There's currently a controversy over the Iowa caucus results, where it's taking several days to finalize them due to a "glitchy app" preventing volunteers from submitting the results immediately after the caucus has been completed. But why is the whole process so difficult compared to a regular election? Tens of millions of ballot papers are quickly counted every November in the US and the results are more or less finalized within 24 hours after the last poll is closed. What makes the Iowa caucus so special?

The same problem happened with the Republican caucuses in Iowa in 2012, so there seems to be a pattern.

4 Answers 4


This question is rather weird. The questions should be why is it difficult to collect the results quickly and accurately. Because they were collected quickly e.g. in 2012, but took 16 days to certify, eventually producing a different winner than that announced on caucus day. The reasons, back then, were mainly typos.

There is no single reason why things went wrong in Iowa, but here's a historical summary of what did go wrong:

1976:Iowa Democratic officials ignored a rule in calculating delegates, resulting in exaggerated projections of delegates for front-runner Jimmy Carter. The Des Moines Register discovered the problem days later.

1980: The problems with the 1980 counts are legendary.

It appeared on caucus night that George H.W. Bush scored an upset victory, beating Ronald Reagan by as much as 6 percentage points.

But computer problems kept 165 mostly rural precincts in which Reagan figured to do especially well from being included in the tally. Two days later, the party’s re-examination showed a 2 percentage-point margin for Bush. CBS News had it even closer, with Reagan leading by less than 1 percent.

The final numbers represented 94.4 percent of the precincts — 142 precincts never reported their results or didn’t hold caucuses, according to Drake University political science professor Dennis Goldford, co-author of “The Iowa Precinct Caucuses: The Making of a Media Event.”

1984: The News Election Service, funded by a consortium of national TV networks and the Associated Press, was created to gather raw vote totals rather than the delegate equivalents that party officials had provided since 1972. But Democratic Party officials refused to cooperate, and the news service managed to tally only 74 percent of precincts.

The party’s own counts proved unreliable: Party leaders were sure Walter Mondale had come in first, but they weren’t certain where other candidates had finished. Many votes were never turned in.

1988: Democrats were still at odds with the news media. Questions were again raised about the validity of the News Election Service results, which were based on just 70 percent of Democratic precincts, Goldford wrote.

Republicans, however, continued to work with the News Election Service to ensure accuracy and legitimacy, and Bob Dole’s nearly 13 percentage-point lead was based on 98 percent of the GOP precincts counted.

1992: Again, Democrats refused to share caucus vote totals, only delegates won by each candidate. The GOP didn’t hold a vote to save its sitting president, George H.W. Bush, any embarrassment from challenger Pat Buchanan.

1996: Republicans decided to do no official party count. Instead, the tabulating was handled by the Voter News Service, which replaced the News Election Service. Still, some GOP candidates thought the VNS results showing Dole first and Buchanan second were flawed.

2008: Republicans handled their own count and experienced some data entry problems. Mike Huckabee won, with Mitt Romney in second. Democrats reported a win for Barack Obama, with John Edwards and Hillary Clinton nearly tied for second.

2012: Iowa Republican officials moved their tabulating center from party headquarters to an undisclosed location to ward against hackers and protesters. County officials reported the precinct votes by phone with live call-takers or logged in to a security-code-protected website.

So yeah, counting is easy, except when people are involved. If I were to summarize from the above, it seems that party-ran caucuses don't have a great track record, but that may be a biased conclusion without comparing it with elections that were held at the same time. It seems somewhat less controversial that finding the right level of outsourcing for party elections has be a tâtonnement process, in Iowa at least.

The NYT now has a more detailed article on this year's problems. The updated instructions don't appear to have been followed everywhere, there were typos involved again, and even data copied in the wrong columns. Because more data was reported (electronically) than in previous years' caucuses, the problem was somewhat different:

Just about every election night includes reporting errors. They can be difficult to identify, but can often be corrected during a recount or a postelection canvass. This year’s Iowa caucuses are the reverse: Errors are now easy to identify, and hard to correct.

  • 2
    VERY impressive list. Just stunning. I've never imagine, that it is done so. Feb 6, 2020 at 14:17
  • 2
    tâtonnement yikes, you dove deep for that one.
    – asgallant
    Feb 6, 2020 at 22:11

Why is it so difficult to collect Iowa caucus results...

It isn't. While other answers explain why it may be more difficult than ballot-election results, past caucus-based election results in Iowa were published in a timely fashion (although TBH I haven't dug back in history very far).

There are nearly 1,700 precincts where caucusing is done, plus nearly 100 satellite precincts out-of-state. Each of those has a "precinct captain", a party official in charge of result reporting. I think but am not 100% sure that's always a single individual.

This time, as opposed to previous caucuses, and for increased transparency, the captains were required to report more figures: Not just the final contribution to each candidate's tally, but also fraction and/or number of voters in the two alignments in the caucus. But reporting 50 numbers instead of 20 numbers, or whatever it is, isn't that much more difficult; and the counting of these numbers is not new.

So, let's say 1700 people needing to reporting a sheet of, say, 50 numbers to a central location. Really, how hard is this? Even if push came to shove and the data had to literally be driven to where the central counting is performed in the state - the drive is under 300 Miles. If the app gave them trouble - they could have sent it via Signal, Telegram or WhatsApp. Yes, that's not very secure, but then - using a personal smartphone is already pretty insecure already. Some volunteers or paid party operatives at the central counting locations can certainly handle adding up 1700 * 50 numbers quickly, even if those arrive as plain text, or a photo of a page, or physically. Even if you add a few hours for reviewing, cross-counting etc. - the results should still have been published the following morning. End of the day max.

