Learning to question statistics to break narratives.

I was listening to a couple of news stories today, and was intrigued about the base assumptions made in arguments, and the statistics thrown as some sort of "proof" that a certain course of action was the only way to fly.

In both cases, the statistics used were dubious at best. 

Now that Toronto is cluing in that theirs is a violent city (so is Hamilton, no airs about that), you have people on the Left calling for a ban on sales because some statistic is being bandied about that half the guns in crimes are domestic/legal in origin.

I doubt this stat for various reasons. For one, we do not know much about the illegally smuggled guns because lots of them get through from the US and there places to do it without detection. So we don't know a lot of things about illegal guns. (And CBC admits that this is an unverifiable stat. No kidding).

And if you ban the sale of legal guns, then all you do is force people intent on killing into getting untraceable illegal guns. That Toronto has a violence problem is beyond dispute, but trying to rig a narrative with some dubious stats will not help solve the problem.

But that was just one story, but there was another one where the journalist in question completely missed the point. The topic is not important, but the author of the piece was on the radio this morning whinging about some program for homeless men that was somehow profoundly unfair, but then bandied about a stat (hard to prove one, but let's say it is a ballpark) that 76% of homeless people are men, and just prattled on.

Hello! Why do you think that it is skewed?

Mostly because homeless women don't live as long as men in the same position. Another ballpark stat had it that, on average, homeless women survive about seven months on the street, which would explain the gender disparity. A lot of those women end up murdered by serial killers, for instance, but there are other dangers as well.

It reminds me of debates about the enslavement of Africans globally in another time and place. The US had far fewer slaves die in their country than others fared in other countries, and that's why you have such a significant percentage of Americans who came from that segment of history. In other countries, the group died out.

This isn't to give white Americans a lollipop -- they were the new country on the block on the time, and it may have been economically more feasible to take care of the kidnapped victims so they can keep working and produce more victims, than in other countries where they just kept importing their victims without worrying about the cost. It could be a simple question of cost effectiveness, but it had different consequences than in other countries.

We rarely ask questions when it comes to statistics, such as challenging them or analyzing what the numbers actually mean, let alone whether those numbers measure what they supposedly do.

But once we start questioning statistics and not try to use them to push a narrative, we can actually begin to dig for facts.

We can pose theories, and then test them -- but not just with stats -- we need other information to confirm or refute the theory.

To blindly defer to numbers often leads us from seeing obvious answers. And statistics is not a hard and fast form of mathematics where the formulas we use are beyond debate: what we use depends on how we gathered the data, and often there is big disagreement on methodology.

It's why so many a hapless student refers to Statistics as Sadistics.

We often are looking for TORTEE: that one rule that will sum everything up so that debate is shut down forever.

Nice try, but life doesn't play by our rules, nor should it...