Little throw away article in the Metro

[Larry Legend]

So, I must admit I was reading the Metro today, but only cause I was on the can and it's a free publication so give me a little leeway.

Ran into an article that parroted an article from the Violence Policy Center:

Washington, DC—States with higher gun ownership rates and weak gun laws have the highest rates of gun death according to a new analysis by the Violence Policy Center (VPC) of just-released 2006 national data (the most recent available) from the federal Centers for Disease Control and Prevention’s National Center for Injury Prevention and Control.

The analysis reveals that the five states with the highest per capita gun death rates were Louisiana, Alabama, Alaska, Mississippi, and Nevada. Each of these states had a per capita gun death rate far exceeding the national per capita gun death rate of 10.32 per 100,000 for 2006. Each state has lax gun laws and higher gun ownership rates. By contrast, states with strong gun laws and low rates of gun ownership had far lower rates of firearm-related death. Ranking last in the nation for gun death was Hawaii, followed by Massachusetts, Rhode Island, Connecticut, and New York. (See chart below for top and bottom five states.



Other than the fact that their gun death rates group ALLLL shootings together (police, suicide, homicide, self defense, accidental, etc.), what are some of the best ways to rebut these articles and their facts.

Try to not get in a tizzy over this one.

 

[gene]

When did the Heller decision come down again? 2008

 

[Jose]

States with higher gun ownership rates and weak gun laws have the highest rates of gun death according to a new analysis by the Violence Policy Center (VPC)

Prima facie evidence of extreme bias and a hidden agenda. That is all you need to refute such drivel.

 

[Dick]

I’d like to share some thoughts on these numbers from the perspective of someone who’s made a good bit of his living over the last 40 or so years number-crunching large data sets for government and private clients. And I apologize in advance for the long-winded response, that's just the way I am (sometimes).

First, Jose is both right and wrong. Right that the VPC is biased and has an agenda (which I would say is not hidden) and therefore their position bears careful examination, but wrong that simply pointing that out is sufficient to refute their argument. In logic, that’s one of the classic informal fallacies, known as argumentum ad hominem (literally “argument to the man”) and it cannot prevail. Consider the situation in reverse: You make the argument that firearms owned and carried by civilians are often used to prevent crime, and often without a shot being fired. The individual arguing against you simply says that you’re a gun owner, therefore your position is biased and invalid. You would be correct if you thought that refutation to be rather unconvincing.

So, with that in mind, I took a look at both the VPC “study” and attempted to apply some basic statistical principles to it. First, I found it curious that they selected the states with highest number of gun deaths by rank and then presented the gun ownership data by actual numbers, only pointing out that they were higher than average. So I went to the raw data and looked at the gun ownership rankings and gun death rankings. When you do that, you come up with the following (I can't get the spacing to come out correctly, hence the slashes):

State/Rank by Gun Ownership/Rank by Gun Deaths

Wyoming/1/8
Alaska/2/4
Montana/3/17
South Dakota/4/31
West Virginia/5/13

Somewhat less convincing, no? It’s now not so clearly evident that high gun deaths and high gun ownership are in fact related – for example, how do you explain South Dakota, with its third highest gun ownership rate being well below the national average for gun deaths? Because VPC also similarly presented the states with the lowest gun death rates, I took a look at the death rate ranking in the states with the lowest gun ownership (note that for the purpose of this and all subsequent discussion, I included DC as a state):

State /Rank by Gun Ownership/Rank by Gun Deaths

DC/51/1
Hawaii/50/51
New Jersey/49/46
Massachusetts/48/50
Rhode Island/47/49

Well, we have the special case of DC, but other than that the states that have very low gun ownership also have very low rates of gun deaths. Taken together, these two sets of numbers are suggestive of a relationship between gun ownership and gun deaths, but is there really a statistically verifiable association here?

