EPEC NEWS

By Jon Lareau, volunteer CTO, EPEC

Reprinted from DigitalPollWatchers.org

We at EPEC have had a lot of interest in the work we’ve done following the numbers of non-citizen registrant removals by the VA department of election (ELECT).

I recently got asked a follow-up question as to how those numbers broke out with respect to estimated party leaning for each of the 435 identified non-citizen voters who cast over 1,000 ballots before they were removed from rolls by ELECT.

The method for performing this calculation is presented below.

The majority (~77%) could not be associated with either party as they either did not have Primary Election voting history or the computation resulted in a neutral score.

There were approximately ~21% with DEM party leaning and only ~2% with REP party leaning.

Method:

VA does not have registration by party, so there is no direct method for knowing a registrant’s party affiliation.

However, for the subset of voters that have voted in party primaries, we can calculate a “leaning” score in a couple of different ways:

Leaning_v1: This method uses the weighted average of the elections each voter has participated in, with Democratic primaries weighted as +1, and Republican primaries weighted as -1, and all other elections weighted as 0.

Leaning_v2: This method computes the voter leaning as the difference in the ratios of how often a voter participates in either a Dem or Rep primary, with a positive result meaning higher Dem, and a negative result meaning a higher Rep score.

Using the results that we presented previously, we have computed the Leaning scores for each identified non-citizen voter and computed the percentage of the non-citizen voters that fall into each category of Republican, Democrat or Unknown.

Both methods of computing the leaning give slightly different, but consistent, results.

See chart below: