Thu, Aug

Wanting to be right *and* popular

Thu, Aug 9, 2007

What is the purpose of polls? Ask most people, and I think they will tell you that the purpose of having a poll is to try and accurately assess what the population thinks about a particular issue, such as who should run this country, what foreign policies we should pursue or not, and what taxes to raise and what spending to cut.

And why should polls be as accurate as possible? Because most of us believe in democracy; we believe that everyone should have some say in the running of this country, that their opinions should have a chance to compete with other people’s opinions, and no one should be denied a chance to participate.

Thus, the only accurate polls are the ones wherein the entire population gets asked (in Canada’s case, defined as all Canadian born and naturalized men and women eighteen years old and older). This, of course, takes considerable resources — about $300 million per federal election — but we wouldn’t think of doing things in such a way that we risk leaving people out. Because to do so would be undemocratic.

But most people running informal polls do not have the taxpayer-supported resources of Elections Canada. So instead the best they can muster is canvassing a subset of the population. Now everybody wants to believe that any poll they run provides representative results, but any statistician will tell you that, if you’re sampling a subset of the population, rather than canvassing the population as a whole, various measures must be taken to ensure that the sample is truly random, so that you can claim with confidence that it is an accurate representation of the population. Either that, or caveat upon caveat must be mounted onto the various polls in order to explain what the limitations are (in a poll of left-handed Newfoundland fishermen, 94% preferred the taste of chocolate to a kick in the head). Canadian opinion poll companies have extensive measures to try and eliminate bias from their sampling systems, but even they can be counted on to get their predictions wrong. Anybody running on fewer resources than the opinion polls have to accept that there is a strong chance that their poll results won’t be representative of the wider population. For online polls, this means that most results just aren’t worth the pixels they’re printed on.

Last month, the Beaver magazine ran an online poll asking readers, and anybody who could access their website who they thought the worst Canadian was. Thanks to a significant online campaign by a number of Conservative Party supporters, prime minister Pierre Elliott Trudeau came out on top (or, bottom, as it were), ahead of such luminaries as serial-murderer Clifford Olsen.

Now, normally I wouldn’t post about something like this, and the Beaver deserves the result it got, thanks to its failure to properly define the question, or ensure that the subset that their online poll sampled was representative of the whole. However, the actions here do serve to highlight how, like spamming, some of the fun of the Internet has disappeared thanks to the malicious actions of a few timewasters. Consider the similar example of CalgaryGrit’s Best Premier contest where heated political action raised a criminally-connected premier over the father of medicare in Saskatchewan, and a one-term NDPer over Big Daddy Bennett in British Columbia.

And the activists are proud of what they’ve done — which is to say, work very hard to skew the results of an online poll so that the final tally reflected their opinions, rather than possibly the opinions of the population as a whole.

Stageleft takes on a blogger Suzanne who commits a fairly blatant example of cognitive dissonance by crowing over the results of their “victory” while at the same time claiming that the results are somehow representative of what the wider population thinks. At least, that’s the take I have after the blogger claims that “all the freeping paid off”, and taking umbrage over Stageleft’s admonition that freeping is pretty close to cheating. The comments are an interesting read.

(Suzanne): it shatters the myth that Trudeau was universally loved and no one really questions his beliefs. In the mind of the media, he embodies the quintessential Canadian, but many people disagree with that, and I’m glad we got that point across.

(Mike): So SUZANNE admits that this is merely a trick and is not representative of the view of Canadians, as it was supposed to be. Yet she still thinks that it means something. Its no different that ballot-box stuffing or voting more than once. Is that not cheating?

It sure means that that “poll” is useless except to show how bloody immature you guys are.

(Suzanne): “I certainly didn’t think of it as a real poll.”

(JJ): Okay… The results are purposely skewed, so they aren’t real, and anyway the poll itself isn’t real to start with — then what’s the point of the whole exercise? And how can anyone see it as a “victory” of some kind? I’m fascinated. Really.

