« America's Dangerous Debt | Main | Does vote swapping increase efficiency? »


Feed You can follow this conversation by subscribing to the comment feed for this post.

One of the interesting stories seems to be that metro-Vancouverites in particular don't have landlines (saving money for rent, I guess). With Vancouver suburbs so contested, this could be significant.

FWIW the polling firm Ekos called my wife on her (unlisted) mobile number last week.

Good to know that the WCI median reader (or least survey-taker) is exactly like me and most of my friends.

Oblivious - "metro-Vancouverites in particular don't have landlines" Perhaps so they have something to do on the Skytrain? Perhaps the city's Asian-friendly culture? Perhaps land lines come with being in one place for a long time and owning your own home - harder to do in Vancouver?

Aaron - It is really difficult to find out exactly how Nanos selects the numbers to be called. I am *sure* I read that it was based on listed numbers, but I can't find that reference, all I can find from the Nanos website is this: "Nanos Research regularly conducts large-scale telephone-based quantitative research across North America through its call centres which are equipped with predictive dialers. Our call centers are located in Toronto, Montreal and Vancouver with full capability in English, French, Mandarin and Cantonese." There's also this article in the Globe and Mail that says "With more young Canadians relying on cellphones, pollsters have to spend more time and money tracking them down – though cynics note that with a lower voter turnout among the young, their absence from polls matters less. But even the traditional telephone samplings have become problematic: The 70-per-cent response rate of a generation ago is now down to 20 per cent as more householders monitor their call display or resist telephone intrusion."

If you don't believe the land line/not land line numbers, then just look at the call screening/not call screening numbers.

a) Poll accuracy NB - t/f I support MRIA standards. b) Yes, but basing ones vote on poll rather than policy, is still their democr. privldge

O.k., I'm wrong. This 2008 Ottawa Citizen article says that Nanos does use cell phone numbers - they pay extra to get cell phone numbers. I can (and have) corrected what I wrote in the post above, but I can't untweet what I posted to twitter.

Oh well. There goes any possibility of gaining a reputation for intelligent commentary.

On the contrary Frances, you have posited a very worthwhile question. All polls need to be scrutinized more closely now due to deteriorating samples and more aggressive data interpretation methods. Just because Nanos passes the test doesn't mean others do; they need to pony up and prove it.

I still have a Maclean's magazine from when Harper was first elected in 2006. It mentioned that there may have been a "rogue poll" in that election that overestimated Liberal support. When pollsters say "accurate to 2% 19 times out of 20", a rogue poll the 20th outlier.

Such occurrences may be becoming more frequent due to deterioration in data quality.

Besides, I live in a Tory-held riding in Central Ontario which hasn't really been contested, but the Liberals are winning the sign war. Especially as their signs are on people's homes, not businesses or public places. I've heard from friends in Ottawa who live in a Tory riding who say the same thing.

As to why WCI readers prefer the Liberals, I imagine the Harper government's ravaging of the tax system with boutique tax credits, the GST cut and for met at least the slap in the face of income splitting has put off many policy-minded readers, which is all of us. WCI has run a series of posts detailing the follies of Tory tax policy and as I recall most readers have disagreed with the underlying government policies.

Determinant - and that 2008 article just suggests that some cell numbers are in Nanos data base. That doesn't mean that the odds of getting called with a cell # are the same as the odds of getting called with a land line.

If Aaron's wife decides that she's fed up with using a mobile company that sells her number to polling companies and switches carriers, her number will be out of the data base.

"Each evening a new group of 400 eligible voters are interviewed."

Each evening:.. Usually at supper time. And at this time of year, for us in the West, during a hockey game. Not conducive to high response rates.

eligible voters:.. How do they know if a person is eligible? You have to be registered to be eligible. Many are not registered now, waiting to register on voting day. Assuming they mean eligible to be registered to vote, do they ask, and if the person says NO (because of age or citizenship) do they call someone else to make up the 400?

are interviewed:.. I'll leave that to your readers to think about...

Someone on twitter calls our attention to this article

Eligible really isn't a problem in Canada, Richard. Same-day registration has been allowed and in fact customary for decades. As long as you have mail supporting the fact that you live at your declared address or a friend to affirm that your declared address is true, the DRO and Poll Clerk will register you on the spot and the Poll Clerk will administer the oath.

