CVR Analysis Mistakes

Citizens Oversight (2022-11-12) Ray Lutz

This Page: https://copswiki.org/Common/M2019
Media Link: https://youtu.be/QsM8OfePs7c%20type%3d%22html%22
More Info: Election Integrity

I frequently see people trying to use the Cast Vote Records (CVR) to analyze the election results, particularly with regard to "The Rule of Large Numbers" i.e. that the results should converge right away to the average, if chosen randomly. Although this rule is correct, it does not apply when not actually using random sampling, but rather sampling "arbitrarily" using some ordering that seems like it might be random, like the order of the precincts or size of precinct, or the order of the CVR records. That might seem random enough, but it is actually not random. For example, using size of precincts is commonly used, from small to large. But smaller precincts might be more rural, larger more urban, and then the split of the vote may be quite different. It may be hard to know or comprehend but if it is not truly random, you can't use this as a basis for analysis, period.

Also, CVR records are not in the order of time, because the record numbers are shuffled, and sometimes they are resorted to precincts, for example. So in the video, we show an example where they actually order the cvr records by reverse number, which makes no sense at all, while also saying "later" and "earlier" as if they relate to time. Also, there are typically four tranches, early voting and early processed VBM ballots, Election Day voting, Later arriving VBM ballots, and finally provisional ballots. It depends on the rules in the state and the equipment used. These tranches tend to be quite different in the margin of victory because early voters may tend to be more conservative (plan ahead) types while later are last-minute voters that are very different and maybe more liberal. But this is not a sure thing anymore because both parties are pushing for early voting. Yet, in the 2020 election, it was leveraged by the Trump campaign NOT to use vote-by-mail voting, which might take longer to process those last ballots, and thus they pushed the existence of what some were calling the "red mirage".

The net result is that this type of CVR analysis is invalid and will provide false notions of hacking that don't actually exist, so beware!

Comments