GUEST COLUMN, Dick Morris
As the presidential election nears, the political establishment is doing its best to hide the increasingly obvious fact that Donald Trump is winning . . . by a lot.
There are three surveys that honestly reflect the reality that Trump is winning.
They are the Harvard Harris poll administered by my former colleague Mark Penn, The New York Times Sienna survey, and the Rasmussen Poll, which was accurate in 2020.
These surveys all are weighted based on actual exit polls from 2020 and all show Trump winning big!
Harvard-Harris shows Trump beating Biden and Kennedy by 46-39-12.
Rasmussen has Trump ahead of Biden and Kennedy by 48-36-8. And The New York Times/Sienna poll has Trump ahead by four among likely voters but only by one among registered voters.
For the rest — mostly showing a very close race between Biden and Trump — there are real doubts.
So let’s take a journey into the weeds and dissect the surveys that show Biden ahead or rate the race as basically tied.
Economist/YouGov on May 8, published findings showing the two-way race tied at 43-43.
The problem with that survey is that among the 1,596 registered voters they surveyed, there were only 444 Republicans or 28 percent.
But in 2020 exit polls indicated that 36 percent of those who voted were, indeed, Republicans. The survey apparently packed its sample with Democrats.
While Democrats and Republicans voted in about equal numbers in 2020, the Economist/YouGov survey sample consisted of 28 percent Republicans and 38 percent Democrats, overstating the Democratic vote by eight points.
If your sample is eight points too Democratic to begin with, your poll’s result will be eight points too Democratic as well.
So the poll is in this case, understates the Trump vote by approximately this margin.
The Morning Consult survey of May 14th suffers from a different defect that undermines its finding that Trump leads Biden by only 44-43.
The Morning Consult poll only surveyed registered voters not likely voters. In 2020 66 percent of voting-eligible people and 71 percent of registered voters actually turned out to cast ballots — a record.
But while 155 million registered and eligible voters did actually vote, 63 million did not.
The Morning Consult Poll reports that among those who were eligible to vote, Trump and Biden are neck-and-neck. But what about the 29 percent of their sample who were nonvoters?
How, by studying nonvoters, are we to draw conclusions about what actual voters will do on Nov. 5, 2024?
The TIPP Survey on May 6 also had the presidential race knotted at 42-40 but with Biden ahead.
Unfortunately, the TIPP poll suffers from the same defect as the Morning Consult survey in interviewing registered not only likely voters so, again, we are left to guess what actual voters think.
A fourth survey, this one by ABC News/Ipsos on May 5, also showed a tight race with Biden in the lead 46-45. In a novel effort to exclude voters who would, ultimately, abstain, the firm asked people to rate their chance of voting on a one to one hundred scale.
But then, having taken great pains to include only actual voters, the polltakers then let everybody into their sample so that 21 percent of the interviews were with people who admitted that there was less than a 50/50 chance that they would actually vote.
Indeed, they proceeded to interview even the 14 percent of respondents who flatly said they would not vote and then they mixed all the respondents together so we are left without a clue as to who those who will vote plan to support.
The fifth survey, taken by Marist College on April 22-25, also showed Biden ahead by 48-50.
But, unlike the other firms it drew its sample from a list of telephone numbers, not of voters. Marist found 1,109 registered voters who it interviewed out of 1,199 people with whom they spoke.
But unfortunately, they were too Democratic.
While the 2020 exit polls showed a nearly even partisan divide with 37 percent Democrat and 36 percent Republican, the Marist sample contained 36 percent Democrat and only 30 percent Republican.
If Marist had adjusted their sample to mirror reality, Trump would have been ahead.
Just this week, we are seeing two new slanted surveys.
As the election approaches and Biden’s ratings remain dismal, the questionable polls are proliferating.
Are pollsters desperate to make Biden look competitive enough to run?
Yahoo posted a survey on May 10-14 showing a tie race between Trump and Biden.
But, once again, it is based on a biased sample with only 27 percent Republicans (as noted, it should be 36 percent Republican according to the 2020 exit polls).
And Economist/YouGov tried again, releasing a survey of registered — not likely — voters this week whose sample was 39-29 Democrat, a bias of 10 points from the 2020 exit poll standard.
It’s finding — a tie race — it can be argued, is worthless.
And now for the key question.
Are what appear to be distortions in these five surveys deliberate, or just mistakes?
In analyses of the 2020 polling, a nonpartisan panel of experts from the Pew Research Center found that “it is clear that Trump’s strength was not fully accounted for in many, if not most, polls.”
Understatement of the year.
And, whether deliberate or not, it looks like the bias continues.