I was a Poll Clerk last election and did exactly that many times, often with students.

As long as you are 18 years old on polling day and are a citizen, you can vote. We have had national uniform election laws since the 1920's and Florida-style problems just don't happen here any more.

If pollsters ask that a person is eligible on polling day I have little reason to doubt their numbers on those grounds.

Nate Silver has done a lot of coverage and analysis of the cellphone effect on US elections in his Five-Thirty-Eight blog. In the 2008 Presidential cycle he found that landline-only polls tended to have a 2.8 point skew against Obama. The percentage of adults with wireless phone service only is 23% down here.

While it might not apply perfectly to Canada, it's worth a look:




You stated: One unequivocal finding from this survey is that people who participate in WCI surveys are wildly unrepresentative of the population as a whole - young, almost 90% male, blisteringly highly educated, intensely political and, for some reason no one on the WCI team can quite figure out, disproportionately likely to vote Liberal.

In the last 5 years, there were at least three scholarly studies (most recent by UT & George Mason profs), that surveyed US and Canadian faculty i.e.all highly educated as with WCI, and found that approximately 80% in US and 85% in Canada (if I recall correctly) self identified as liberal, left liberal, social democrat or beyond while the 15% to 20% self identified as a variant of conservative (and 14.9% are hiding deep in the academic closet).

These scholarly surveys prompted me to suggest to my colleagues that the academy is just like Fox News! Fox News offers a token liberal in any discussion and all other analysts are conservatives - thus Fox claims it is "fair and balanced". In the academy, we have a token conservative or two and all the rest are liberal, left liberal, social democrat and beyond - just like Fox News, we too in the academy are "fair and balanced" :)

But Schumpeter understood this phenomenon long before us in Capitalism, Socialism and Democracy, when he brilliantly turned Marx on his head by arguing that capitalism would fail not due to depression per Marx, but because of its world historical material successes which, inter alia, funded an enormous growth in professors, NGOs et al who would become relentlessly critical of the very system or conditions of their own possibility - "an atmosphere of almost universal hostility to its own social order". These demands would eventually destroy the vitality of creative destruction that made the cornucopia possible.

Ian - I wonder if appreciation of political positions is a bit like appreciation of music. The ill informed will enjoy the catchy, easy melodies of pop music. Those with more sophisticated tastes will only enjoy Justin Bieber if his music is slowed down 800 times.

Frances - it is not the appreciation but the distribution that I noted, - as your initial post raised methodology issues concerning the representativeness of the samples used in the polls.

The distribution of political positions in the academy appears somewhat skewed at approx 80:20 and not representative of the larger population.

Those who support proportional reprersentation in elections or employment equity in hiring appeal to representativeness as a foundational argument which in turn is based on equity.

I merely noted that in the academy - as may be the case with pollsters - we do not always meet our own perhaps demanding standard concerning representativeness.

I don't know how you're drawing inferences about opinion in academia from that poll.

I was providing a possible interpretation in response to Frances' comment:

"blisteringly highly educated, intensely political and, for some reason no one on the WCI team can quite figure out, disproportionately likely to vote Liberal"

I drew on Schumpeter who developed the core of his theory in C, S and D, in the chapter, Sociology of the Intellectual, wherein he argued that and explained why highly educated individuals are disproportionately liberal.

This "class" (don't recall that Schumpeter called them a class - Djilas did in the 1960s as did David Brooks in his Bobos book much later), was enabled and created by the vast wealth generated by markets, as firms had needs for experts. Galbraith later called it the technostructure.

However, this group or "class" of professors, NGOs and public servants, is independent of firms in their own institutions and becomes - Schumpeter predicted - hostile to the market economy and continually demands reforms to "tame the market" of creative destruction.

“the very success of capitalist enterprise paradoxically tends to impair the prestige or social weight of the class primarily associated with it [i.e., the entrepreneurial bourgeoisie] and that the giant unit of control [the large corporation] tends to oust the bourgeoisie from the functions to which it owed that social weight” (p. 139). The decline of family-owned businesses “takes the life out of the idea of property. . . . Dematerialized, defunctionalized and absentee ownership does not . . . call forth moral allegiance as the vital form of property did” (p. 142). Hence capitalism loses much of its moral legitimacy, and an anticapitalist social atmosphere emerges.

