Some time ago, Washington Post presented an article about flaws of mail voting. Another article tells about "tens of thousands" such ballots.

And that caused me to ask this question:

Is there any known average amount of missing/rejected/not counted mail ballots?

For example, in 2016 elections?


1 Answer 1


Yes, the Election Administration and Voting Survey - conducted biennially since 2004 by the Election Assistance Commission - collects this data.

The 2016 report can be found here. In particular:

Approximately 80.1 percent of absentee ballots that were transmitted to voters were returned and processed, with 1.4 percent of transmitted ballots returned as undeliverable and 2 percent reported as spoiled (e.g., the voter returned the ballot and asked for a replacement).

Ninety-nine percent of absentee ballots categorized as “returned and submitted for counting” were ultimately counted in the 2016 election.

enter image description here

To put a raw figure on the number of ballots rejected, the report also states that "Nationally, by-mail voting constituted 23.7 percent of all votes cast in the 2016 election".

The FEC gives the total number of votes cast in 2016 as 136,669,276, so, according to the EAVS, there were approximately 32,390,000 by-mail ballots cast. As these, presumably, are the 99% which were counted, this means that approximately 327,000 by-mail ballots returned were rejected. As 19.9% of ballots sent to voters were not returned, we can calculate this figure as around 8,128,000, of which around 572,000 were returned as undeliverable, and 816,000 reported as spoiled.

  • 1
    This is an excellent answer; while the question doesn't ask for it, the only thing I might hope for would be a comparison to the percentages of in-person votes that are rejected and reasons therefor. Oct 15, 2020 at 17:08
  • Just to make sure I'm interpreting that table right, the rows below "rejected (total)" are a subset of "rejected (total)," so late ballots made up 23.1% of that 1% of ballots that were rejected. Is that the right way to read that?
    – Davy M
    Oct 15, 2020 at 21:12
  • 2
    @DavyM Exactly - it's a pretty weird way to lay it out, but that's how I think it's meant to be interpreted.
    – CDJB
    Oct 15, 2020 at 21:17
  • So would that mean of all of the ballots 0.275% (1.0% x 27.5%) had a mis-matched signature? Which lets say 50% of those were because the signature changed (say changed name and signature after getting married) So it is possible that 0.1375% (0.275% x 50%) were fraudulent. Is my understanding of those numbers correct (Assuming my 50% of mis-matched have legitimate signature change assumption is correct)? Oct 15, 2020 at 21:40
  • 2
    @darcy I mean, 50% is totally made up, so I would hesitate to say you're correct. Furthermore, check out NYT's article on signature matching. I think guessing at how many signatures are fraudulent is pointless. There's no evidence tens of thousands of ballots were forged that way in 2016 Oct 16, 2020 at 5:14

You must log in to answer this question.

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