Keyboard worrier

Wednesday, May 02, 2012

Gun murders revisited

Yesterday, I grabbed some Wiki data to look at whether a higher rate of gun ownership means a lower chance of being killed with a gun. I divided one by the other and it didn't look like the argument stood up.

Today I'll do it a different way: I'll MULTIPLY one by the other, on the assumption that if the theory is correct, one figure gets lower and the other gets higher, so the line should be reasonably even, even if it might be angled (I'm sure a statistician can put me right, but at least I'm trying). Here's the data:


... and here's the graph (in block form):


The last 5 on the right leap out of the trend. My explanation is that higher rates of gun murders are more a function of lack of social cohesion and weak official control. What's yours?

8 comments:

Paddington said...

Perhaps the South Africans are just much better shots?

Nick Drew said...

what's with Thailand ?

(pardon my ignorance)

& where's Russia ?

Sackerson said...

I was only correlating the evidence I could find in those two places, hence omission of Russia and others.

A brief glance round the internet suggets that Thailand is indeed a violent place, though how much of that is down to Communist insurgents I don't know.

Sackerson said...

Sorry, I'm out of date, that would be Islamic insurgents now, in the South.

Paddington said...

By all accounts, Thailand is quite violent, especially in Bangkok.

Paddington said...

You would probably get a better model by dividing the gun death rate by the gun ownership rate.

Sackerson said...

@Paddington 15:46 - I did the inverse of that on Monday, would that be completely wrong?

Paddington said...

You would get the reciprocal. What I suggested would approximate the deaths per gun.