Following and applying Schumpeter in a straight forward way, suggests that highly educated individuals are much more likely to subscribe to liberal ideas and values.

Thus, to answer your question, the relationship, per Schumpeter, is between highly educated and political views.

One more quick point.

There is a scholarly "niche" with a long tradition analyzing the sociology of the intellectual. Some economists in the German Historical School including most importantly Max Weber and Schumpeter, studied this issue.

In modern times, Tom Wolfe (BTW has a PhD, Yale) satirized the "chattering class" in 1970 in Radical Chic and MauMaing the Flack Catchers.

Most recently, the NYT David Brooks in his book: Bobos in Paradise: The new upper class and how they got there" addressed this issue.

The larger point is that professors have much in common in terms of values and sociology with NGOs and public servants. One could probably write a decent paper arguing that some think tanks emerged as a "substitute" for the academy as an antidote to perceived scholarly bias against markets and private firms.

Ian Lee: "I merely noted that in the academy - as may be the case with pollsters - we do not always meet our own perhaps demanding standard concerning representativeness."

Indeed.  The stupid are grossly underrepresented.

My theory on why at least urban Vancouverites don't have land lines: they live outside of their smaller homes much more of the time: at the park or seawall, at the coffee shop or restaurant, on the ski hill or at the beach, on skytrain, etc.

I suspect suburban vancouverites are more likely to have landlines.

-Signed a Vancouverite with no landline-

Wendy - at least you didn't start talking about daffodils or cherry tree blossoms - but family and the ocean are what I miss the most -

- Signed an ex-Vancouverite with a landline -

I only have a bachelor's degree. I guess I represent the less-educated end of WCI.

K: "Indeed. The stupid are grossly underrepresented."

That's of course the danger. "We can't hire or promote him; he's conservative, so must be stupid". Megan McArdle did a couple of good posts on this a month or so back; how the liberal and conservative positions suddenly switch when we talk about representation by politics, rather than race or sex.

I'm still puzzled why it's so overwhelmingly Liberal on this blog though. Why not more NDPers, if it's so leftie? Maybe NDPers don't like "orthodox" economics much?

Nick: "We can't hire or promote him; he's conservative, so must be stupid".

I agree with you that this would be a danger, if it actually occurred. Perhaps it does some other places. But certainly I've never experienced that in a Canadian academic economics department. I can tell you my colleagues' food preferences (who can't eat onions or shellfish/peanuts or spicy food or meat, who keeps kosher, etc). I know Ian's political preferences because he's run for public office, other than that I only know about the political preferences of economics/business colleagues who are also close friends.

But I suspect that liberal political preferences - if indeed academics have such preferences - are much more likely to be the by-product of selection on other types of criteria.

Let's think about it from the perspective of an academic hiring committee. We want someone who is smart enough to be a highly paid corporate lawyer. But we want that person to reject corporate law in favour of a job that offers irregular work hours, the opportunity to work from home, and freedom to wear what you want. Shaving is optional. For both sexes. There are great opportunities to travel - or not. You will meet loads of new people, all the time, people who will challenge your ideas and expose you to new ones. You can work on whatever kind of research you feel like working on.

Only certain types of people will go for that particular job description - and liking the freedom of an academic job might well be correlated with personality traits commonly found in liberals.

If you'd asked me to guess the political character of the average WCI reader, I'd have guessed libertarian - fiscal conservative - social liberal. That was why I was surprised by the extent of the Liberal preferences, because I'd have thought a fair number of fiscally conservative/socially liberal people might find themselves affiliated with the Conservative Party of Canada. But perhaps the CPC's socially conservative positions and not-so-fiscally responsible history is turning off our readers.

Either that or the on-line poll wasn't even representative of WCI readers, to say nothing of the population as a whole.

Pretty much, I think. The Liberal's balanced budget streak and the deficits under the Tories (I don't blame them really, the recession shredded their plans) refutes any simple correlation between parties and budget balance.

Taxation is the issue here. The Tories want less taxes and more service cuts, the Liberals want more taxes and less service cuts.

It seems both the Liberals and the Tories have the same economic roots (stimulus during a recession, balanced budgets when possible) but differ on the method to get there and the credibility that a voter can or will give to such a policy choice.

