Friday, November 16, 2007

Scientific method ya’ll

Yes, it’s a method. It’s not about using sciencey words. It’s about finding some effect that’s measurable, observable and repeatable. Laymen like me can use it to troubleshoot a computer, or figure out why my car won’t run, or why I have to go to the bathroom too much. It's the way we know things. Sciencey words don't make science.

I’ve read a fair amount of abstracts, and I know that there are some creative ways of holding things constant in order to measure your key factor. That’s good.

I know there are some number crunching science types, that read this blog, that can easily tell what makes a good experiment and if the conclusion follows from the data and the collection methods. You probably certainly have a much sharper eye for this than I do. Sometimes the comparisons are drawn so lopsidedly that I have to wonder how these people can dare call themselves Scientists. I’m not talking about mere bias. I’m talking Piltdown type stuff. It’s as though the conclusion has happened before the data collection, based from their world view.

Ok. So here’s the quick version of my example. CNN announced that suicide rates among servicemen who have deployed is “Epidemic”. This was quickly pounced on by liberals who then announced that all these suicides were clearly because servicemen don’t like the mission and because of the guilt they feel from all the atrocities they’ve committed. The liberal template is that veterans are either homeless, suicidal, or alcoholic. Probably all 3. It’s been on NPR, and on other news outlets. It’s what they’ve been running with for a while. So lets look at the numbers.

Their study found that veterans were more than twice as likely to commit suicide in 2005 than non-vets. (Veterans committed suicide at the rate of between 18.7 to 20.8 per 100,000, compared to other Americans, who did so at the rate of 8.9 per 100,000.)

Wow. That’s a huge difference. I would expect Americans who have been badly injured to have a slightly higher suicide rate than the rest of the population, but not that high. 8.9% vs. 20%. Holy smokes. That’s substantial. Wait a sec. It’s not out of 100. It’s out of 100,000. Let me get my calculator quick. That’s.....er....not quite as “epidemic” as they’re promising. But yet even a small difference should be looked at closely.

Here’s the other part that should strike you as odd. Do you know the demographic that has the highest suicide rate in the United States? Young men. Males. Folks with one X chromosome and one Y chromosome. For the 20-24 age group, The World Health Organization tells us that in 2004 females (in the general population) committed suicide at a rate of 3.59 per 100,000; the male figure was 20.84.

Men just flat out commit suicide more.
Mostly young men.

Guess what demographic almost exclusively comprises deployable servicemen on their first enlistment?

Yep, you guess it. Men aged 18-22.

Why would someone compare group of almost exclusively men against a mixed gender population when looking for a cause of suicide rates, when they know full well that the biggest factors are age and gender?

Because they’re assholes, that’s why.

How can the words “liberal” and “scientist” go in the same sentence. Stop calling them scientists. They’re jerks.

There’s no way that they just accidentally missed those things. It was intentional, because fit a certain agenda. Nobody that uses scientific method for a living could do that by mistake.

All The News That’s Fit To Print

All The News That Fits We Print

Labels:

4 Comments:

Blogger moif said...

Fuck em.

Amid the madness I saw this today:

Mitsuoko Oshiro recalls how she was given a grenade by a soldier, who told her that if she failed to use it to kill herself and her family, she would be raped and tortured by the Americans.

"I wanted to die, but I couldn't do it. We fled to the hills when the Americans invaded, but they didn't harm us - they just let us go," she says.

9:33 PM  
Blogger brando said...

I like your first two words Moif. Sometimes I get too wordy and wrapped around the axle, when a simple "Fuck em" would serve just fine.

9:43 PM  
Blogger Michael L. Heien said...

you are correct. This is a common problem in these types of studies. People do not compare the right things to support their conclusions. The numbers are solid, but it's when people start making conclusions about why the numbers are higher they get in trouble. What fits in "their" picture is the war causes soldiers to kill themselves. Suicide is a serious thing, but the other underlying factors (race, age, gender) were overlooked for a narative to play on NPR.

A "classic" example is the correlation between ice cream sales and incidences of rape (r^2 = 0.84). Does it mean ice cream causes rape? No, it doesn't. They left warm weather out of the equation, both seem to go up with warm weather. Correlation does not equal causation, no matter what Michelle Norris says...

5:05 PM  
Blogger brando said...

MIH, I was actually thinking about you when I wrote about numbers guys who know how to look at data. Your posts about data make my head spin.

6:35 PM  

Post a Comment

<< Home