I have long been suspicious of polls of any kind, but especially of political polls. It’s not just that they can be mistaken, but that it seems obvious that in the age of computers and AI, polls can be manipulated to produce desired outcomes. Look at the disparity in Presidential favorability ratings in the chart on the left.
My suspicions were confirmed this week by an op-ed in the Wall Street Journal co-written by Harry Crane, a professor of statistics at Rutgers University, and Koleman Strumpf, a professor of economics at Wake Forest University. They discussed “two examples of fraudulent polls from the current election cycle.”
The first fabricated poll “prominently overstated the chances of a far-left candidate in the Democratic primary for Wisconsin Governor.” The second actually had an impact on the Michigan Democratic Senatorial primary. The fake poll numbers led one candidate to drop out of the race after it seemed hopeless to continue. The poll “led donations to dry up and left staffers demoralized.” As it turned out, Abdul El-Sayed, the Democratic Socialist candidate, barely won the Democratic primary after leading in some polls by 20 points.
These two examples of fraudulent polling would seem to be just the tip of the iceberg. In the run up to the 2016 and 2020 Presidential races, polling numbers for Donald Trump were five points lower than the actual result. That’s well out of the margin of error. The 2020 polls were called the worst in 40 years. Polling results improved in 2024 but some still failed to call seven swing states for Trump.
In each case the polling error would appear to have been a form of voter suppression. Democrat leaning polls were a way to discourage Trump supporters from voting. Democrats often complain about what they consider to be Republican voter suppression tactics like voter ID, but they and their pollster and media allies are masters of voter suppression.
The professors suggest that popular prediction betting sites might be better indicators of future election results, but I wonder if these sites could also be manipulated. After all, if a campaign has a million to spend on useless TV ads, maybe a bet on Kalshi might be more effective in swaying opinion. The last time I looked almost $14 Million had been bet on the Michigan Senatorial contest. This astounding figure makes you wonder if the affordability crisis is also a product of fake polls.
###



