Read on Twitter
The "expert" witness for the Kraken plaintiffs analyzes a survey of people who have a record of casting an absentee ballot. It is a survey! Must be 100% accurate, right? But, let's dive into it anyway https://www.courtlistener.com/recap/gov.uscourts.gand.284055/gov.uscourts.gand.284055.1.1.pdf
The key piece of evidence of vote fraud are the "troublesome" ballots from people who are recorded as returning an absentee ballot from people who supposedly never requested a ballot. Must be evidence of massive vote fraud, right?
The most obvious point of rebuttal is found in this table (this one is for Georgia). See anything that might explain why people have a record of voting a mail ballot but supposedly never returned them?
Let me help you a little further
A person who returned a mail ballot in-person would have truthfully responded to this question that they did not mail back their ballot. That is not evidence of vote fraud
That strikes down half of the argument. The response might be there is much more vote fraud! If we include the people who returned ballots in-person, there's even more people who returned mail ballots *without ever requesting them*. So, I now have to address the request part
First is that you have to assume, as the expert does, that
"I assume survery respondents are representative and the data is accurate."
Yes, he wrote "survery" in true Kracken form
"I assume survery respondents are representative and the data is accurate."
Yes, he wrote "survery" in true Kracken form
One problem is that it appears in some cases the actual voters were not contacted, but a "member" (I'm assuming household member) answered for the voter. All of these responses should be removed from the analysis since they are hearsay. It is unknown how this affects the analysis
Another problem is it is well-known in survey methodology that people make errors in responding to polls. Any polls with small percentages should be taken with a huge grain of salt. That is the case here. E.g., the "troublesome" ballots account for 1.3% of all GA mail ballots
The expert does a sleight of hand to cover up the fact that he's analyzing a very small number of respondents. He reports 95% confidence intervals at the first stage to get to his troublesome ballots, but doesn't actually construct a confidence interval for the second stage
Wrt to memory errors, I'd be very cautious about Arizona, Pennsylvania, and Wisconsin. AZ and WI have permanent absentee ballot lists. A PA primary ballot request carried forward to the general election. Voters might not remember/understand their initial request
