Exit polls vs machine counts or, how you might rig a primary

These kinds of large disparities between the exit polls and the machine counts might be dismissed as an anomaly but so many cases in so many states should probably have raised alarm bells.


After the debacle in Iowa on February 3rd that resulted in the winner of the popular vote in that state’s caucuses, Bernie Sanders, receiving fewer delegates than former South Bend, Indiana mayor, Pete Buttigieg, many on the progressive left questioned the fairness of what had taken place. It didn’t help that an app designed to make the process less complicated failed spectacularly.

After two more wins for Sanders, the fourth state to vote, South Carolina, was called very early, with exit polls showing that the endorsement of Rep. Jim Clyburn had played a major role in Joe Biden’s decisive win. This led to what seemed like a coordinated effort to push the moderate vote to Biden, with all but two of the remaining candidates dropping out of the race and seemingly coordinating their endorsements of the former vice president, establishing the narrative of his inevitability on March 3rd, Super Tuesday.

This made for a good story but is also one-sided and far from the whole truth.

Even in just following what was happening on the ground in two major states, California and Texas there were serious problems, with long lineups and some voters waiting up to six hours to finally cast their ballots, usually in areas with large communities of color. While primaries and caucuses are organized at the state level, this should have caused more concern to mainstream journalists and party officials than it did. After all, these were Democratic primaries and we are always told that these communities, often victims of Republican gerrymandering efforts, are the bedrock of the party.

Young voters, a crucial demographic for the Sanders campaign, were among those who faced real obstacles in exercising their right to vote, with the U.K. Guardian reporting hours long lines at the University of Texas in Austin, where one first time voter they spoke to, Fernando Miranda, waited an hour and a half earlier in the day before having to go to class then returning after 7 PM, at which time he waited for two more hours and was finally turned away without being able to cast his vote.

As the 19-year-old told the paper, “This was going to be my first time ever voting in a primary. I really wish I could have gotten it in, it’s my civil duty as a citizen to vote and it’s a bit disappointing that it didn’t work out.”

Further, there is still no final result in California and might not be one until April 10th,, hurting any momentum that might have been gained by Sanders after winning the country’s most populous state.

All of this is troubling enough but, according to Theodore de Macedo Soares, who runs the web-site TDMS Research, which appears to be the only place to find analysis of the differences between exit poll data and the unobserved machine votes that give the final results, has shown some alarming inconsistencies in several states crucial to the Vermont senator, including Texas, California and Massachusetts.

Exit polls, which allow the press to see the likely result of all but the closest contests, take place after individual voters cast their ballots and are asked as they leave the polling place about how they voted and usually their opinions on some of the issues. Throughout these Democratic primaries these polls are being done by Edison Research.

Interestingly, only in South Carolina were exit polls, which are usually remarkably accurate in predicting the final results, the focus of media attention due to the correlation with Rep. Clyburn’s endorsement and the African American vote, especially among this community’s older voters.

Early exit polls, which were said to have a 3.9% margin of error, seemed to favor Senator Sanders in many of the Super Tuesday states, with reporting showing that the issue sets he has put forward, from Medicare for All to free college and trade school were winning issues with voters in most on Super Tuesday, even if they didn’t seem to make a difference in terms of the vote, especially in two of them, Tennessee and North Carolina.

And this is where things in several important states that former Vice President Biden won, most importantly in California, Massachusetts and Texas become strange, as the initial exit polls and unobservable machine counts diverge by margins well outside the supposed margin of error.

The machine counts in these three states apportioned all important delegates in greater numbers to Biden than predicted by the exit polls. As Soares writes in his analysis of the California primary, “The current (3/9/2020) apportionment of California delegates, in accordance with the computer counts, stands at 185 for Sanders and 143 for Biden. (These numbers will change in the following days). The estimate derived from the exit polls calculates to 207 delegates for Sanders and 122 for Biden.”

In Texas, despite exit polls showing a tie or close win for the Vermont senator, Sanders wound up losing by 4.5%. As Soares notes, “His (Sanders) projected vote proportion fell 4% in the vote counts—an 12% reduction of his exit poll share.”

In what may have have been the biggest surprise, Sanders’ loss in Massachusetts, where he seemed likely to win, Biden’s machine count result was 16.2% higher or just over 65,000 individual votes than predicted by the exit polls.

Even in the Senator Sander’s home state, Vermont Biden received 26.1% more of the vote in the machine count than the exit polls, which seemed to show him within the margin of error of not reaching the 15% threshold to receive delegates.

One instance of these kinds of large disparities between the exit polls and the machine counts might be dismissed as an anomaly but so many cases in so many states should probably have raised alarm bells. If it had played out in favor of Senator Sanders it most likely would have been the subject of agonized opeds about the possibility of unnamed ‘Russians’ hacking the primaries.

Also, as I mentioned before, the machine counts are not observable, something that would be criticized were it to take place in another country like Venezuela where such practices would likely provoke calls for a judicial or military coup. This is why here in Canada and in much of Europe, ballots are still counted by hand out of the caution that machines can be gamed and, in some cases, hacked.

While Sanders still has a narrow path to the nomination and with a debate coming up with Biden on Sunday that could still be a game changer, it should embarrass Democratic Party officials that, once again, the progressive left have ample reason to believe that the collective thumb of the corporatist establishment has been placed on the scale to ensure that Senator Sanders is stopped and their big money donor sponsored gravy train keeps on rolling, regardless of whether their nominee actually wins the general election in November.


If you liked this article, please donate $5 to keep NationofChange online through November.