 percent of support in the reallocated preference vote.
Despite that, the Decision Desk said it found at least 15 precincts where a candidate received SDEs, despite being below the 15 percent threshold.
In all but the smallest precincts, the rules state that the viability threshold is determined by multiplying the number of voters by 0.15 and rounding up to the nearest whole number. However, it appears that the viability threshold was calculated by rounding down in several precincts Monday night.
No single candidate seems to have benefited disproportionately from this misapplication of the rules. There were at least six instances in which Sanders benefited, four favoring Buttigieg, two for former Vice President Joe Biden, one for Sen. Amy Klobuchar, D-Minn., and one for Sen. Elizabeth Warren, D-Mass.
An example of this is in the city of New Hampton Ward Three precinct in Chickasaw County. Forty-two voters participated, meaning that a candidate would need seven votes or more to be above the 15 percent threshold. In the reallocated preference, however, Sanders received only six votes, or 14.29 percent. Despite receiving below 15 percent in the reallocated vote in the precinct, Sanders received 0.1 SDEs from the precinct — when he should have received none.
NBC News has reached out to the Iowa Democratic Party about these potential errors and inconsistencies and has not received a response.
In a statement released Thursday afternoon, Iowa Democratic Party Chair Troy Price said that, "Throughout the collection of records of results, the IDP identified inconsistencies in the data and used our redundant paper records to promptly correct those errors."
Charlie Riemann, Josh Clinton, Stephen Pettigrew and Marc Meredith of the NBC News Decision Desk contributed reporting to this article.