Statisticians use something called correlation analysis to determine if there is a relationship between two sets of paired data. The analysis, which I conducted using the ranks rather than the raw data, examines all of the data to determine if a relationship exists. It produces a statistic called the correlation coefficient, which can vary from +1.0, indicating a perfect relationship (in this case, that the highest ownership state is the highest death state, 2nd is 2nd, 3rd is 3rd, etc.) to -1.0 (highest ownership is lowest death, 2nd is 50th, 3rd is 49th, etc.). The closer the value of the coefficient is to either end of the spectrum, the closer the relationship. The value of the correlation coefficient can also be tested for statistical significance, by which we mean it’s unlikely that these results occurred by chance.

Using VPC’s own data – which they said came from CDC, and I accepted it as such – the correlation between gun ownership and total gun deaths by state is .4713, which is strongly indicative that such a relationship does in fact exist, and is also very highly statistically significant. So, whether we like it or not, there is a definite relationship between high rate of gun ownership in a state and a high rate of gun deaths in that same state.

However, you put your finger on the problem with VPC’s position when you correctly pointed out that their gun deaths include all types of death related to firearms. This is important because their discussion of gun deaths is very clearly focused on gun homicides (i.e., they talk about “permissive carry laws,” “restrictions on assault weapons,” “permits to purchase,” etc.) and, of course, that’s what the average citizen is worried about, as they well know. So the data of interest therefore, are not gun deaths, but gun homicides – and those data are readily available from the same source. I won’t go into all the details, but if you calculate the correlation coefficient for gun ownership vs. gun homicide by state the result is -.0818, indicating that there is essentially no relationship whatsoever. So what’s causing the relationship between gun ownership and gun deaths in general?

Data on “unintentional” deaths by firearms (which I assume means “accidental” although most of us here might prefer the term “negligent”) and suicides by firearms are also readily available, and it turns out that both of those categories are very highly correlated with gun ownership (both about .87), which shouldn’t come as a big shock since it’s reasonable to assume that when more guns are around there will be more gun accidents (after all, when more cars are around there will be more car accidents, when more boats more boating accidents, etc.), and I believe it's been known for some time that an individual determined to commit suicide will most often use a gun if there's one available.

So, that’s what appears to be going on here – VPC is using a true relationship between gun ownership and accidental/self-inflicted gun deaths to claim (or at least imply) a false relationship between gun ownership and gun homicide. It is scientifically disingenuous, and they most certainly know it. (As an aside, the numbers also clearly show that the relationship between gun ownership and non-homicide gun deaths is very real and argue that we should all be careful about using and storing our guns properly to reduce the number of accidental and/or self-inflicted deaths due to improper use).

 

[Marcus5aurelius]

I read that this morning on the T, such BS

 

[Rob Boudrie]

An observation that flies are found near garbage does not, in an of itself, prove that flies cause garbage.

I'll bet a statistical analysis of towns in MA would reveal that there is an inverse correlation between the issuing authority's tendancy to place restrictions on LTC-As and crime which, using the VPC logic, would establish that restricting LTCs increases the crime rate.

 

[livefreeorhere]

So, that’s what appears to be going on here – VPC is using a true relationship between gun ownership and accidental/self-inflicted gun deaths to claim (or at least imply) a false relationship between gun ownership and gun homicide. It is scientifically disingenuous, and they most certainly know it. (As an aside, the numbers also clearly show that the relationship between gun ownership and non-homicide gun deaths is very real and argue that we should all be careful about using and storing our guns properly to reduce the number of accidental and/or self-inflicted deaths due to improper use).

this is what irks me the most. As a scientist myself it's frustrating when people bend the numbers or fudge things to prove their own point. It's unacceptable to do this because all it does is confuse the people who are casual observers of the argument. Most people reading that on the T aren't going to check the numbers. What a bunch of asshats.

 

[livefreeorhere]

An observation that flies are found near garbage does not, in an of itself, prove that flies cause garbage.

