-9

I need a comprehensive dataset showing when individual votes were cast and for whom across the USA in the 2020 general election. I can find totals on the county websites and presidential candidate totals by county, but what I need at a minimum is a dataset that includes the timelines of all updates from each precinct during the vote-counting process. That is, what were the actual numbers of votes for each candidate reported by each precinct, and when were they reported? I think it is likely that many journalistic sites had access to this info but I have not found it publicly available anywhere. (It should be made public.) If possible I need a dataset that includes anonymized voter IDs and the complete ballot contents, (including rejected ballots and why they were rejected, e.g., whether a signature match succeeded or failed), and very importantly, their source (mail-in, drop-off, or voted in person). Timestamped data would be greatly preferred, including if possible, the marked time of receipt and the time each ballot was tabulated.

The icing on the cake would be to find historic voter counts and ballot details (and if possible, correlated IDs) for past elections for comparison.

Anyone know of a public dataset I can obtain that fits any substantial part of this description?

Improve this question
8
  • 1
    This type of stuff is not available until long long after the election is over. – Michael Mormon Nov 19 '20 at 17:04
  • @MichaelMormon Then surely there would be data available for past elections? (Why would it be withheld until after the election results are finalized, by the way?) – pygosceles Nov 19 '20 at 17:09
  • Any reasons for the hasty downvotes? – pygosceles Nov 19 '20 at 17:09
  • 5
    Most of the data you are asking simply does not exist ("the time each ballot was tabulated???") and the tone of the question is rather demanding ("I need" and other expressions) for such a high order of detail. – SJuan76 Nov 19 '20 at 20:40
  • 1
    @MichaelMormon what on earth are you talking about.. I'm looking at the 197*2656*4 row precinct tabulation of Georgia right now and I've plotted it and there are hilarious signs of fraud from all angles of statistical analysis – Lewis Kelsey Dec 4 '20 at 12:11
3

There's probably no such database freely available, although as you noted some media organizations have gathered the updates... and e.g. 538 has produced timeline graphs such as these based on those data sets.

enter image description here

I don't know if e.g. ABC and/or Edison Research charge for access to such data. Your best bet is probably to contact them and inquire on the conditions under which they'd make the raw data available to you.

In principle, anyone interested and determined enough could have created such a data set since the updates themselves were public information; it's just that they were overwritten on the state/county boards' websites etc. every time there was new data. (I've checked a few such sites and they only have the final tallies now.)

A different q here actually provided the raw data that NYT pulled from Edison; it looks like the raw files are still available by state, e.g.

Beware however that you don't get absolute vote counts except for the total; for each candidate you only get a (three digit) approximation of their vote share:

{"vote_shares":{"trumpd":0.593,"bidenj":0.384},"votes":958809,"eevp":17,"eevp_source":"edison","timestamp":"2020-11-04T02:11:23Z"},
{"vote_shares":{"trumpd":0.593,"bidenj":0.383},"votes":959682,"eevp":17,"eevp_source":"edison","timestamp":"2020-11-04T02:13:39Z"}

(Amusingly, this three-digit approximation is apparently a good part of a conspiracy theory.)

Additional searching finds a github page that pulled all this NYT in into one giant results.json file for all the states!

In the opposite direction, the NYT state data is actually obtained by them by "concatenating" precinct data (the state json files reference these)... and these precinct-level json files are also still available, e.g.

They contain actual vote counts for candidates (not approximations), but these seem to be totals and don't seem to have any timestamps, e.g.

{"precinct_id":"Bethlehem 11th Ward 1st District","locality_name":"Lehigh","results":{"bidenj":410,"trumpd":218,"jorgensenj":10},"vote_type":"total","is_complete":null,"geo_id":"42077-BETHLEHEM 11","precinct_name":"","locality_fips":"42077","is_geographic":true,"votes":638,"is_reporting":true},
{"precinct_id":"Bethlehem 12th Ward 1st District","locality_name":"Lehigh","results":{"bidenj":439,"trumpd":169,"jorgensenj":12},"vote_type":"total","is_complete":null,"geo_id":"42077-BETHLEHEM 12-1","precinct_name":"","locality_fips":"42077","is_geographic":true,"votes":620,"is_reporting":true},

Further investigation indicates that there probably are separate files for each such update, named with the exact timestamp of the update, e.g.

But the catch is that the state file misses the final portion of the timestamp (the decimal points after the seconds) in the update list it contains, so I don't know how to get all the precinct-level update files.

Improve this answer

You must log in to answer this question.

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