[extoc]
Preface
Why Japan isn’t a Counter Example
Now before I get into the statistics and the analysis which can be a little dry, I’d like to point something out. Since my initial posting, there have been a few discussions in which the effectiveness of gun control laws can be “proven” be cases like Japan. The problem with using Japan as a lone example, or even an example with one or two other nations is that Japan is a very different nation from the United States.
It’s indeed true that since gun control laws were enacted, there have been very few gun related homicides. However it’s important to note that the gun was never Japan’s weapon of choice. Very few people ever owned guns in Japan. Beyond that, statistics at least indicate that Japan has, for a very long time, had a relatively low homicide rate over all. Now this does not mean that Japan is without violence. Japan has an usually high suicide rate: over twice that of the United States.
It seems that a lot of the differences between Japan and the United States can be summarized as internalized as opposed to externalized aggression. A Japanese person seems to internalize his aggression and stress until he snaps and take it out on himself while an American takes it out on someone else. This partially has to do with the culture of respect for others in Japan and it most likely also has to do with the massive media attention and almost level of fame received by someone in this nation who kills others. Therefore it can be assumed that it’s the difference in culture rather than the legislation that results in the differences between the United States and Japan.
In any case, Japan is not a good comparison to the United States. It is foolish to try to use Japan as a shining example of how gun control works since gun control was never really needed in Japan in the first place.
Data and Analysis
Update: A new analysis (in table form) that combines data from 2008 through 2011 can be found here. At four years worth of data (190 data points) there is enough evidence to suggest that there is indeed a difference. The proportion of homicides caused by firearms seems to be higher in states with the strictest gun laws. I’m working on including an analysis relating poverty and homicide rates as well.
On obtaining another year’s worth of data, I compared the pooled 2010 and 2011 information to get a better estimate on star vs no star states and also compared the change in gun related homicide rates in states that reduced in gun control score and those that did not in order to start getting a better feel of how the rates are changing over time. Of importance to note is that I am assuming the Brady Score methodology did not change from 2010 to 2011. If it has, this analysis needs to be redone. However under this assumption, the results are similar to my initial investigation.
One argument that was made that may have merit regarding the analysis is that guns can move too easily between states. Because of this, gun laws in a specific state do not greatly effect the availability of guns and so do not reduce gun related violence. However if that were the case, we still should have seen a significant difference in overall firearm homicide rates from 2010 to 2011 because overall, the Brady Score of the states dropped. That’s not the case. More importantly, it does not in any way explain why the proportion is higher in states with stricter laws.
The next step in my analysis would be to expand beyond the data in the United States. It seems that the European Union would be a good test platform since there should be plenty of statistics and there is a lot of cross traffic between regions within the European Union.
An unbiased, but admittedly older, meta-analysis of gun control studies can be found here.
Old Data
Calculations
At Least One Star vs None
| Group | NO STAR | STAR |
|---|---|---|
| Mean | 59.9300 | 62.2900 |
| SD | 11.9600 | 14.3600 |
| SEM | 1.5440 | 2.3933 |
| N | 60 | 36 |
Intermediate values
t = 0.8674
df = 94
standard error of difference = 2.721
Unpaired t test results
P value
The two-tailed P value equals 0.3879
By conventional criteria, this difference is considered to be not statistically significant.
Declining Score
| Group | DECREASE | NO DECREASE |
|---|---|---|
| Mean | 62.5300 | 60.8000 |
| SD | 11.4600 | 12.3200 |
| SEM | 3.3082 | 2.0533 |
| N | 12 | 36 |
Intermediate values
t = 0.4282
df = 46
standard error of difference = 4.040
Unpaired t test results
P value
The two-tailed P value equals 0.6705
By conventional criteria, this difference is considered to be not statistically significant.
Explanation
The first test is comparing the states that received at least one star with those that did not. 18 states received a star in each year, so all together that’s 36 data points with a star and 60 without. The conclusion is once again, no statistically significant difference between percentage firearm in the states with at least one star and those without a star.
The second test is the test comparing states that went down in Brady Score vs those that stayed the same or went up. Only one state went up (California) while 12 states decreased in score. Again the conclusion is the same. There’s no statistically significant difference.