Thursday, March 02, 2006

Statistics

There is something I hate so much. People who claim "they don't trust statistics." Generally people who do this use it mid argument so that their unsubstantiated claims can continue existing. In other words, they distrust statistics that disagree with them.

Actually, a healthy distrust of statistics is reasonable- if someone quotes you a statistic, without telling you where he/she read it, then it is perfectly reasonable to discount it, although it really does depend who you are arguing with- if with a friend you can probably a bit more loose than a formal argument.

The problem is the person who just categorically says statistics are wrong. And they do exist. This if, of course, ridiculous. While a single statistic, quoted out of context, is close to meaningless (yet dos pollute political debate), a paper using statistical methods is clearly much better. Statistics is how medicine works, for example. Every single drug is tested randomly on a sample to make sure it works and does not have major side effects. Without statistics we simply would not have the healthcare system that currently exists. Yes, mistakes are made using this method, but you really can't argue with results- less people die from diseases every year thanks to this research.

Statistical research is very useful. Ultimately you cannot claim to know peoples opinions without statistics- otherwise you are really just guessing. That's what political polls try to do, although they are flawed, thanks to a reasonable proportion of people you ask to survey refusing to participate, creating a non-random survey (all statistical results are created from the assumption of randomness), so errors are made. Still, a survey taken of 1000 people selected at random is far more likely to be more accurate than someone saying "well I just know it's so." People make the mistake of assuming that just because something is sometimes wrong (as stastics can, and will be), they are therefore more accurate. It's similar to the creationists fallacy that, if they disprove the theory of evolution, therefore creationism is true, while clearly this is not true.

(to see why, consider this statement. This cat is not black. Therefore it is white)

8 Comments:

At 7:28 pm, Blogger Kelly said...

Good point about statistics. They have their value, but they are not everything. It kind of makes me want to eat some gummy bears.

 
At 4:40 pm, Anonymous Alex Pett said...

Has Alex, or should I say Falex read this? I am sure she can give you a million and one things to do with a standard deviation. Come to think of it, how many statistic modules to you take?

 
At 9:44 pm, Blogger The Venomous Bee said...

I agree that statistics can be manipulated, but if done honestly are really the only way of finding if, say, the Art Gallery you work at is meeting the needs of Older Adults in a way comparable to other galleries of its size (as I compiled endlessly last summer).

One of my best friends is in biology and economics and he LOVES statistics, because you can play with them. i.e. only one in six billion people is Clint Eastwood. The existence of Clint Eastwood is negligible. I'm sure, however, that one could argue for his existence by measuring the global irritation that "Unforgiven" won an Oscar.

Or, every Canadian is entitled to 1.09 breasts.

I'm out of my depth here, I know, and you wise number people are cringing.

 
At 12:21 am, Blogger cait said...

I think the thing about statistics is that it depends on how sound they are. Which sounds excessively stupid, I know, but really--if you have a large random sample group, and experienced knowledgeable people asking the most neutral questions possible, then statistics are lovely.

However, you also get surveys (like one I did) that ask questions like:

Are you a racist bastard or do you think that Native Canadians should be subject to the same laws as everyone else?

Stupid Jim Pankiw.

 
At 8:24 pm, Blogger Phil Plasma said...

Ever since I read the Foundation series I figured that statistics can play an intergral part of life as Psychohistory was at least partly based on statistics.

 
At 1:48 pm, Blogger Marc AndrĂ© BĂ©langer said...

As someone once said, statistics are like bikinis: more interesting by what they hide than what they reveal...

Seriously, statistic are a damn good tool. But they are just that: a tool. Not an end in themselves. If the science behind them is crap, the result is crap.

For example, I've often seen this fallacy: A and B covary. Therefore, A causes B. The statistics presented are not at fault, just the science.

 
At 5:53 pm, Blogger Kelly said...

Phil: I have only read Foundation and the second one . . . was it Foundation and Empire or Second Foundation? Anyway, I have the whole series now (as of yesterday, but Foundation is from a different edition, which makes me sad) and will start reading it again when I finish the Dune series--which is amazing, by the way. I can't believe how much time I wasted reading sci-fi/fantasy franchise series (BattleTech and Dragonlance--although the latter can be good sometimes).

 
At 8:35 pm, Blogger Mr K said...

Ah you gotta love the Foundation books, although they go a bit mad come foundation and earth....

 

Post a Comment

<< Home