Friday, March 26, 2010

Correlation Between Gun Laws and Deaths?

Gun laws are a subject many people feel strongly about, and as with any subject people feel strongly about there are no shortage of data and statistics that seem to support either side.  I decided to look into any correlation I could find between gun laws and deaths in the US.  Just like with political parties and deficit spending I didn't really have any expectations, but would probably guess that there wasn't much correlation.

Before I go any further I'll pretend like it is necessary to give a brief background on my stance on guns in general.  I've never owned a gun, and don't really have any strong desire to own one.  On the other hand I was in the Marines for 4 years and certainly am not uncomfortable around guns (I literally slept with a rifle for 22 days straight).  I am opposed to gun laws as I believe the largest threat to our society is government itself.  A well armed populace will be harder to subjugate.  A slight increase in deaths would be worth adding a layer of protection of our basic freedoms.  I'm also a big fan of personal freedoms, and I think people should be free to make decisions for themselves.  I think most of the opposition to guns in general comes from people that simply don't like guns because they are scared of them, and not because of any rational reason.

I've seen both sides of the gun control debate claim statistics support their side.  I'd like to think I'm being as impartial here as possible (as opposed to most people, who think they are biased).  I decided to do this comparison after I read this web page, which implies a correlation between Brady score and murders.  The Brady Campaign is an anti-gun group; they give each state a score 0-100 based on how strict its gun laws are.  I'll use their scores as a gauge of how strict gun laws are.  It is simply the only convenient measure I can think of, and I think it is likely to fairly accurate.

The first problem though is what to compare the score to.  This is likely the first place people looking to make statistics support their cause will diverge.  There are a number of different stats that could be used, all of which would appear similar to a causal inspection.  Deaths from guns are sure to higher in states with more guns.  Deaths from X are sure to be higher in states with more X.  On the other hand it is no secret that the upper New England states have both some of the laxist gun laws and lowest crime rates out of any states in the Union.

It is important to remember the old adage: "Correlation does not imply causation".  If it turned out that states with less gun laws tended to have less crime it wouldn't necessarily mean that the lack of gun laws caused the lesser crime.  Rather, it could be that the lesser crime caused the lack of gun laws, or that some third unknown variable caused them both.  Of course, it could be that the lack of gun laws did cause the reduction in crime.  The point is that if you do discover a correlation it is a starting point, not an ending point.  You need to find out what is causing the correlation.  You can't just assume that one thing causes the other.

I grabbed a bunch of different stats and found the correlation for all of them.  I got the data from here:
http://en.wikipedia.org/wiki/Crime_in_the_United_States
http://en.wikipedia.org/wiki/Gun_violence_in_the_United_States_by_state
http://www.statehealthfacts.org/comparemaptable.jsp?cat=2&ind=113
I then popped them into OpenOffice Calc, and used the correlation function, which "Returns the Pearson correlation coefficient of two sets of data".  If you don't know what correlation coefficient is, it is a number that ranges from -1 to 1 and tells how strongly related two sets of data are.  A correlation of 0 means no relation at all, while 1 means completely dependent.  In practice anything < 0.1 is no correlation, 0.1 to 0.3 is small, 0.3 to 0.5 is medium, and 0.5 to 1.0 is large.  A correlation of -1 is also completely dependent but in the opposite direction.  For our data this means that as the number gets closer to 1 there is a connection between a high score (stricter gun laws) and a higher whatever rate.  As the number gets closer to -1 it means there is a connection between higher score and lower whatever rate.  Since we are generally measuring something negative (murder, robbery) this effectively means that closer to -1 supports gun control, while closer to 1 supports gun freedom.

Here are the correlation results:
More Lax Gun Laws
Gun Deaths-0.545
Rape-0.398
Larceny-0.313
Burglary-0.257
Property Crime-0.219
Murder Rate-0.029
Serious Assault-0.025
Gun Murder Rate-0.001
Violent Crime0.020
Motor Vehicle Theft0.216
Robbery0.417
Murders Total0.434
Population Total0.476
More Strict Gun Laws

First allow me to explain some of the categories "Gun Deaths" is the only one that doesn't come from Wikipedia.  Based on the description from the site it appears to be exactly what it sounds like, all deaths from a gun, accidental or intentional.  I would guess "Rape" is only forcible rape (and not statutory rape), although the Wikipedia page doesn't specify that.  All of them are the per capita rates, except the last two, "Murders Total", and "Population Total".

What does the data tell us?  It should be no surprise that as the number of guns increase the number of deaths from them increase as well.  The pro gun counter would be that the increase in accidental deaths would be countered by the decrease in murders.  The rape one is interesting.  However, the fact that the rest of the violent crimes are so close to 0 leads me to believe that it is arbitrary.  The property crimes are pretty much meaningless.  You'd be hard pressed to convince me that an increase in guns somehow leads to an increase in burglary or larceny.  Yet there is a clear small to medium correlation between the two.  If nothing else this should show how meaningless these connections are.

Then come the big four.  Murders, Serious Assaults, Gun Murders, and Violent Crimes.  They are all amazingly close to 0.  I wouldn't have guessed there would be so little correlation, negative or positive, between violent crimes and gun laws.  I doubt gun laws have any effect on car theft so, like rape, I'd guess that is just noise.  Robbery is fairly high, I suppose it is possible that the fact that a potential victim could be armed could be deterring some robberies.  Perhaps it is distinct from the other violent crimes in that if someone is going to murder someone else a gun won't stop them; while if they are only looking for some money the thought of getting into an armed conflict may stop them.  Still I'd lean towards just random chance or some other factor as the reason.

The last two, "Murders Total" and "Population Total" don't mean much as far as gun laws and murder go.  It's no secret that as population goes up so do murders.  The only mildly interesting thing here is that there is a clear correlation between high population states and stricter gun laws.  Really though, that shouldn't be a surprise either.

So in summary it would seem gun laws have little correlation with violent crimes.  Feel free to distort and skew these results to fit whatever your pet beliefs are.

No comments:

Post a Comment