Recently there has been some crowing about North Dakota having only two murders in 2008 (and neither were committed with a gun):
- Gun Murders and Gun Laws
- But guns cause murder!
- Via SayUncle: Congratulations to North Dakota; Brady Campaign? Not so much.
- VPC: Epic Fail
- Despite Being A Top Gun-Owning State, North Dakota Recorded Exactly Two Murders In 2008
And of course North Dakota has a very poor score as far as the Brady Campaign is concerned.
While the claims made above are true this isn’t the whole story. It’s an example of cherry picking the data and is very risky if you are a scientist.
Using the FBI Uniform Crime Report (2005) and the Brady Campaign Scores on the states (2007) I came up with some interesting information. Starting with the FBI data I added the Brady Scores and added some additional rows and columns. The result is here*.
The bottom line is that the assuming someone is murdered then correlation between the Brady Score and a particular method of murder are as follows:
- Weapon:Correlation
- Firearm:0.009
- Knife:0.215
- Weapon other than firearm (including knives):-0.014
- Hands, fists, feet, etc. (including being pushed):
-0.1800.007
Correlation is a number between -1.0 and 1.0, inclusive. What the above numbers mean is that a good Brady score does not mean there is a reduction in the the percentage of murders committed with firearms. It may mean there is a slight increase in the percentage of murders committed with knives and a decrease in the percentage of unarmed murders in Brady favored states. Another interesting set of data would be to compare the total violent crime rate to Brady Scores. I have done that in the past with the result of discovering there was essentially no correlation. Perhaps I’ll have time to look at that this weekend sometime.
Update: As noted in the comments I made a mistake on one of the formulas in the spreadsheet. This only affected the correlation of the “Hands, fists, feet, etc.” murders with the Brady Score. Instead of there being a slight negative correlation there is instead essentially zero correlation with Brady Scores. I have updated the spreadsheet accordingly.
* Please check my numbers and formulas. The Brady Scores were hand entered and hence are error prone. Also note that the original FBI table does not list Florida. I did not remove it intentionally.
You also need to add the total population so that you can create a Firearm Murder Rate/Knife Murder Rate/etc.
Your current numbers only show the likelihood of a firearm murder given a murder. In otherwords, it’s only a statistic of “If you are murdered, what’s the likelihood of a gun being used” not “what’s the likelihood of being murdered with a gun”.
The difference being that if North Dakota’s two murders had been gun murders it would show up as 100%. But this doesn’t accurately reflect that you are extraordinarily safe from being shot in ND.
Absolutely correct. This is why I said I wanted to compute the violent crime rate when I get the time.
I think there is a mistake in your sheet. Column N should be the result of dividing column J by B, not I by B.
Michael,
You are correct. Thank you! It is fixed now.
The correlations are so low as to mean that there is effectively no correlation between the Brady score and murder. Which means the Brady score is completely useless at measuring which laws keep people safe.
This appears to be the data file you want for population
http://www.census.gov/popest/national/files/NST-EST2008-alldata.csv
Found at
http://www.census.gov/popest/datasets.html