An Actual Vote Analysis is a comparison of values transcribed from poll tapes to the corresponding officially reported results as an independent double-check of the accuracy of the vote reporting. 

For the United States, America Counts defines the scope of an Actual Vote Analysis as any vote reporting in one specific election for one specific state. An Actual Vote Analysis typically covers multiple counties in that state, and multiple (if not all) precincts from each of those counties. A different election in the same state, or a different state in the same election, are each considered a separate analysis and would get separate results pages. The scope of the Actual Vote Analysis on this page is given by the Scope box above.

In North Carolina, officially reported results from an election are given in a format that can be easily transformed into one row per combination of County, Precinct, Voting Method, Contest, and Choice, with a single Officially Reported Vote Total for that combination. The four possible values for Voting Method are Election Day, Early Voting, Absentee, and Provisional.

In some cases in certain North Carolina counties, Early Voting, Absentee, and Provisional votes may be reported in different ways that are not directly at a Precinct level. The methodology presented here gets adapted to work in essentially the same way in these cases, so we don’t go into further detail here. Meanwhile, it seems that Election Day votes are always reported directly at a Precinct level.

Each printed results poll tape from that election in North Carolina represents one specific triple of County, Precinct, and Voting Method, and gives tabulated Poll Tape Vote Totals for in-scope Contests and Choices for that triple. In North Carolina, we have been told that it will always be one tabulating machine per triple, so we never have two or more tabulating machines for one triple each producing its own tape. If we did have that situation, we’d have to add up the corresponding Contest and Choice values for each poll tape from a given County-Precinct-Voting Method triple.

In an Actual Vote analysis, we say that one submission is the set of all videos and/or pictures submitted through the Actual Vote app and/or sent directly to America Counts that contribute to depicting one specific thing. This thing is most often a single Results Tape or a single Zero Tape, but it could also be a Configuration Report, a photo of part of the polling place closing procedure, and so on. Recordings for the purpose of practicing using Actual Vote are not counted as submissions. If a recording features two poll tapes instead of one, we count that as two submissions (though we ask users to stick to one poll tape per recording when possible). A submission may not depict all values on a poll tape that it was intended to document, since some values on the recording may be too blurry to read, or part of the poll tape is obscured.

We say that a transcribable submission is a submission (or excerpt from a submission, or combination of multiple submissions) that depicts one specific results tape, has state, county, election date, voting method, and precinct clearly readable in the header (so we know what official results to compare it to) and contains at least one legible contest/choice/vote total triple for us to transcribe. and compare to the officially reported results.

Submissions containing zero tapes or configuration report tapes are not transcribable because they don’t contain any vote totals that can be compared against the officially reported results. Actual Vote users are free to record zero tapes with the app if they wish, but valid zero tapes (ones that correctly show zero votes in all vote total fields) can’t tell us anything about the vote reporting process itself. Meanwhile, in the exceedingly rare case that 

A submission containing a summary report tape with vote totals aggregated at a high level can sometimes be compared to officially reported totals, depending on what specific information it contains. If such a submission can be analyzed in a reasonable amount of time and gives useful information about vote reporting, America Counts will classify it as a transcribable submission. Otherwise, we will classify it as a not-practical-to-transcribe submission.

We transcribe Poll Tape Vote Totals for each Contest and Candidate on each transcribable submission. In practice, many transcribable submissions are simply a well-composed and well-focused video-recording from the Actual Vote app that clearly depicts the entire poll tape from top to bottom, making the values easy to transcribe. In other cases, a transcribable submission might consist of a handful of hard-to-transcribe photographs taken as best as the user could during the polling place closing procedure.

The poll tape transcription for the analysis on this page was done by hand on a highly-structured Google Sheet.

In a given county, the number of transcribable submissions per Voting Method we have is a good measure of the extent to which Actual Vote is double-checking the vote reporting in that county. For example, if we get one transcribable submission for election day votes from each precinct in a county, we can do a complete audit of the election day vote reporting in that county.

County-Precinct-Voting Method-Contest-Choice-level Poll Tape Vote Totals transcribed from transcribable submissions can each be compared directly to the corresponding number in the Official Results. If the totals don’t match, the value is marked as an apparent discrepancy. In practice, the ingesting and pre-processing of the Official Results and the Poll Tape Transcriptions, as well their direct comparison, is done using Python, an industry-standard programming language for data science. It also produces the summary results shown on this page. 

An Actual Vote Analysis has two possible Analysis Windows: before the election certification date and after the election certification date. In both cases, our independent audit of vote reporting increases election transparency. But if all submissions are received and all analysis is completed in the before election certification date Analysis Window, any apparent discrepancies we find can be escalated to the relevant County election administrators or further while it’s still possible to update the official reporting if it is warranted. This outcome would mean an improvement in election accuracy.

On the other hand, the evidence from an Actual Vote Analysis whose Analysis Window is after the election certification cannot change the already-certified results. But such an analysis is still of interest, as it does give information about the state of vote reporting in the areas that it  covers, which increases election transparency.

America Counts reports results from the each analysis window separately.