Another shooting in the USA. Remind me about the reason for having guns again. - by SubJeff
Jason Moyer on 7/9/2012 at 01:16
Quote Posted by Subjective Effect
Going to need some axis labels there fella.
Unless he changed the graph since he posted it, it would appear to be labelled exactly the way you would label a line graph depicting multiple sets of data.
LarryG on 7/9/2012 at 01:31
Quote Posted by Jason Moyer
In areas where the rate of legal firearm possession is highest, people don't fucking shoot each other. This is, in a nutshell, why 99% of gun control arguments are complete bullshit.
No. That's not true. There is no correlation, neither negative nor positive. See above. No correlation. Worldwide, the number of guns per person has no relationship to the number of gun homicides per person. I suspect that the number of gun homicides is driven by other things. As you suggest, poverty may enter into it, as may political strife. But (by itself) the number of guns available is not a predictor of the number of gun related homicides (except for the trivial case of zero guns should predict zero gun homicides).
Jason Moyer on 7/9/2012 at 01:51
See above? You mean the chart that shows that as the rate of gun ownership increases, the rate of firearm homicides decreases?
june gloom on 7/9/2012 at 01:57
you dragged that out because i posted a song from GTA3 on fb didn't you
Jason Moyer on 7/9/2012 at 02:05
Maybe
LarryG on 7/9/2012 at 02:47
Quote Posted by Jason Moyer
See above? You mean the chart that shows that as the rate of gun ownership increases, the rate of firearm homicides decreases?
That's not what it shows. It shows no correction at all.
I also "ran the numbers" on the correlations between Firearm homicide rate per 100K pop and Average firearms per 100K pop using minitab. The bottom line is that there is no correlation. This is most easily seen by looking at the P-value of 0.296*. That means there is no statistical correlation between those two sets of data. The slope of the regression equation is practically zero (- 0.000085), which also indicates that there is no correlation. If you want the results:
Correlations: Firearm homicide rate per 100K, Average firearms per 100K pop
Pearson correlation of Firearm homicide rate per 100K and Average firearms per
100K pop = -0.102
P-Value = 0.296
Regression Analysis: Firearm homicide versus Average firearms
The regression equation is
Firearm homicide rate per 100K = 5.72 - 0.000085 Average firearms per 100K pop
107 cases used, 78 cases contain missing values
Predictor Coef SE Coef T P
Constant 5.718 1.435 3.99 0.000
Average firearms per 100K pop -0.00008527 0.00008115 -1.05 0.296
S = 10.7430 R-Sq = 1.0% R-Sq(adj) = 0.1%
Analysis of Variance
Source DF SS MS F P
Regression 1 127.4 127.4 1.10 0.296
Residual Error 105 12118.3 115.4
Total 106 12245.8
Unusual Observations
Average Firearm
firearms homicide
per 100K rate per
Obs pop 100K Fit SE Fit Residual St Resid
36 5900 27.09 5.22 1.16 21.87 2.05R
52 5800 39.90 5.22 1.16 34.68 3.25R
59 45300 0.45 1.86 2.88 -1.41 -0.14 X
68 13100 34.81 4.60 1.04 30.21 2.83R
73 6200 68.43 5.19 1.15 63.24 5.92R
84 8100 39.40 5.03 1.09 34.37 3.22R
163 45700 0.77 1.82 2.91 -1.05 -0.10 X
170 1600 27.31 5.58 1.35 21.73 2.04R
177 88800 2.97 -1.85 6.30 4.82 0.55 X
180 10700 38.97 4.81 1.05 34.16 3.20R
183 54800 * 1.05 3.61 * * X
R denotes an observation with a large standardized residual.
X denotes an observation whose X value gives it large influence.
- - - - - - -
* the standard rule of thumb is that if P < .05 then there is a 95% likelihood that a correlation exists and otherwise, there is not. A value of 0.296 indicates a near certainty that no correlation exists
A "simple" explanation: (
http://www.gifted.uconn.edu/siegle/research/correlation/alphaleve.htm) Is there a relationship (difference) or isn't there a relationship (difference)? In our case r=-0.102 and P=0.296, so no correlation.
Vivian on 7/9/2012 at 07:26
R squared of 0.1? Why is it in percent? But yeah, other than that, seems pretty convincing given the correlation value. It's not a regression analysis though, so why the reg equation? Mind you, I pretty much only do model 1 or 2 regression, so I'm not that familiar with the pearsons.
But anyways, nice find! I'd still like to see that culled to countries in more equivalent economic states, say western europe vs US or something, but it's interesting.
Chimpy Chompy on 7/9/2012 at 07:57
Wait, shouldn't we be comparing gun ownership to *total* homicides? Not just gun homicides. Being killed with a gun isn't somehow intrinsically worse than being killed with anything else.
SubJeff on 7/9/2012 at 08:08
None of this matters. Correlation is not causation.
If there are more deaths/1000 pop on the road in country A than country B, and also a higher percent of Toyotas in country A than country B there is a correlation between deaths/1000 pop and percent of cars that are Toyotas.
This is not a cause, and it does not mean that the more Toyotas you have the higher your road deaths will be.
