At the time this article was written Dr. Lorne Bozinoff was Vice-president of Gallup Canada. He has held senior positions with the Coopers & Lybrand Limited consulting group and Bell Canada, and taught courses in survey research at Pennsylvania State University. Bill Blaikie, MP, represented Winnipeg Transcona in the House of Commons. William Roberts, MLA, represented Edmonton-Centre in the Alberta Legislative Assembly. Eric Allaby, MLA, represented Charlotte-Fundy in the New Brunswick Legislative Assembly. James Wilson, MLA, represented Fredericton North in the New Brunswick Legislative Assembly. Stephen Rogers, MLA, was the Speaker of the British Columbia Legislative Assembly and represented the constituency of Vancouver South.

This round table took place at the 31st Commonwealth Parliamentary Association Regional Conference in Fredericton in July 1990

]

Lorne Bozinoff: I want to talk about four myths and realities of public opinion polling. Myth One is that polls have a direct influence on public opinion. This is probably the most misunderstood aspect of polling. Polls are a measure of public opinion. A thermometer does not create the weather, and a poll does not create public opinion. In my opinion, polls do not create a bandwagon effect unless it is clearly in the voters' interests to support the winning party. In some parts of the country, if you do not support the winning party, your road might not get paved. In that case, there may be a bandwagon effect, but I do not think that happens very often. The reality is that polls have an indirect influence on public opinion. It is certainly the case that polls focus upon the horse race aspects of elections but that is what elections are all about – they are horse races. There will be a winner and a loser; that is their purpose. The polls have simply made that more objective. We did not create that situation. An election is an adversarial contest, and the polls are simply there to measure it in an objective manner.

People have commented that polls take away from the discussion of issues. All that people want to talk about are the polls. To go back to the 1988 federal election, people told me that there were too many polls. They felt that that was all that people talked about, and that we needed more time to talk about free trade. In my opinion, we talked about free trade far too much. I cannot imagine anyone wanting to hear more about free trade than we heard in that election. I cannot believe that anything was crowded out by the reporting of polls.

I also feel that, to a certain extent, the Canadian public has heard enough from experts, pundits, journalists, editors and the like. They want to hear from the average citizen. There is a great need for that kind of information. There is also the feeling that polls can reduce the amount of media coverage provided to trailing candidates. The decision about whether or not a candidate is trailing will be made anyway; someone will make that judgment. In the past, reporters conducted what they called "man-on-the-street interviews" to get a sense of how a campaign is going. The important thing is, that's exactly what Gallup does, except when we do it we talk to 1000 individuals. When we do it we make sure we ask everyone exactly the same question. When we do it we make sure a random sample is selected.

The last point I want to make in terms of the impact that polls can have concerns the feeling that the media can ignore candidates because they are trailing in the polls. I do not think that happens either. If a candidate is trailing but has something important to say, I think the coverage will be there. I do not think polls will lead someone to think that a candidate is trailing, so he or she should not be covered.

Myth Two is related to Myth One, and it concerns the bandwagon effect. I will make two logical points to argue that there is no bandwagon effect. We do not have to do a study or conduct a survey to figure this out. If there is a consistent bandwagon effect, then two things will happen. First of all, final pre-election polls will always differ from the actual results on voting day. When the final pre-election poll is published, if it influences the campaign, then the numbers will vary from the actual election results. Gallup has 50 years of pre-election polls and, as everyone knows, they are deadly accurate.

Secondly, if there is a bandwagon effect, then the party that was leading at the beginning of the campaign would necessarily increase its lead. Tell John Turner that with regard to 1984: he was leading at the beginning of the campaign, and he can tell you that did not happen. There was no bandwagon effect. All the polls at the beginning of the 1984 campaign showed the Liberals to be ahead. As we all know, that was not the case by the end of the campaign. Unfortunately, the reality is that politicians sometimes seek to ban election polls. A federal process is underway right now – the Royal Commission on Electoral Reform and Party Financing — which is receiving briefs suggesting that polls should be banned. In fact, some media outlets in the country will not publish polls during the last 10 days of a campaign. They think that their viewers or readers are dumbbells and are swayed by polls. When they make this kind of judgment, I think they greatly underestimate the intelligence of their viewers or readers.

I understand that politicians do not like media polls. They love their own internal party polls, but we're talking about the banning of the publication of polls, not the banning of taking polls. If it were simply a ban on taking polls, I would have nothing to worry about, because I know no politician in the country would agree to such a ban. The point I want to make here is that politicians do not like these media polls because there is no opportunity to control them. In a certain sense, these polls are totally outside the influence of the media experts, consultants and image consultants that the parties tend to hire these days. The numbers that come out are very objective. Generally speaking, the pollster has no connection with a political party and cannot be challenged in terms of credibility. I think this makes some politicians uncomfortable.

