At the time this paper was written Professor
Paul Fox was Principal, Erindale Campus, University of Toronto. This is a
revised version of an address to a joint conference of the Canadian Public
Relations Society and the Couchiching Institute of Public Affairs held at Hart
House, University of Toronto, February 12, 1985.
In any answer to this question, it is
obvious that polling companies and pollsters are beneficiaries of public
opinion polls. They seem to make a pretty good living – at least the best known
The second element to gain is the public.
People learn something from polls which often provide interesting and useful
information. Anyone in the knowledge information business ought to be in favour
of maximizing knowledge, and therefore in favour of polls, at least those that
one could call honest polls.
Third, the media gain because polls make
good news stories. There are facts and figures in them that give a hard edge to
news and are easily understood by the public. This is undoubtedly the basic reason
why the media seize on them. It is easy to make a good story out of polls,
dramatizing, for instance, the way in which the movement of the poll affects a
party's or an individual's fortune.
Fourth, the manipulators gain: politicians,
parties, advertisers, lobbyists and special interest groups – all those who
want to ascribe to themselves the authenticity that comes from quoting
apparently indisputable facts and figures.
This is a point one ought to consider. Why
do the results of public opinion polls seems to have so much impact? The fact
is they quote specifies, and this is a specifics-oriented culture, whether it
has to do with your income or your age, or frequency of sex, or whatever. It
does not matter much whether the poll results are true or false. They still can
be used to advantage by the manipulators.
Now who loses? Obviously, the public loses
when they are misled. This may occur in several ways. First, by outright error.
Although the public seems to have considerable confidence in polls and appears
to believe what the polls say, the polls are wrong on occasion and some of the
occasions are rather important.
The polls were wrong in the 1984 Australian
federal election in not forecasting the swing away from Labour which diminished
Prime Minister Hawke's party from the majority of 40 seats they were projected
to win to a majority of 16. It was even worse in 1980 when almost every
pollster predicted a Labour victory and yet Malcolm Fraser's Liberal National
coalition won a very handsome victory.
The pollsters were wrong again in the recent
election in Denmark, and of course they were far out in their forecasts of the
election outcomes in Great Britain in 1970 and 1974 and in Gallup's forecasts
for Canada in 1957 and 1980. In the 1984 Canadian federal election, the. polls
all missed predicting John Turner's win in Quadra and the extent of Brian
Mulroney's breakthrough in Quebec. These were honest errors, but they were
errors nonetheless and should not be forgotten.
The second way in which the public is misled
is by deliberate misrepresentation or connivance. A special interest group
wants a certain result and goes out and hires a pollster or makes use of an
honest pollster's poll in a way he did not intend it. A group may even conduct
its own poll which is very unscientific, but which has the great virtue of
proving the results it wants.
There are many examples of that practice. In
the 1984 Conservative leadership race in Ontario every candidate except Larry
Grossman issued poll results which boosted his cause. And those poll forecasts,
when examined in the light of what happened, did not bear much relation to the
ultimate reality. That is par for the course. Candidates in a leadership
campaign or an election are wildly optimistic and they frequently use polls to
their own advantage to confirm their hopes for success.
But it is not only parties and politicians
who pervert the polls. Special interest groups such as the pro and anti
abortion forces have used polls to support their own respective interests.'
There was another interesting illustration
in Toronto when George Cohon, of Macdonald's Corporation, a man who is also
chairman of the Canadian National Exhibition Board, had decided to counter the
Macaulay committee's recommendation of Downsview for the domed stadium. It was
decided to pay Martin Goldfarb $5,000 to conduct a poll on where the site ought
to be. 2
Not surprisingly the survey found that about
twice as many respondents wanted the dome at the CNE as at Downsview: 46% for
the CNE, 28% at Downsview, and 16% for Woodbine. It is rather amusing that the
irrepressible Mayor of North York, Mel Lastman, got in the last word on the
subject when he put such tendentious polls in their place with the remark,
"statistics like that are like a girl in a bikini. They are interesting in
what they reveal, but they are vital in what they conceal." Perverted
polls are like Tammany Hall at its peak when it was said that Tammany Hall was
giving corruption a bad name!
The third way in which the public is misled
is the pollsters' practice of allocating the undecided respondents in the same
proportion as the decided. This has not been discussed extensively, vet we are
all aware, of course, that in most polls a very large percentage of respondents
say they do not know, or they are undecided, or they will not say. It is not
unusual that 20, 30 or even 40% of a sample give that answer. Yet Gallup, in
particular, and most other pollsters, round up the total to 100%. They may
mention that a certain proportion was undecided but when they give the figures
of how the vote breaks down, say in a pre-election federal contest, they give
figures for the Liberals, Conservatives, New Democratic party, and others which
are rounded up to 100%. The way in which they round it up is to allocate the
undecided in the same proportion as those who made their choice known despite
the fact that it is highly questionable to give the same degree of certainty to
dubious votes as to decided votes.
This seems to be a real weakness in public
opinion polls, particularly in an election. Yet very little has been said on
this subject. The danger in this practice was demonstrated in the Carleton
University School of Journalism's poll of delegates to the Conservative
Convention in Winnipeg in 1983 which led to Joe Clark calling a leadership
convention. The Carleton poll misallocated the "undecideds" and
forecast that Clark would get 76.7 per cent of the votes when in fact he got
about ten percentage points less.
Who else loses? Certainly justice loses when
we have a plethora of private polls whose results are not revealed. The results
of private polls are usually kept secret, in contrast to public polls which are
published. The well-heeled interests in society can afford to do a lot of
private polling and use the secret results to their own advantage.
We know, for instance, that governments,
parties, corporations, and interest groups do a lot of private polling from
which they alone benefit. The Ontario Conservative government was accused of
spending $500,000 a year on private polls whose results they did not reveal.3
The federal Liberal government employed Martin Goldfarb for years to do
continual private polling for them. Alan Gregg said that he had polled 500 to
1,000 Canadians every day during the September 1984 election campaign on behalf
of the federal Conservative party to whom the results were given.4 We know that
Ronald Reagan and the Republicans were able to use to their advantage the very
sophisticated polling technique of Richard Wirthlin and Richard Beal in the
1980 U.S. election.
Now such extensive private polling obviously
confers a great advantage on those who are wealthy enough to command the
resources. This practice is all the more questionable in a democracy when a
government uses public funds to acquire knowledge which is used for private
The argument that we should ban public
opinion polls during an election really misses the mark. What we should be
talking about is what to do with the growing new secret private polls which
confer such an advantage on those with the resources to conduct them.
Finally, it seems that truth loses when we
over-emphasize polls. We can all recall how in the 1984 United States
presidential election Gary Hart was manufactured by his private pollster who told
him how to dress, how to speak, how to present his personality, and what
policies to put before the public. This seems to be a perversion of the
Some observers believe that candidates will
be strong enough to resist these pressures but it is a cause for concern that
many candidates are not. They want to win and they are prepared to modify,
twist, pervert their presentations, themselves, their psyches, their personae,
and their policies to suit the interest that is revealed to them by their
Such candidates also avoid contentious big
issues such as mounting deficits, or the problems of abortion and capital
punishment, because the pollsters have told them the public will become so
divided on these issues that if they talk about them, they will alienate some
faction and lose. Obviously, the manufacture of candidates or the ignoring of
big issues is not healthy for democracy.
Finally, are polls here to stay? The answer
is definitely yes. There is little we can do to abolish them, and in any case
it may not be wise to abolish them. It is amusing to note that even the Roman
Catholic bishops are now into polling. They conducted a poll recently to find
out whether or not the Pope is popular, which seems to be an ultimate of some
kind. Incidentally, it is not surprising that they found that 88% of the poll
thought the Pope was well liked. Undoubtedly, a good journalist or observer of
public events could have told them that and saved them some money. However, if
the Catholic bishops are into polling, it is likely we are all becoming
Honest public polling is useful and can be
informative. However, perverted and private polls are something else and we
should worry about them.
1 The Globe and Mail, January 26,
2 Ibid., February 21, 1984.
3 The Toronto Star, February 25,
4 The Globe and Mail, October 12,