A fair chunk of the comments is a debate over whether the actions of the campaigners amounted to “freeping”. Suzanne used the word herself and claimed there was nothing wrong with it. “Freeping” is one of the uglier Internet words I know (ugly in terms of how it rolls off the tongue, not necessarily in what it describes), used to describe a grassroots or Astroturf political campaign to stack online polls so that the results match the desires of said group, rather than possibly reflect the opinions of the population as a whole. Such actions completely invalidate such online polls, since the opposing viewpoints rarely have a chance to express their opinion. That was the express intent of this campaign: to ensure victory by active numbers, rather than an accurate reflection of what the true numbers were. To then turn around and claim that the numbers mean anything, even “shattering the myth that Trudeau was universally loved” (a myth that doesn’t exist, incidentally), is an attempt to have one’s cake and eat it too.

In real voting, freeping would be little different from the hard work required by the various parties to get their vote out, but in real elections, all voters have a chance to express their opinion, and it is officially assumed (though not generally accurate) that a refusal to vote equals consent with the results. It is impossible to “freep” a general election because the interested parties are so much smaller than the population as a whole; if this wasn’t the case, well then their opponents would have no cause to complain because the numbers would still provide an accurate reflection of the population. When you’re canvassing the whole of the population rather than the subset, it’s impossible to get the numbers to lie.

But online polls, remember, do not have the resources of a general election, and they explicitly poll a small subset of the population. Most of the people who might have an opinion one way or the other are going to be left out. This allows small groups of dedicated followers to throw their weight around. If a significant number of people on one side of the debate are made aware of the poll, and not the opposing side, that allows them to skew the results. Thus online polls do not become representative samples of opinion, but opinion-generators. It highlights how active a small group of people are, which I suppose is fine in and of itself, but these proponents should not be surprised if their activities are viewed with suspicion, and the skewed results taken with a big grain of salt. Being politically active only affects how well various groups can convince undecided voters whether or not to follow their agenda, but in the end, democracy functions on total numbers, not the strength of opinion of a subset of numbers.

What is discouraging is that this sort of activity is common, and it exists on both sides of the ideological divide. And I think it’s a bitterness that comes from finding oneself often at odds with the national “consensus” that comes when a truly representative poll is taken. In their minds, a key line has been crossed: polls are no longer supposed to represent what society as a whole think, but should be tailored so that it appears that society as a whole thinks the way they do. In short, such an activity expresses a desire to eliminate one’s ideological opponents from society, so that one doesn’t have to confront their ideological opposition to one’s own point of view in the framework of civil society. It’s a frustration that comes with having one’s preconceived notions challenged, and not having the intellectual wherewithall to argue back. Better to just put yourself into a sub-community of like-minded friends, and try to silent dissent through a redistribution of the numbers.

Unfortunately, this attitude has affected the political process here in North America. While the federal election results remain pretty representative and free of these shenanigans so far (to the frustration of some of those desperate for a Conservative majority government), elements on both sides of the ideological divide in the United States obsess over getting their vote out (nothing wrong with that) and keeping the other side’s vote at home (which is where the problems lie). In the 1995 Quebec referendum, there was a similar debate over the voters lists, and questions over who was “Quebec enough” to vote.

We need to give ourselves a shake, here. It’s one thing to try and get the vote out, but it’s another matter entirely to hope that your ideological opponents are somehow blocked from expressing their own opinions. That is not a democratic way of thinking. It is a failure to acknowledge the worth of our fellow citizens in the land that we share, and we all deserve better than this.


One of the better ways I’ve seen to limit the impact of these immature vote-stuffing activities is to block access to the results until polls close or, as was the case in the Canadian Blog Awards, show the numbers, but not the names of the options that held those numbers. In the same manner that keeping the results of the ballot secret until the polls close, it limits an organizational rush to try and get a flagging competitor over the top.

While it’s silly and impractical to, say, block transmission of Atlantic Canada’s poll results in Ontario until the Ontario polls close, I like Damien Penny’s idea to not announce any of the poll results in any of the provinces until the polls close across Canada. That would surely survive a Charter challenge and strikes me as being fair for all. Why should there have to be an organizational rush to get the vote out when you know a vote is close? These people you are getting out to vote should be voting anyway, and you should be working to get out the vote regardless of what the interim results are.

On This Day

blog comments powered by Disqus