  • 8
    So why were there quickly published initial GOP results wrong in 2012 then? politics.stackexchange.com/questions/50099/… Your answer says that that should (or could?) never happen basically. Feb 5, 2020 at 20:11
  • 3
    But why did it take 16 days to find the error? Your post says "reviewing" the results should take a day at most. Feb 5, 2020 at 20:24
  • 3
    Also "The problems with the 1980 counts are legendary." Care to explain that too? Feb 5, 2020 at 20:55
  • 4
    This answer provides nothing but speculation and adds nothing of value to the discussion. If you have references for the information, then please add them.
    – Joe
    Feb 6, 2020 at 16:16
  • 10
    @einpoklum Honestly, you should source that, also; but that's not the majority of my qualms with this. "Really, how hard is this" and such - you're just criticizing without facts. Unless you're a precinct captain with specific knowledge of this, this is no better than a reddit post.
    – Joe
    Feb 6, 2020 at 17:08

One of the major gripes that came out of the 2016 caucuses was that a close and contentious vote was not backed up with vote totals, meaning no recounts were possible.

Mr. Sanders said on his flight to New Hampshire late Monday night, after a virtual tie with Hillary Clinton, that he would ask the Iowa Democratic Party to reveal the raw vote count underlying the percentages it reported showing Mrs. Clinton defeating Mr. Sanders, 49.9 percent to 49.6 percent.

The state party does not report raw votes. Caucus attendees at each of 1,681 precincts elect delegates to a county convention, who in turn will elect delegates to a state convention. The results from caucus night are reported as “state delegate equivalents,’’ expressed as a fraction. After 171,109 Iowans turned out, larger than almost all projections, the party said Mrs. Clinton won 700.59 delegate equivalents and Mr. Sanders had 696.82. A single precinct, Des Moines 42, worth 2.28 delegate equivalents, had not been counted.

A major function of the new app was to collect that data. Unfortunately, there were some major problems, most notably that the app was rolled out at the last minute

"Caucus chairs in many cases apparently were attempting to download and install the app on their phones on caucus night. That's extraordinarily difficult, to do that kind of thing under pressure," says Jones. "Downloading an app at the last minute is crazy."

Elesha Gayman, the Democratic Party chair in Scott County, said their attempts were complicated by measures put in place to protect the reporting system against outside attacks.

"We had a lot of our precinct captains and temporary chairs, and permanent chairs, that were trying to log in to the system, and quite frankly there were so many layers of security, they would get messed up," she said. Out of frustration, they resorted to calling a backup hotline, which was quickly overwhelmed with calls.

And to top it all off, this wasn't being run by the State of Iowa (which has systems to collect and count votes), but by local party bosses who had little experience in doing this to begin with

Iowa caucuses are different from primaries and general elections because they're conducted by party officials, not by people who run elections as a full-time job.

It's important to note that Democrats and Republicans run their caucuses differently, so the 2012 Republican problems are unrelated to the 2016 and 2020 Democratic Caucus.

  • They could've just drived with the records to a central location. Would've taken 5-6 hours at most for the most distinct precincts. Feb 5, 2020 at 21:08
  • 1
    Possibly, but let me offer something else from personal experience. They were asking them to enter data into the app and it appears that it took them a long time to realize there was a problem. As such, any attempts to get more data were probably hampered by having to contact people (who thought they were done) that they had to pass the data manually now.
    – Machavity
    Feb 5, 2020 at 21:10
  • ...and if you thought you were done reporting, you might not actually have a backup of the data to report any more.
    – T.E.D.
    Feb 6, 2020 at 10:31
  • "Caucus chairs in many cases apparently were attempting to download and install the app on their phones on caucus night." does not necessarily mean the app was rolled out at the last minute.
    – Jontia
    Feb 6, 2020 at 11:47
  • @Jontia They deployed the infrastructure, yes. Nobody thought to do a dry run and make sure that people could use it. In this case, they ran into the same problem Mitt Romney did in 2012, except that ORCA was tested to work, but not under load, while this app was apparently not tested by actual precinct workers at all. That's why that person said it was "crazy" to be downloading it election night. Testing things in production can end very badly
    – Machavity
    Feb 6, 2020 at 13:33

Primaries are run by the parties in accordance with the rules of the state on primary voting, so problems with collecting the votes in primaries are exclusive to the state's chapter of that party. The Republican Iowa Caucus of this year released it's results in a timely fashion (yes there were other candidates besides Trump. If you haven't heard of them, keep in mind that the state went ~95% for Trump overall with many precincts as high as 100%. That means the remaining 3 candidates split 5% of the vote between themselves... the Libertarian party candidate has a better percent of the general election popular vote than that.).

If I recall, the delay in 2012 was that the numbers were really close while the 2020 delay is due to technical difficulties. In the long run, it's not the end of the world, and historically, Iowa has been a poor predictor of Party nominees (I believe in 2012 Rick Santorum won the cacus but second place Mitt Romney won the nomination). Where it's important is that, when the winner is determined in a timely fashion, they're victory speeches get wide politicization in other early primary states, which can translate to momentum. This is why you saw all the Dems declare victory, make a speech, and then hop on the plane for New Hampshire (well, they may have gone to DC for the state of the union and then some of the front-runners have to vote on the verdict of Impeachment the following day. But it's usually New Hampshire by sunrise or bust... if you haven't made your victory speech then, it's over and there is no recovery of that moment.).

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .