Research companies have existed since the 1930s – in fact, George Gallup founded the first one, the American Institute of Public Opinion. This is the same pollster who has since stated that “the Institute is in a good position to know how ill-informed, how prejudiced, how stupid are some voters.” But this isn’t how polls are used. They don’t serve to remind us how unformed the general population’s views are on many topics, they are used as political ammunition. We hold polls up as shining examples of “what the public thinks”, when in reality, we often don’t think at all.
The US
isn’t the only country whose politicians rely on polls to gauge public opinion.
France is one of the biggest consumers of polls in the world – during the 2017
French presidential campaign, spanning just one month, 560 polls were
published.
So what? So
we’re obsessed with polls, who cares? The problem is that polls present
themselves as tools of direct democracy, based on the assumptions that voters
are rational and that the polling method is accurate. The truth of the matter
is that the math behind polling is highly suspect, and that, in the words of
Pierre Bourdieu, “Public opinion doesn’t exist.”
There’s
been a lot of research done on the so-called sociology of opinion. Patrick
Champagne argues that polls attempt to quantify opinions provoked by the
pollsters – that is to say, these are not spontaneous opinions, but rather
responses to targeted questions. These opinions are equally not divided by
their intensity or by how well-researched and thought-through they are. A
pollster could get me on the phone and ask me what I think about the effects of
quantitative easing, and group my poorly-thought out response with that of an
economist. So not only do polls assume that you even have an opinion on
something, but they assume that every opinion is equal in its intensity. My
favourite example of this is an experiment conducted by Gill in 1947, whereby
70% of surveyed people expressed an opinion on the (invented) Metallic Metals
Act.
And that’s just the sociology. Once we look at the way these ridiculous results are actually transformed into comprehensible statistics, serious questions can be raised. There are two main ways pollsters attempt to form a representative sample of the population: they can either dial telephone numbers at random, or they can follow a system of quotas based on four criteria (gender, age, profession and location – who knew you could define a population just based on four attributes?!).
And so this
leads us to 2012, when Gallup
Inc. (formerly the American Institute of Public Opinion) incorrectly predicted
that Mitt Romney would win the 2012 US presidential election. Following this
minor disaster, Gallup spent six months reviewing its methodology, and
established that it had insufficiently contacted the Eastern and Pacific time zones, overestimated the white vote, and, through
its reliance on listed landline phones it had overrepresented the older
demographic. But in
this day and age, who has a landline? And out of those who do, who’s going to
agree to answer a survey about current affairs? The answer is the same everywhere:
the middle-aged and middle-class. Once on the phone, a pollster will ask you a
series of carefully worded questions, with pre-formed responses. And if you
dare to say you don’t have an opinion? This is simply discounted. Once they’ve
got their answers, pollsters will look at their four criteria and start
correcting the data. They’ll shrink down the percentage of over-represented
groups and multiply under-represented groups, stretching and distorting the
data. What often gets underestimated, however, is the percentage of abstainers,
and the percentage of those voting to the right (especially the ALT-right), partly because of the so-called "social desirability bias".
It’s this kind of mistake that led pollster Frank Lutz to tweet in November 2016:
In case I wasn't clear enough from my previous tweets:
— Frank Luntz (@FrankLuntz) November 8, 2016
Hillary Clinton will be the next President of the United States. #ElectionNight
The problem with this
kind of mishap is that opinion polls are used by politicians as a gauge of
public opinion. The themes that dominate electoral campaigns do so because
these are the issues that pollsters claim the public cares the most about.
Politicians who claim to be representing public opinion are given false
legitimacy by polls, a mandate to carry out policies which poll well. There are
even laws that have been made as a result of polling.
The other big issue
with the inaccuracy of polls is that they are first and foremost a product
which can be sold to newspapers. In France, research companies’ biggest
customers are newspapers. Results of surveys are then further distorted by
journalists: “think” becomes “want” or “demand”, and “62% of surveyed people”
becomes “62% of French people”.
Finally, the last problem with this kind of polling is that voters are influenced by how they think others will vote. There are two main effects, the bandwagon effect and the underdog effect. Particularly in countries with a First-Past-The-Post electoral system, where tactical voting is often necessary, voters will look to polls to see which candidate they should vote for in order to keep their least-liked candidate out of power.
As Loïc Blondiaux expertly sums up, “A voter should make their choice based on their own conscience, not what their neighbour thinks!”