Monday, February 15, 2010

Gay Marriage and Abortion

These are very serious and controversial issues. I will not talk about them at all. Instead, I'll talk about some other issues which concern me.


In Canada, it is illegal to charge more than 60% yearly in interest. But every day, thousands of Canadians are charged much higher rates at payday loan companies like Money Mart and Cash Money. The maximum amount these companies can charge in interest and "fees" is 0.13% per day. For example, if you borrow $1000 you can only be charged $1.28 per day! These stores get around the law by adding fees to the transaction, in addition to the interest rate, which can drive the actual annual interest rate up into the thousands of percent.

Payday loan companies pray on the poor and those who may not understand how punishing compound interest is. They must be regulated nationally.

Statistics and Surveys

There is a subject I think requires more explanation in school. That is Statistics.

Almost every day, we are exposed to some survey in the media. But we should not treat them as all the same. Surveys are only valid if 1) they are random, 2) enough people participate, 3) the questions are good.

First, many surveys aren't truly random. Telephone surveys can be slightly biased because they only survey people who own phones, have their number in a phone book, and (more importantly) agree to participate in the survey. Internet surveys are terrible because only people who are truly passionate tend to respond.

Second, the more people who participate, the more accurate the survey. But keep in mind the actual population size. 100 people surveyed may be fine to know the opinion of Kia drivers but not to know the opinion of all Canadians.

Third, the questions need to be clear, not-loaded, and unbiased. For example, "Should it be illegal to beat your child?" will give different results than "Do you think the government should be able to tell you how to discipline your child?". This is why it is important to look at the source of the survey if the results look wonky.

Finally, what do those margin of error statements mean (e.g., plus or minus 5 percentage points, 19 out of 20 times)? The first number is accuracy and the second is repeatability. Accuracy is how close the survey is to the actual opinion of the population being surveyed. Repeatability means if they did the survey again, how likely are they to get that same accuracy. Luckily, the repeatability on most surveys is 95% (19 out of 20 times) so looking at the percentage accuracy is a good judge of the survey.

Surveys are just one part of statistics. Many people confuse correlation and causation (e.g., most prisoners are smokers, but that doesn't mean most smokers are criminals) and put far too much weight on anecdotal evidence (e.g., violent crime has steadily decreased since the 70's, but people still yearn for the "good ol' days").

Okay, there I'm done. Oh right, I hate ads before movies. Yeah. Okay, now I'm done.