@jennycohn1 @Gwenwatel1 If a variety of machines were used, it is possible to estimate how many flipped votes were missed and actually cast by mistake. If we know what fraction of o/w straight-ticket Dem votes nonetheless chose Cruz on the other machines, compare that with the % on the HART e-slates.
@jennycohn1 @Gwenwatel1 Example: suppose 97K ballots cast through other machines show straight-ticket votes for Dems, and 3K were straight-ticket Dem except for votes for Cruz. If the HART e-slates show a different ratio, say 92K : 8K, then while other factors cannot be entirely dismissed, there...
@jennycohn1 @Gwenwatel1 ...would be strong reason to suspect that as many as 5% of that specific subset of Cruz votes (those where all down-ballot votes went to Dems- not 5% of all Cruz votes) had in fact been mistaken votes intended for Beto.
@jennycohn1 @Gwenwatel1 Careful quantitative analysis of this should be sufficiently persuasive for a court to consider becoming involved- or it might reveal that the problem was less significant than we had feared. How can we make this analysis happen?
@jennycohn1 @Gwenwatel1 If no other machines except HART e-slates had been in use, then these numbers would remain largely unknowable. The fact that other machines were also used, however, means we essentially have a scientific control group- a basis for comparison which permits quantitation.
@jennycohn1 @Gwenwatel1 Caveat: the comparisons between types of machines must be made only between districts with similar voting profiles: a different area might have a larger or smaller percentage of "all Dem + Cruz" than would be representative of districts with the HART e-slates at issue.
@jennycohn1 @Gwenwatel1 @threadreaderapp please unroll