We should think of a poll as an electronic town hall meeting. We have not brought everyone together physically in one room, but we have talked to these 1000 Canadians individually. They have all been asked to express their opinions on the very same issue. We have taken very great care to record that. In a sense, we are talking about a type of participatory democracy—an individual has been given the chance to express his or her views. This does not concern the chronic complainer or the person who writes to his MP every week about something. The average Canadian is asked about national issues. I can tell you that we get a great positive response and participation because people do want their views known. We know that when the polls are published, the readers like to read that information.

Myth Three is that the media understand polls. There are four comments I want to make with regard to that myth. First of all, the media are not trained in the interpretation of polls. The media do not understand things like sampling. I have tried to explain it many times, and I can tell you it's not understood. They do not understand the technical aspects of the polls, such as questionnaire design. In fact I am beginning to suspect that the media do not understand the public. The media spend so much time on Parliament Hill in Ottawa looking at national issues that I am not sure that they really understand what is on the average voter's mind. I look at the judgments that are made by many prominent members of the media, and I compare them to our own polling results. Journalists will say that the public will never support a certain thing, but I have got the numbers sitting on my desk which indicate that they do support it.

We have media conglomerates now who own polling organizations–a sort of unholy alliance. Can you imagine this sort of one-stop shopping? Your poll is taken by one department and analyzed and commented on by another department, all within the same organization. That is an inherent conflict of interest.

Another thing I should comment on with regard to the media's understanding of polling is that the media like to hire party pollsters. Gallup makes a very great effort to be non-partisan, refusing to work for political parties. I have found that when a media outlet hires a party pollster, that pollster's party tends to be the same party that the editors of that newspaper supports. It is quite a coincidence.

Of course the media love polls that they do not have complete access to. They love a leaked party poll. They would rather not know about the sample size or who was surveyed. Whenever a Gallup poll comes out, there is always one party that says that its poll shows something different. The media love this because they do not have access to it. The media ask to be shown the poll results, but the party states that it cannot do that. It says, "Take our word for it." Unfortunately, the media does.

Let's try to get the analysis, what's really going on; let's get the story behind the numbers. It seems to me, the weirder the pollster–the bigger the earrings or the fuzzier the hair the more the media seem to like it.

Myth Four is that the pollster is an important person. The Gallup philosophy has been that it's the poll that is the news and not the pollster. Gallup has not had a identifiable spokesperson in this country; that has been deliberate. George Gallup passed away several years ago and, having been an American, never had too much to do with Canada. The Gallup organization is not connected to George Gallup in any ownership sense. The idea of a poll is to take the responses of the average Canadian, not to take one individual's interpretation or my own interpretation, but take the numbers as they come in as representing the Canadian public. In that regard, I have to say that a pollster does not speak on behalf of millions of voters. All a pollster can do is tell you his polling results.

It is true that a pollster is an important person, but for a different reason. Pollsters do have access to the front page. Gallup produces two news releases a week, all year long. A number of those get very good coverage. The media would rather talk to a pollster than report seriously the results of a poll. They want to talk to the guru. We call this "personality cult polling."

Pollsters are important because they are responsible for selecting the questions that will be asked, and they do issue the survey results. For example, all during the Meech Lake process, we knew what the public was thinking. Gallup was doing monthly polls to find out who was in favour of Meech Lake and it was reported every single month and obtained fairly good coverage. I think pollsters have a fair degree of credibility with the Canadian public.

Finally–and this is something we really have to think about–public opinion is difficult for anyone to influence. I do not believe that a TV clip will do it. I do not believe a poll or a pollster will do it. The public should not be underestimated. It has become very weary of politicians and news media. Political parties and the media do not have a high degree of credibility with the Canadian public. Consider again the Meech Lake process. We had a policy which was supported by all three federal political parties and, at one point, all ten provincial governments. We had a policy that was never accepted in English Canada all during that time. Think of it. We went through a three-year process, and during a large part of it no one opposed the accord. I believe that the media was giving the accord fairly sympathetic coverage, but the public never bought it. Where did the opposition to the accord come from? I will leave you with that question.

Bill Blaikie MP: I do not think that using the thermometer and the weather as an analogy is fair, because the weather is not conscious and does not know what the thermometer is saying about itself. This is the fundamental distinction between the social sciences and the physical sciences. What is involved here is a kind of behaviour in which the action of the observer and what the observer reports about the observed affect the behaviour of the observed. It may not always have the same effect, and sometimes it may not even have a significant effect. However, I think it is logically absurd to claim that it has no effect. This is to deny the consciousness of human beings and the fact that when people are part of a situation, they are part of everything that has to be taken into consideration. That is the position that pollsters find themselves in. Frankly, I think it is ridiculous to make the simplistic claim that it does not affect things. What we are trying to determine here is not whether or not there is an effect but what the effect is and what it is in relation to all the other factors that have to be taken into account, such as the synergistic effect with TV that sometimes occurs or certain regional matters. I do not think you can analyze the 1988 federal election without explaining the differences between Quebec's attitude and that of the rest of the country to the free trade agreement. So that is at least one thing that has to be taken into account.

