Tag Archives: correlation

Correlations

Spend much time debating gun rights on-line, and you’ll be told that “stronger” gun laws correlate to better outcomes in terms of deaths.

All right, let’s find out. I’m drawing data from the following sources:

1. Homicide rates by state, 2013: http://www.deathpenaltyinfo.org/murder-rates-nationally-and-state

Numbers are homicides per 100,000.

2. Suicide rates by state, 2012: http://www.worldlifeexpectancy.com/usa/suicide

Numbers are suicides per 100,000.

3. Brady Campaign state scorecard, 2013: http://www.bradycampaign.org/2013-state-scorecard

Their methods are their own, but since they are opposed to gun rights, I presume that their scoring favors gun control over loose gun laws.

Both graphs use an X axis with values from 0 to 4 and a Y axis from 0 to 30. The X values are keyed to a four-point grade scale:

A: 4
A-: 3.75
B+: 3.25
B: 3
B-: 2.75
C+: 2.25
C: 2
C-: 1.75
D+: 1.25
D: 1
D-: 0.75
F: 0

The Y numbers are suicides or homicides per 100,000.

First the homicide numbers compared to the Brady score:

Homicide to Brady score

Notice the lack of a pattern, other than groups of states with the same gun laws having homicide rates at wild variance from each other? If the Brady Bunch were correct, there should be a strong correlation, not nothing.

Now let’s consider the suicide numbers:

Suicide to Brady score

Here, there is a low negative correlation, though as with homicide rates, the large grouping at X = 0 is significant.

Of course, as any student of statistics knows, correlation doesn’t imply causation. But if there is causation, there must be correlation. The lack of correlation in homicide rates and the weak correlation in suicide rates demonstrates that we cannot claim that the strictness of gun laws determines lives saved.

Feel free to share this next time you’re dealing with someone promoting gun control.