Posted by Kromey at 5:28pm Jul 3 '12Looking at the time period from 1991-2010, I plugged in all the data for all 50 states (plus D.C. and the US aggregate), then looked up the Brady Campaign's "scorecard" for 2010 and added that.
The goal here is to do some preliminary investigation into your premise that more guns hinders the rate of decline of violent crime. This is not meant to be conclusive in any way, just a cursory examination of the data in order to arrive at a preliminary understanding of what might be uncovered in a more in-depth study.
I used the crime data for 1991 and 2010 from that Disaster Center website you've provided to look at violent crime rates in 2010, as well as percent decline from 1991-2010 (calculated as ([1991]-[2010])÷[1991]; a negative result therefore means an increase, a probably counter-intuitive result based on my expectation that we would see declines across the board).
As a metric for gun control laws, I chose the Brady Campaign's "scorecard" because they use a system that looks at several different factors to arrive at a numerical indicator of how strict gun control laws are in a given state (higher numbers are stricter states); for consistency with the crime rate data I dug up Brady Campaign's 2010 "scorecard". I chose this metric because it is the most comprehensive one I know of, taking into consideration the widest array of possible gun control measures. (While an interesting analysis would be to compare the change in a state's grade to the change in their violent crime rates, Brady Campaign did not begin "grading" states until 1997, and have changed their criteria multiple times since then -- making a valid year-to-year comparison extremely difficult.)
The first analysis I did is one that I've seen done numerous times before -- a simple, straight-up scatter plot of the Brady Campaign's grade for a state plotted against that state's violent crime rate:
The first thing you notice is that the overwhelming majority of states have very low scores on Brady's scale. The next thing you notice is that everywhere along the Brady Campaign's grading scale, there's significant variation in crime rates -- suggesting that the two data aren't related. In fact, a regression analysis (red line) results in an r-squared value of 0.0016 -- a very strong indicator that there is no correlation between the Brady Campaign's scoring system and violent crime rates.
What little correlation there is suggests that a higher Brady Campaign score is associated with a higher violent crime rate -- exactly the opposite of what they contend the results of their preferred gun control laws would have! Again, though, we can see from the r-squared value that there really is no correlation at all here -- "better" gun control laws has no bearing on violent crime rate.
For my second analysis, I chose to address the question of whether or not the prevalence of guns -- or rather, the prevalence of gun control laws -- has an influence one way or the other on the change in violent crime rates over time, as you suggested in your comparison of Illinois to Arizona:
A few things to note about this graph:
First, as already mentioned, a negative percentage here (y-axis) means an increase in violent crime. Second, we see a few states with highly negative changes, so much so in fact that they would seem to skew any attempted regression analysis. Were we doing a proper statistical study, we'd look seriously at whether or not these states would even be valid to consider when looking for a trend; I did not do this, however.
Before we go any further, there's two additional things we must keep in mind about this graph:
1) We've already shown that there's no correlation between Brady Campaign score and violent crime rate; it stands to reason that there wouldn't be a correlation between the rate of change, either.
2) This analysis is flawed from the get-go, because it looks only at each state's ending Brady score, without consideration to how those laws have changed over the same period. A more rigorous examination would take the criteria for the 2010 Brady score and apply that to each of the states' laws as they were in 1991, and then compare the change in Brady score to the change in violent crime rates.
So, with all of that out of the way, what's left? Well, as I'm sure you've already noted, there are 8 states that saw an increase in violent crime over this period, and 7 of them have 2010 Brady scores below 10. And in fact, a linear regression does suggest that a higher Brady score means a greater reduction in violent crime rates -- although the r-squared value is so low (0.0860; sorry for not rounding that one on my chart) that we can confidently say there is no correlation.
Going slightly deeper into the raw data, we see that 7 of the 8 states with negative violent crime decreases (i.e. with increases) still, in 2010 (i.e. after the increase), have violent crime rates significantly below the US average -- in fact, around 50-60% of the US aggregate. (The 8th state, Alaska, has about 140% the US average, but also saw the smallest increase of any of them.)
In conclusion... well, there isn't really a conclusion. As stated before, this is a very cursory glance at the data. However, we already see some pretty clear signs that gun control laws do not correlate to violent crime rates, nor to changes in violent crime rates.