One of the things that MLAs and MPs have to take into account in asking themselves how the polls affect the political process is how polls have affected their role. You say that no politicians have complaints about internal polls. I would say that many politicians do have complaints about internal polls because the party brass–the party organizers, platform designers and so on–now depend more on the polls than they do on their caucuses for determining how the public thinks. That is why we had caucuses in the first place, so that the Prime Minister or the Leader of the Opposition could ask their party members what they believe the people are thinking and what they are hearing on the street. They don't do that now. They ask Angus Reid or Allan Gregg.

I think that the concern of people here today should be the way in which polling has diminished the roles of politicians. You said it was like having an electronic town hall meeting. The logical extension of that argument is that someday everyone will simply have some kind of computer hookup in the living room. When the Prime Minister wants to know what kind of decision to make, he will punch the question in, the question will appear on the screen, everybody will punch their answer back, the tally will be made and the government will make its decision. Why do we have to go to the trouble of having MLAs and MPs in the first place? I think this erodes the tension between the varying descriptions of the role of Members of Parliament or Members of the Legislature. The image of the Member of Parliament that it erodes the most is the traditional Tory one, if you like: the Burkean notion of a Member of Parliament as someone who owes his constituents more than his best judgment. If everything is reduced to polls, then you have the view of a Member of Parliament as simply a delegate of his constituents. If you are simply a delegate, if you are simply responding to polls, the whole question of right or wrong is removed from the political calculation that MPs have to make. It reinforces the view that members are simply there to do whatever their constituents want them to do in a demonstrated majority. Maybe some people here agree with that view, but I have been in situations where I disagreed with the majority of my constituents, as I am sure other have—for instance, with regard to capital punishment or a number of other issues. What the whole poll culture reinforces over time is the view that you should not be rendering judgments of your own and being accountable for them at election time.

On the nightly news, one broadcaster talks to another broadcaster who talks to another. Why do we go to all this trouble to get elected, so that we can watch other people talk about us instead of us talking to the people through the media? After all, we are the ones for whom the people voted, not the media representatives who get to talk to the people all the time.

That is my biggest reservation about polls. I do not expect that they will be banned or that their publication will be banned. I think that we need much more discussion about the effect that they are having on our political culture, both inside political parties and in their relationships to the media. You have the media taking their own polls.

We are developing an entire layer of political discourse on top of the legitimate layer of discourse, which is the parliamentary and the legislative. So far we have not figured out the full effects of that, and they are related to the media and the polls. I think there is an underlying feeling that people want to shake this. I do not know if it is a question of getting back or going forward to something other than politics being reduced to entertainment. One observer talks to another observer who talks to another observer, and very seldom do the real participants get anything more than an 30-second clip. Unfortunately, you cannot describe reality in a 30-second clip.

James Wilson, MLA: I would like to start from the premise that political polling is here to stay. Whether or not one likes its intervention in political campaigns, it is a legitimate activity. Does it influence voters' behaviour? It probably does, but no more than other factors such as the leader's credibility, an unpopular tax, scandals, or various interest groups that are mad at individual politicians. The list goes on and on. These are all very important factors.

The brutal reality is that in recent years polls have generally been quite correct in terms of the winners and losers in federal and provincial elections. The 1987 provincial election in New Brunswick was no exception. I am told that some of those who commissioned the polls outside of the political parties refused to believe the indications of the polls as they came in. In fact, I did not believe them either; in about the third week after the writ was issued, the polls had me at about 15% and my opponent, the incumbent, at 85%. After two weeks of knocking on doors and travelling around my riding, I knew for sure that that figure was incorrect. In fact, some of my campaign workers subsequently travelled with me on various occasions to see the impact of going door to door and to see how people reacted. They could not believe it. There was never another poll. If polls have any bearing on election campaigns, as far as the candidates are concerned, I do not think that one carried too much weight.

William Roberts, MLA: I would like to build a bit more on the sense of how we as legislators use this resource, this snapshot of what is going on out there, this kind of sociological data in a political way. I would like to pick up on what Mr. Bozinoff said in terms of his belief that the media do not understand how to interpret polls. It might be fair to say that politicians do not understand how to interpret polls or how to use them in certain ways. It seems to me that we need to have more strategic thinkers within our ranks who can bridge the gap between what this sociological data tells us and what our party tells us in terms of what we are about. We need more strategic thinking to help us position ourselves with respect to certain data or to find what is salient among that data to develop our best message or how to best put the spin, as they say, on our message. In Alberta, we New Democrats are very highly positioned on health care—all the data supports that. It is a very salient issue for Albertans that there be a good health care system. We need to find better ways to talk about health as an issue for us.

