Why is how badly the polls performed in the recent presidential election a business story? While
many Americans think of polling as being part of the political or media process and the results
sacrosanct, it is for all practical purposes a business. I should know because I was trained as a
researcher and have conducted many surveys over my marketing career. In most cases, I was
hired to either prove or disprove a hypothesis. What that means is that I will go to great lengths
to craft a survey (the format, the questions, and the target group) to fit the conclusions that my
clients are paying me for.
None of the news organizations do their own polling. They hire a polling or research firm and in
most cases the objective is to prove that their narrative is sound and even scientific. So, if Joe
Biden is ahead of Donald Trump by 10 points, it simply fits the narrative that has been so carefully
scripted over a long period of time that in Trump’s case began almost immediately after he
stepped off the escalator to announce he was running for the White House. They are out to prove
that Trump’s election was a fluke and that most Americans are anxious to rid themselves of this
unconventional bully president. At least that was the case for most of the liberal media.
In a presidential election, the polling organizations will be releasing as many as four polls a day,
perhaps as many as 400 in all by time the election takes place. And why not? Each and every poll
is paid for and the more polls, the more fees and more profits. It’s a great business because, most
Americans do not have a clue how these polls are conducted and for all practical purposes it is an
industry without accountability. In reality, I learned early on that the science of polling depends
on the validity of the sample. For example, if one wanted to poll a community, the starting point
would be to gather the demographics of that community and then take a representative sample
that in the end could prophesize what that entitize community is thinking.
Many people in the polling industry have tried to impress on me that getting the correct sample
nowadays is a challenge. Many more calls have to be made just to get a fairly representative
sample (i.e. voice mail, caller ID). Think of it this way: We know that approximately 140 million
Americans voted in the past election, but a good polling organization might speak to as little as
800 people to predict the outcome of a presidential election. If those 800 people would truly be a
“representative” sample, it could work, but what if the “random” sample is not really so random.
What if a news organization pinpoints the communities they want sampled because they have a
preconceived notion of what their responses might be, then the results will, of course, be skewed
and I suspect that it is exactly what happened in the 2020 polling when they ferreted out the
Biden voters.
I have always wondered how valid these surveys are when the people surveyed don’t actually tell
the truth, meaning they might tell the pollster that they will vote for Biden but then at the booth,
they vote for Trump. Perhaps there was someone standing nearby who would frown about a
Trump voter. Telephone polling has become even more complicated as computer generated lists
now have to include cell phones since for so many Americans cell phoned have become their
primary contact number. While polling might arguably be a science, it relies on several key
factors. The first is to make sure that there is a valid and representative random sample. For
those still doing manual telephone surveys, how do you make sure that the interviewer does not
in some way lead the subject to the conclusion he desires. For example, imagine a question like
this: “If Joe Biden could guarantee a solid economy, full employment, and peace throughout the
world, would you vote for him?” A mere answer of yes could put that person in the Biden column
although even after the telephone interview, he has no intention of voting for Biden.
Here is another pollster trick, called “weighting” that I read about: What happens if a pollster
finds that only 6% of his sample is African-American, compared to 12% of the American public.
When using weighting, in effect, counting African-American responses twice towards the overall
results. But for news organizations that are aware of these flaws in the polling process, there is no
worry because they have so many polls to chose from that they can always pull out any survey
they like into what is now called their “news cycle.”
As a business, polling is a good business. Organizations get paid for every survey, cover the
expenses of polling workers, and tack on many fees like for charts and reports that adds up to a
very substantial profit center. What is amazing is that while other businesses would have some
accountability if they fail, the polling industry just goes on with their lives. What consequences
did they suffer from totally missing the call on the Trump-Clinton race in 2016? Aside from a few
nasty editorials, and snide comments from news commentators, very little. Imagine that it now
happened twice in a row and they’re still in business. I think that every business would love to be
able to get such a free pass if they should happen to mess up. I know that I would.
Nowadays, many surveys are done on-line. Anyone can download Google’s Monkey to do a
survey. I know that in our community, a sizeable percentage do not use the Internet. In fact, 16%
of Americans don’t use the internet, requiring an additional layer of the weighting technique to
try and get close to a representative sample of the public. This is why it has often been difficult to
get an accurate random sample in the frum community. For these frum voters or consumers,
phone or personal interviews would be required. But pollsters often shy away from going into the
frum communities, sometimes encountering issues like the gender of the pollster. Media
organizations trust live caller polls over other methods largely because their cost and less
difficulty to obtain means that the groups doing the surveys have a bigger incentive to produce
high quality results and get things right.
I don’t know too many people that ever spoke to a pollster. According to Pew Research Center,
only 6% of the people it calls are willing to talk. Those numbers represent a significant problem
— and a huge opportunity. At the turn of the century, Pew’s response rate stood at 36%. So
again, here is an industry that most of its consumers don’t even want to talk to them; yet they
are out there predicting who the next leader of the free world will be.
Just how much of a business is polling. Over the past 3 years, the Marketing Research and
Public Opinion Polling industry in the U.S. has averaged an annual growth rate of 2.3% to reach
$20.6 billion. It might be a business that some of our young entrepreneurs should consider
especially if they can come up with creative models that are not as outdated as the current
models are. Most polling organizations will continue to do business as usual. Sure, they will
make some modifications like switching to cellphones but by and large the random sample
model will stand. Pew is looking to making changes by shortening interviews and offering
financial incentives to those who participate in its surveys. Some are moving to the on-line
models while others are using a mixed bag. As for me, I will continue to be skeptical of an
industry that is flawed and seems to be on the beck and call of biased news gathering agencies.