Linked below is my spreadsheet where I did all of this, so you can double-check my work as well as take the analysis further, if you wish.
The goal here is to do some preliminary investigation into your premise that more guns hinders the rate of decline of violent crime. This is not meant to be conclusive in any way, just a cursory examination of the data in order to arrive at a preliminary understanding of what might be uncovered in a more in-depth study.
I used the crime data for 1991 and 2010 from that Disaster Center website you've provided to look at violent crime rates in 2010, as well as percent decline from 1991-2010 (calculated as ([1991]-[2010])÷[1991]; a negative result therefore means an increase, a probably counter-intuitive result based on my expectation that we would see declines across the board).
As a metric for gun control laws, I chose the Brady Campaign's "scorecard" because they use a system that looks at several different factors to arrive at a numerical indicator of how strict gun control laws are in a given state (higher numbers are stricter states); for consistency with the crime rate data I dug up Brady Campaign's 2010 "scorecard". I chose this metric because it is the most comprehensive one I know of, taking into consideration the widest array of possible gun control measures. (While an interesting analysis would be to compare the change in a state's grade to the change in their violent crime rates, Brady Campaign did not begin "grading" states until 1997, and have changed their criteria multiple times since then -- making a valid year-to-year comparison extremely difficult.)
The first analysis I did is one that I've seen done numerous times before -- a simple, straight-up scatter plot of the Brady Campaign's grade for a state plotted against that state's violent crime rate:
The first thing you notice is that the overwhelming majority of states have very low scores on Brady's scale. The next thing you notice is that everywhere along the Brady Campaign's grading scale, there's significant variation in crime rates -- suggesting that the two data aren't related. In fact, a regression analysis (red line) results in an r-squared value of 0.0016 -- a very strong indicator that there is no correlation between the Brady Campaign's scoring system and violent crime rates.
What little correlation there is suggests that a higher Brady Campaign score is associated with a higher violent crime rate -- exactly the opposite of what they contend the results of their preferred gun control laws would have! Again, though, we can see from the r-squared value that there really is no correlation at all here -- "better" gun control laws has no bearing on violent crime rate.
For my second analysis, I chose to address the question of whether or not the prevalence of guns -- or rather, the prevalence of gun control laws -- has an influence one way or the other on the change in violent crime rates over time, as you suggested in your comparison of Illinois to Arizona:
A few things to note about this graph:
First, as already mentioned, a negative percentage here (y-axis) means an increase in violent crime. Second, we see a few states with highly negative changes, so much so in fact that they would seem to skew any attempted regression analysis. Were we doing a proper statistical study, we'd look seriously at whether or not these states would even be valid to consider when looking for a trend; I did not do this, however.
Before we go any further, there's two additional things we must keep in mind about this graph:
1) We've already shown that there's no correlation between Brady Campaign score and violent crime rate; it stands to reason that there wouldn't be a correlation between the rate of change, either.
2) This analysis is flawed from the get-go, because it looks only at each state's ending Brady score, without consideration to how those laws have changed over the same period. A more rigorous examination would take the criteria for the 2010 Brady score and apply that to each of the states' laws as they were in 1991, and then compare the change in Brady score to the change in violent crime rates.
So, with all of that out of the way, what's left? Well, as I'm sure you've already noted, there are 8 states that saw an increase in violent crime over this period, and 7 of them have 2010 Brady scores below 10. And in fact, a linear regression does suggest that a higher Brady score means a greater reduction in violent crime rates -- although the r-squared value is so low (0.0860; sorry for not rounding that one on my chart) that we can confidently say there is no correlation.
Going slightly deeper into the raw data, we see that 7 of the 8 states with negative violent crime decreases (i.e. with increases) still, in 2010 (i.e. after the increase), have violent crime rates significantly below the US average -- in fact, around 50-60% of the US aggregate. (The 8th state, Alaska, has about 140% the US average, but also saw the smallest increase of any of them.)
In conclusion... well, there isn't really a conclusion. As stated before, this is a very cursory glance at the data. However, we already see some pretty clear signs that gun control laws do not correlate to violent crime rates, nor to changes in violent crime rates.
Linked below is my spreadsheet where I did all of this, so you can double-check my work as well as take the analysis further, if you wish.
Link: Click Here
Did some analysis of this premise - Kromey