However, I do not know how we can get the kind of training, experience or the strategic ability we need to best use the data to advance the cause of those issues to which we as New Democrats, Liberals or Conservatives are most committed. It is something that we cannot reject; it is here to stay. We should not ignore it or think of it as an evil that we have to accept; it is something that we can use much more effectively than we have to date. I am all ears about how better to do that.

Eric Allaby MLA: We have been dealing with the power and attractions of polling. I think one of the attractions that certainly has to be considered is the fact that information can be conveyed in very concise packages. Television relies on being able to put a message across in 15 seconds. Polling lends itself to that kind of transfer of information. A graph can very concisely express an entire complex question in a very short order through a colourful graph on the television screen. In that sense it is a very attractive means of conveying information. It is also very powerful. This power is also in the hands of the pollster, and the spin of the question will suggest the spin of the answer. I recognize that polling organizations pride themselves in being extremely objective in the way in which they put the questions across, but there is no doubt that the organization that commissions the poll and indicates the kinds of questions it wants to be asked in a sense poses leading questions to provide the answers that it is looking for. In that sense there is a great deal of power associated with polling. I am wondering if there is sufficient regulation or self-regulation of the methodological approach to polling to ensure that a fair level of objectivity is maintained in the polling industry.

The power of polling is indeed awesome. It is quite scary that the government of Canada puts so much weight on Decima Research, that Decima results are widely assumed to affect government policy. Indeed, the province of New Brunswick pays some $24,000 annually to get the Decima results, to learn what influences Ottawa's thinking and planning, so people take this pretty seriously. To me it is quite scary to consider that polling has such influence in the governing of our country. There certainly is power and influence in polls.

In this context, I would like to deal with the question of the access to polling information by the Legislature. Set aside the importance of polling information and consider the relationship between the executive and the Legislature; this is something that is a concern of all parliamentarians. The Legislature holds the executive accountable to the people for what it does and accountable to the taxpayers in terms of how it spends' that is why we have several readings of bills and we undertake to discuss money matters in the Legislatures to approve the government's plans. So we are indeed holding the executive branch of government accountable.

I feel it is a matter of grave concern for parliamentarians if executive polling is not available to Legislatures.

If polling information is commissioned by the executive and not shared with the Legislature, it could be seen as an assault on private members' access to information, particularly if this information is paid for by the taxpayers of their jurisdictions. Therefore, I think it is reasonable to feel that any poll commissioned by the executive branch of a government should be tabled in the Legislature. What a private member then does with this information is up to him. I was interested in Mr. Blaikie's comments on this with regard to whether the private member should choose on the basis of the polling information to be a delegate of his people or a representative of his people; that is a decision for the private member to make. However, I think the important question is the access to information that the private member should have.

This brings us to another interesting question. As I understand it, governments that subscribe to Decima Research are under contract not to share the information, yet the subscription is paid for by public monies. I would put this question to you: Does this not place the executive in an awkward position with the Legislature? It is under contract not to table information that has been purchased with public monies authorized by the Legislature. I would be interested to hear the comments of others on the relationship between the executive and the Legislature with respect to polling information and on whether there should be a general move to require all polls commissioned by the executive to be tables in the Legislature.

Stephen Rogers MLA: I represent a British Columbia constituency in which the majority of constituents speaker neither of Canada's two official languages. I wonder what pollsters who speak only of Canada's two official languages do when they get to my constituency. Do they, as I suspect, hang up and find somebody else to poll? If so, their polling is not terribly representative.

Every now and then, the CBC pats itself on the back and issues a poll about what is happening with the three federal parties, when in fact there are five federal parties. In many polls that were issues by the CBC last year, there was no mention of a small group in Western Canada that is going to make Mr. Mulroney's life quite miserable. I never see any reference to that. Either someone in Toronto does not know or care, or they are living in a total vacuum. There are other political organizations sprouting in the West. Of course, that news does not travel terribly well.

About two years ago, I moved to a neighbourhood that was very average—in fact, it was startlingly average, because in the last federal election the poll showed that it was nip and tuck between the two candidates. It was the same in the provincial election. The houses in the area were modest. I recorded the number of times I was polled. I would like to know, is it always the same neighbourhood that is polled? I had never been polled before in my life, and I got polled there on everything, from disposable diapers to political parties.

Mr. Blaikie made a very good point about capital punishment. If you poll people about what they want to do about capital punishment they will tell you. If you ask them to pull the trigger, they give quite another opinion. That is one difference between an elected person and a person simply giving his or her opinion. The person might say that he or she is in favour of capital punishment, and therefore the poll will indicate that. However, if that person were asked to assist in tying the knot or releasing the gas, he or she would change his or her opinion. So there is a difference between doing it one way or the other.