As much as I love and respect my late maternal uncle, a highly intelligent, articulate diocesan priest who had earned his licentiate (like a Master's degree in theology) from a prestigious seminary in Montreal (he had to write his exams in Latin (like my Mom and Dad, bilingual in French and English)), he, very conservative politically, was a total skeptic on the issue of election polling: "After all, I've never participated in one. How can they predict how I would vote?" Well, I had taken several applied math courses in probability, statistics, sampling, and research methodology; I tried, to no avail, to get him to understand how unlikely it is for sample results to radically depart from general population ones, never mind the proven track record of polling and related election results. Maybe in theology he had unquestioned expertise, but his understanding of statistics was uninformed.
Let's even start with the right-wing's favorite case in point: Trump's unlikely win in 2016. It really wasn't as unlikely as people might believe; if you go to RCP, you'll see Clinton had nearly a 3-point lead on Trump going into the election, and she won a plurality of popular votes by about 2 points. This was well within sampling error. Trump's victory was due to a highly improbable run of photo finish victories in WI, MI and PA and probably due to a number of favorable factors: including a change year election, Trump's lack of a track record in politics, a highly unpopular Clinton, and a disunited Democratic base after a divisive battle for the nomination. I'm not yet forecasting a Biden victory, but Trump's approval ratings have usually trailed his 2016 share of the popular vote, which is never good for an incumbent and his handling of the COVID-19 crisis has been and remains unpopular. He has trailed all 3 normally Dem states responsible for his 2016 victory, isn't competitive in any Clinton state (they are targeting MN, but that's unlikely to flip), while Biden is very competitive in a large number of Trump states, including AZ, TX, NC, FL, IA, and OH. (If Biden flips TX, I'm sure Trump is done; I think Trump is generally favored to win there but by a small margin, within upset range.) Still, Biden's lead is within the margin of error of battleground states, and Trump has 3 debates over the coming weeks to try to win over voters.
Ironically, what led to this rant was a relative by marriage's recent Facebook post basically trying to link a certain Pennsylvania kerfuffle mail ballot controversy to Trump's self-serving, manipulative attacks on mail ballot elections. Now I've published on this issue recently and made it clear: (1) I generally disapprove of the irrational fear-mongering, especially by Democrats, that in-person polling is a COVID-19 death trap; (2) I am worried about the less robust internal controls of vote-by-mail (how do you validate the target ballot was completed by the legitimate voter, what about the chain of custody of ballots, etc.) and whether validation of completed ballots (say, signature checking) can scale up beyond the usual volume for absentee ballots. I'm concerned about ballot harvesting from distribution to poorly maintained, invalid poll mailing lists.
My brother-in-law (who isn't outwardly Trumpkin but makes use of anecdotal evidence in thinly-disguised support of Trump's voter fraud nonsense) pointed out a recent Pennsylvania issue where voters weren't following the prescribed two-envelope directions and hence their ballots were being disqualified. (I don't know if we are talking Democrat voters, but it's highly probable most are.) From a partisan standpoint, I would think Trumpkins are happy that ballots are being validated. My estranged female Trumpkin cousin remarked the mail-in voting was making the election a "shitshow".
Just a side note on the "two envelope" issue. I have a research interest in human factors/ergonomics and have published on the concept of usability design, roughly designing for human error. It's actually incredible to me (especially after notorious incidents like the 2000 Florida butterfly ballot debacle) that public officials would roll out novel ballot programs without usability testing which would have provided early evidence of issues. Just another example of government-run incompetence and chaos. But there are other preventable election validity problems, like crosschecking for duplicate voting registrations (consider out-of-state college students, who may also be registered in their own states).
Let's be clear; voting fraud can and does occur, and this election, especially when trying large-scale novel voting changes, will be no different. But there is little evidence that fraud materially affects election results, if and when there are anecdotal incidents detected. Most of the fact-checking literature I've seen suggests similar conclusions. We have 50 different states running elections plus DC, all with varying processes. Trump wants to force in-person voting which he feels plays to his advantage. But there are practical issues; for example, I pointed out in a recent post that Maryland is having trouble recruiting election day judges (they intend to provide in-person voting as an option). This also poses an internal controls issue. Trump is asserting unsupported, unspecified problems with election processes that he has no conceptual understanding of. This is little more than same-old, same-old FUD (fear, uncertainty, doubt). It's immoral and disgusting behavior, unworthy of any sitting President.