I'll bet a statistical analysis of towns in MA would reveal that there is an inverse correlation between the issuing authority's tendancy to place restrictions on LTC-As and crime which, using the VPC logic, would establish that restricting LTCs increases the crime rate.

ahh the golden rule of correlation! Correlation does not imply causation!

 

[n1zfg]

If you torture the data enough, it will confess.
or
If you don't like the outcome, manipulate the sample.

There may be a correlation between gun related deaths and gun ownership. I'm also sure that motor vehicle death rates go down in regions where automobile ownership is less prevalent. It also does not refute John Lott's "More Guns, Less CRIME" thesis.

It's just worded to sound like it does.

 

[Hiltonizer]

My boss put this in front of me yesturday in a "take this" type fashion.

1) The figures are skewed against places like Alaska and Wyoming given their low populations. Despite this, NH with extremely lax gun laws is #6 in terms of safety.

2) You aren't afraid to walk down the street at night in Alaska or Wyoming, you are afraid to walk at night in Mass and NY.

3) Mississippi and Louisiana's other issues might have something to do with it.

 

[greycar]

a day later

On the same page number, an article about the shooting of a Malden man in Fresh Pond. With a comment that "There have been three homicides in Cambridge this year, already more than the last two years combined." Today's Fresh Pond story is not found on their web site either; just the VPC story is on the web site. Had to poke my finger in their eye at [email protected] . GC

 

[sgtdemeo]

My boss put this in front of me yesturday in a "take this" type fashion.

1) The figures are skewed against places like Alaska and Wyoming given their low populations. Despite this, NH with extremely lax gun laws is #6 in terms of safety.

2) You aren't afraid to walk down the street at night in Alaska or Wyoming, you are afraid to walk at night in Mass and NY.

3) Mississippi and Louisiana's other issues might have something to do with it.

Actually, NYC is the Nations safest major city this year, according to the FBI, with homicides at their lowest level since 1963.

 

[Dick]

Actually, NYC is the Nations safest major city this year, according to the FBI, with homicides at their lowest level since 1963.

But that doesn't address Hiltonizer's point - he's talking about states, not major cities, and the fact that NYC homicides are at their lowest point since 1963 doesn't say anything about how the homicide rate compares with anywhere else.

According to the data, your chances of being killed by a gun in Alaska and New York (state) are about the same - about 2.6 out of 100,000, which places them almost exactly in the middle of the range. Wyoming, in spite of (or maybe because of) it being the state with the highest percentage of gun ownership, has the 4th lowest rate of gun homicide, only 0.6 per 100,000. Massachusetts is lower than average, at 1.5.

I think another point that's worth mentioning is that you don't "walk down the street" across the entire state - meaning that it's difficult to generalize from figures that lump many different areas. For example, the Massachusetts rate is probably too low for Boston but much too high for many rural areas of MA where people don't lock their doors. Also, what would the state figures for Michigan look like without Detroit, or Louisiana without New Orleans? I'm sure there are even places in DC (with its gun homicide rate of around 30) where it's perfectly safe to be out and about at night, and I've been out and about in many of them.

 

[Bob P]

I think another point that's worth mentioning is that you don't "walk down the street" across the entire state - meaning that it's difficult to generalize from figures that lump many different areas. For example, the Massachusetts rate is probably too low for Boston but much too high for many rural areas of MA where people don't lock their doors. Also, what would the state figures for Michigan look like without Detroit, or Louisiana without New Orleans? I'm sure there are even places in DC (with its gun homicide rate of around 30) where it's perfectly safe to be out and about at night, and I've been out and about in many of them.

I think the above does the most to address the fallacy of trying to correlate gun ownership numbers with gun homicide and/or crime numbers. There are SO many other factors involved that gun access is merely a small one of many. More important are percentage of urban population, percentage unemployment, percentage of citizens below the poverty level etc. Let's face it, you give eveyone a gun in some affluent or even middle class suburb or rural area and you're not going to suddenly see a spike in crime stats. So, big surprise that cities like Detroit with rampant unemployment, poverty and really not much in the way of opportunity has soaring crime stats while Boston, which is really a lot better off economically them most other cities its size doesn't by comparison.

 

[sgtdemeo]

My post was a direct response to the poster who said that he didnt feel safe walking down the street in ny. Based on stats, or perception? Prob not stats. Seeing as how he chose 2 cities with gun control laws. My point is that nyc, with its gun control laws is still the safest in the country. Not the perception shares commonly on this site.

 

[livefreeorhere]

My post was a direct response to the poster who said that he didnt feel safe walking down the street in ny. Based on stats, or perception? Prob not stats. Seeing as how he chose 2 cities with gun control laws. My point is that nyc, with its gun control laws is still the safest in the country. Not the perception shares commonly on this site.

that still doesn't make me feel any better about walking down the street. NYC has a much higher crime rate than Concord NH. They're not lumped together because they're not both large cities but here's some data from the FBI website and areaconnect (Concord doesn't have 100,000 people):

Boston per capita "violent crime" .01 in 2008
NYC per capita .006
Billings, MT per capita: .003
Concord, NH per capita .0015

 

[Gadsden]

The subject of this thread pretty well sums it up.. "Throw away" and "Metro". It is a lousy liberal rag that they have to give away, and it seems they have a hard time doing even that.

 

[JRyan]

My post was a direct response to the poster who said that he didnt feel safe walking down the street in ny. Based on stats, or perception? Prob not stats. Seeing as how he chose 2 cities with gun control laws. My point is that nyc, with its gun control laws is still the safest in the country. Not the perception shares commonly on this site.

As has been pointed out in previous posts, that makes no sense in terms of what your trying to prove.

 

[gene]

I feel much safe walking around Boston, Dorchester, Lynn, Chelsea, Jamaca Plain, Roxbury, Lawrence, Fitchburg and Springfield at night to name a few knowing just how well we are doing with crime rates and stats here in Boston myself.

 

[CMF]

Thanks for the excellent analysis, Dick. The result seems to be in line with the finding in other studies that gun control/ownership does not correlate with violent crime.

I also wrote to Metro about this article. I'm not expecting to see any reply published, although they have been quite good about covering both sides of issues recently e.g. abortion, climate change.

I wish I could avoid this liberal rag, along with the others (Globe, NYT) but they are freaking everywhere and I'm a textaholic.

 

[KMaurer]

In addition to their use of rankings rather than actual levels and their unsupported assumption of causation and its direction, there's also a problem with the data itself. While firearm death data are readily available from the CDC (actually the National Center for Health Statistics and the National Center for Injury Prevention and Control within the CDC), there's no really reliable source for gun ownership data. (This is a much greater problem when the presumed causation runs from gun ownership towards violence than if it ran the other direction.) Few states require that ownership of all firearms be registered, and even that registration is widely believed to represent a significant undercount. Survey data, particularly data from various government surveys, tend to be systematically biased downward, with much larger downward biases in jurisdictions with strict firearm laws. Finally, the proxy variables often used for gun ownership (e.g. subscription rates to Guns and Ammo) tend to be laughable at best. As a result, anyone who purports to demonstrate statistical evidence of causality from gun ownership to gun violence has to be presumed to be either a charlatan or a fool in the absence of compelling evidence to the contrary.

 

[Bill Nance]

I’d like to share some thoughts on these numbers from the perspective of someone who’s made a good bit of his living over the last 40 or so years number-crunching large data sets for government and private clients.

...So, that’s what appears to be going on here – VPC is using a true relationship between gun ownership and accidental/self-inflicted gun deaths to claim (or at least imply) a false relationship between gun ownership and gun homicide. It is scientifically disingenuous, and they most certainly know it. (As an aside, the numbers also clearly show that the relationship between gun ownership and non-homicide gun deaths is very real and argue that we should all be careful about using and storing our guns properly to reduce the number of accidental and/or self-inflicted deaths due to improper use).

Excellent post.