rant is where the heart is

diaryland: sirilyan.diaryland.com: entry for 2005-11-26 (15:52)
In which our plucky young hero gets it right, 19 times out of 20.

Q. Is it time for an election?

Yes. Ottawa Owen, the official woodchuck of Elections Canada, will have had seen his shadow on Monday, according to a motion of confidence that the leader of the Opposition, Jack Layton, put forth in Parliament.

Q. How will I know who to vote for?

Lost your flippin' coin, eh? Don't worry. There's a simple method of gauging which party will govern wisest and bestest for the next two months to five years.

Q. Are you referring to carefully reading all of the party platforms, judging the likelihood of each of them working as described, and then voting for the best plan with the best chance of success?

No, silly. I'm talking about polls.

Q. What's a poll?

A poll, or in French a poile, is when a media company hires a polling company to call you right the hell in the middle of dinner and ask you who you intend to vote for, who you intend to not vote for, why you intend to vote for or not vote for them, and what color underwear you have on.

Q. Pink. With unicorns!

After contacting several hundred or thousand people, the polling company reduces this huge amount of information to a few bar charts, which your local newspaper prints the next day.

Q. So this "poll" shows me who to vote for?

Not necessarily. Some polling techniques are better than others at getting an accurate measure of the electorate.

Q. What is an example of an unreliable or flawed poll?

Unreliable polls have questionable methodology and use so-called "push" questions to get the desired response. For example:

• Would you vote for the Liberal Party?
  
• In the next election, will you vote for the Liberals?
  
• In your opinion, does the Liberal Party have a chance of getting your vote?

Note that all of these questions assume that the Liberals will receive "votes" in an "election". The possibility that people will not vote for the Liberals, or that an election will not be held, is not even entertained.

Fair and unbiased alternatives to these "push" questions include:

• Did you know a Liberal backbencher might think you're getting fat?
• Which is better: the Liberals or the number five?
• How many angels can dance on the head of a pin?
• There are those who say the Liberals make small, cute ponies drink battery acid. Is this unsubstantiated but also completely non-disproven rumor much more likely to make you hate them, more likely to make you hate them, or do you hate ponies?

Polls can also be influenced by sample size. For example, a poll that asks 1000 people across the country for their opinion will produce skewed results, because there is a good chance that all one thousand respondents will be cousins of Dave Dingwall. He has a very large family. The only reliable sample is one taken of every single Canadian of voting age. (Sadly, this means that every election result since Confederation is unreliable, but such is the price of statistical soundness.)

Q. How can I detect an unreliable poll?

Unreliable polls will often show unusual results at odds with what you know to be true. Some examples of these unusual results are:

• The Liberals poll higher than that bear, you know, the one that ate people. Good people.
• On the question "Which party's leader do you find most trustworthy?", "Liberal" comes in ahead of "Communist, of Russia, circa 1933 or thereabouts".
• Canadians rank health care, the environment, social spending, and the economy as priorities, not corruption, corruption, the crippling tax load that bends the spine of every citizen of the country, and corruption.
• The Greens are listed with a "2%" or "insignificant" level of support, rather than "99.97225% if the goddamn media would just give us any coverage I'm going to rabble.ca to rant now".

Q. This all sounds very complex. Can you give me an easy set of guidelines?

Yes. Every poll that shows your favorite candidate is losing is flawed, and every poll that shows your favorite candidate is winning is underreported.

Q. What, all of them?

You haven't been reading enough political blogs, have you?

sirilyan.diaryland.com | sirilyan dotcom
anything said in lowercase sounds profound. say it to me.