Frances - for the record - as you raised it - I ran in 1993 in Ottawa Centre for the PCs. Since then, I have NOT belonged to any political party and am NOT involved in partisan politics at all since then. However, I will not pretend that I do not hold certain positions concerning fiscal and monetary policy or economic ;policies concerning economic growth.
I selected Ottawa Centre as it had voted Liberal since 1896 in that riding and its predecessor ridings. Thus, I could not win - which was ideal - a platform to support certain policies without fear of being elected.
I wanted a platform to support liberalized ("free") trade, the GST and economic growth policies as intellectuals were overwhelmingly opposed to each in 1993 (as many are in this year's election in 2011 guise e.g. CIT). During that campaign, the most outlandish, absurd, nonsensical claims were posited by intellectuals including the suggestion that NAFTA would force Canada to dismantle public health care, Canada would be forced to sell water and energy to the US et al.
I had no substantive interest in social policy (although I supported womens right to choose, gun control in the urban and gay marriage - I was an "early adopter" to use the language of business strategy).
To bring this back to the current debate, my argument in the posts above today, is that as the German-American philosopher Hannah Arendt argued, each "person has interests in the world, because we have bodies". (The only truly neutral people in the world are buried in the cemeteries.
Thus, it is profoundly disingenuous of the academy to pretend we are neutral, above the fray individuals concerning public policy.
Indeed, we are just as "biased" as Fox News, for we structure ourselves in a way somewhat similar to Fox to create the patina and illusion of neutrality and balance.

"We can't hire or promote him; he's conservative, so must be stupid."
"But certainly I've never experienced that in a Canadian academic economics department."

I would be very surprised if there were conscious malice towards conservatives in academia. The issue is less direct, but a world where 80% of professors are liberal is going to result in some biases.

Firstly, I suspect that our ideological leanings have at least some impact on the research questions we investigate. In my own field, for instance, "democratic peace theory" is the source of much inquiry, even though there is also evidence of a "communist" peace (I suspect for similar reasons). People that take on research questions that fail to address an existing literature often find themselves with no audience, need to construct datasets from scratch, and may have trouble getting grants from people that are already situated in the literature. Is that because such questions are inherently uninteresting, or just uninteresting to the people that are inclined to become professors?

Secondly, some research projects that conservatives might be more inclined to take up are inherently taboo. For instance if we think of Larry Summers' infamous remarks about women in science, they also represented an empirical conjecture that could be studied. I'm guessing anybody poking their nose in that direction would experience a similar fate if their coefficients were significant in the wrong direction (I hope this isn't interpreted as an endorsement of say, the work of Philippe Rushton - I think he has the absolute right to study the research question he has selected, but that he deserves to be castigated for his slipshop methodology).

Thirdly, the critical process going on BEFORE somebody is up for tenure is not exactly sacrosanct. Reviewers, journal editors and grant... granters are all critical gatekeepers for the field, subject to little scrutiny. Even if only 10% exhibit some measure of bias against illiberal proposals (which need not be conscious - humans find it taxing to hold conflicting ideas in our heads, and have a strong tendency to expunge those that don't fit), that can pose an additional burden for non-liberal scholars - particularly where consensus is required.

Fourthly, networking does matter in academia. Most academics will have no problem befriending conservatives, but some do. Moreover, conservatives may be less able to include themselves in political conversations - or may be more withdrawn in general, for fear that their deep dark secret might be revealed.

Hey Frances,

Even if polls are fundamentally poorly constructed and biased, is that relevant if their results are, say with 3% margin of error, mostly accurate? It would be easy to verify for Angus Reid (but not us I suppose) to simply compare their predictions versus election results over time to see if polls are a good predictor of actual results. We often must use variables that we know aren't perfect but make acceptable proxies, and the ultimate test of the proxy is to see how accurate it is.

"... for some reason no one on the WCI team can quite figure out, disproportionately likely to vote Liberal."

The Globe ran an article a couple weeks ago on an Ekos poll which found that the Conservatives do poorly among the university-educated (it's not just academics, Ian), while they have a lock on traditionalist voters who value leadership over expertise (about one in four voters). The cancellation of the long-form census illustrates the divide between the two.


The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad