6

According to uselectionatlas.org, Clinton is currently leading Trump by 466,220 votes. Of course, there are still 400,000 votes to be counted in Arizona alone, so it's far too early, in my opinion, to say who won the national popular vote, but that still does not prevent folks from trumpeting otherwise.

In any case, the final tabulation will be very close and Clinton might likely come out on top. With that in mind, how accurate will the final vote count be?? There will be miscounted, uncounted and fraudulently counted votes to be sure, but I have no idea what percentage of the total they will represent.

|improve this question
  • 3
    We usually don't call those kinds of issues part of the "margin of error". The margin of error is a statistical term representing the likely accuracy of a sample in predicting the performance of the whole. The vote is the actual whole in this case. Therefore it wouldn't have a margin of error. – Brythan Nov 11 '16 at 16:20
  • I think miscounted counts as part of the Margin of Error. Uncounted and Fraud are different issues outside of margins of error. – user1530 Nov 11 '16 at 16:25
  • 1
    @Brythan If you can suggest a better term, I will be happy to update the question. How about: "How accurate is the final vote count? ", clearer? – Michael J. Nov 11 '16 at 16:48
  • 1
    @blip I don't know that this is the place to get into this particular issue (Cross Validated: stats.SE would probably be better), but Margin of Error is a purely statistical term regarding how the sample reflects the whole. It doesn't include any source of error outside sampling error (meaning picking the wrong people to ask). It's a purely mathematical measure based on sampling randomness. People often act like it includes other sources of error, but it doesn't. That's why it can be so precise. – Brythan Nov 11 '16 at 18:02
  • 1
    @blip Yes to both. – Brythan Nov 11 '16 at 18:44
6

Since voting varies so much by state, it may be hard to come up with an absolute error rate. The Al Gore election in Florida resulted in a number of 3%:

The National Opinion Research Center at the University of Chicago, sponsored by a consortium of major United States news organizations, conducted a Florida Ballot Project comprehensive review of all ballots uncounted (by machine) in the Florida 2000 presidential election, both undervotes and overvotes, with the main research aim being to report how different ballot layouts correlated with voter mistakes. The total number of undervotes and overvotes in Florida amounted to 3% of all votes cast in the state. The review's findings were reported in the media during the week after November 12, 2001.

https://en.wikipedia.org/wiki/Florida_election_recount

Note that that would be on the extreme high side, as there were major issues with the ballots in Florida.

In other states, they have automatic recount laws if the margins are at a level they consider "within the margin of error". In Minnesota, for example, will trigger a recount if the margin is less than .25%:

For federal, statewide and district judicial races, a losing candidate can request a recount if the margin of victory is below 0.25% of the total number of votes counted or is ten votes or less and the toal number of votes cast is 400 votes or less. See MS 204C.35(1)(a)(2) and (b)(2).

http://www.ceimn.org/ceimn-state-recount-laws-searchable-database/states/Minnesota

So based on those two random anecdotes, the completely unscientific wild guess is that the margin of error could fall somewhere between .25% and 3%

|improve this answer
-2

That last conclusion is mistaken since it does not take into account that there are states where the margin of victory for one candidate is enough so that it does NOT trigger a recount, even though the count can be fairly inaccurate. That being said, the national vote tally is exacerbated by the amount of states that are lopsided one way or another. Based on this, the likelihood of the national vote tally being even within 2 percent of the real tally is very unlikely.

|improve this answer
  • We generally want answers that completely answer the posted question. This doesn't do a great job of addressing the question ("How accurate is the popular vote count?") and will likely attract downvotes. You can edit your answer to improve it at any time. – indigochild Nov 28 '16 at 17:03

You must log in to answer this question.

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