Grandinite

The folks over at the CBC seem to be having a rather embarassing Canada Day over their recent Facebook poll called The Great Canadian Wish List, and for the most part, it definitely blew up in ther faces. It only serves to prove that inserting the words “Great Canadian” into any online poll gets you results, as I did with my Great Canadian Blog Surey.

The number one wish of facebook users: abolish abortion.

Now that’s a story because the state-sponsored propaganda mill responsible for defining Canada’s self-perception has now fallen victim to what appears to be the Canadian version of Ron Paul supporters - highly active, tech savvy folks who have the means, time and energy to spam polls by setting up fake accounts or taking a poll multiple times. I’m not denying the wishes or the rights of these people to vote multiple times, but online polls fly in the face of democracy because they are inherently not only are they a self-selected sample, but completely misrepresentative of the total population.

A basic lesson in statistics is in order here. Self-selection bias in samples means that only the people who feel most strongly about an issue will respond to a survey. Here’s the lowdown, from wiki, a medium which is itself not immune to the issue:

http://en.wikipedia.org/wiki/Selection_bias

“Selection bias, sometimes referred to as the selection effect, is the error of distorting a statistical analysis due to the methodology of how the samples are collected. For example the sample selection may involve pre- or post-selecting the samples that may preferentially include or exclude certain kinds of results. Typically this causes measures of statistical significance to appear much stronger than they are, but it is also possible to cause completely illusory artifacts. Selection bias can be the result of scientific fraud which manipulate data directly, but more often is either unconscious or due to biases in the instruments used for observation.”

So, what we have here is direct manipulation of the data. Online polls that do not follow the “one person, one vote” principle are inherently non-representative of the total population. So, this data can’t be extrapolated across the greater population, and it is impossible to rank these wishes by votes. The data ought to have a caveat: it represents people who:

1. Have a facebook account. - Not everyone in Canada has one, and it’s definitely geared towards higher-income, urban youth.
2. Have plenty of free time. - Usually, students who have a low opportunity cost of time. If you’re working 80 hours a week, you’re using facebook as a communication medium, and that’s about it.

The CBC needs to understand several things about online polling.

1. You need to control for multiple responses by limiting the number of responses to one per IP address. Sure, someone could use proxy browsers and IP scramblers to bypass this, but such people are in the minority.
2. You must get demographic data on respondents, such as age, province of residence, income, gender, marital status, etc., and then compare these numbers to known population means and averages. If the CBC had done this with their Facebook poll, they would have likey found the demographic to be younger, low-income, single sudents.

I’m glad this happened to the CBC. I suppose there’s no such thing as failure, only opportunities to learn. I’d say the CBC has a few things to learn before venturing into the world of the new social media.