Article claims "States with higher gun ownership have higher murder rates"

[Sharps-shooter]

so, the purpose was essentially to establish a correlation, and it appears from the presented data that there is a correlation between gun ownership and murder rates. What we don't know includes:

What role does an atmosphere of violence play in people's decision to get guns? (are people going out and buying guns because they live in dangerous areas?)

How does the murder rate change proportionally to changes in the rate of firearm ownership in a given state? (do they both increase together? or does one go up and the other down?)

and, in my professional opinion as a published research scientist , most importantly:

How reliable is the original data? In other words, can the survey of firearm ownership be relied upon?

Now, I'm pretty private about my firearm ownership. I wouldn't respond to it on a survey, especially not one in which they called my number and therefore could be assumed to have my name. I don't do firearm registration, and I go out of my way to avoid IBC's. And I'm a respectable upstanding member of society.
MUCH LESS would I answer this question in the afirmative if I were not lawfully allowed to own a firearm, or were in a situation where I felt there was a likelihood I would be committing a crime with one in the near future.

In the scientific community, surveys, and especially phone-iin surveys, are considered one of the least reliable forms of gathering data, because you can't really verify the truth value of the data so gathered.

I think that a realistic determination of the rate of firearm ownership, ESPECIALLY among the criminal element in society, would be extremely hard to determine.

 

[mountainclmbr]

Why have they not looked at the number killed (murdered) by their own governments in the last 100 years? The leftists have killed some 150 million or so. The answer is that if the gun banners asked this question, they would probably be looking at their own faces in the mirror.

 

[Sharps-shooter]

It's beyond the scope of the study to look at something like that. There's a huge difference between "correlation between murder rates and gun ownership" and "Ugh! Guns Bad!", but unfortunately the media is going to report this as "ugh! Guns bad!", and that's what the public will think the scientists said in the first place. Which may well have been the funding agencies' plan from the outset.

 

[wuchak]

Looks like the CDC is doing the same thing Congress spanked it for before, using taxpayer money to promote an anti-gun agenda. With Dems taking control it's unlikly the CDC has anything to fear now.

The Dems will use this in conjunction with the recent conviently timed school shootings to push for major new firearms restrictions. Most people hear Harvard and CDC and think they are credible sources just like they have with the BS reported in the New England Journal of Medicine.

we are all doomed.

 

[Heywood Case]

Sharps-shooter,

I'd say from their publication record that these particular scientists are absolutely saying "Ugh! Guns Bad!". As to the reliability of survey data, there are certainly problems with any survey let alone asking questions about whether there is a gun in a household. But it is still the only way to get many kinds of information in a way that allows inferences to be made about a population. In this case the survey itself is well executed but we can expect that gun ownership was most likely under-reported to a significant degree. Whether that makes a difference to the conclusions is another question (and it might very well not); in my opinion it is incumbent upon users of these data (i.e., Miller et al) to demonstrate that the measure is reliable and valid for their application (to my knowledge, they have not).

Before dismissing all surveys out of hand, remember that survey results are also used to show the value of firearms ownership (e.g., Gary Kleck, John Lott).

Darth Muffin, firearm ownership in the study was aggregated by state and is used as a percentage of households - a rate, not the raw number.

I still think there are serious questions about the reality of the relationship presented in this study, partly because of my previously mentioned argument that the confidence intervals look too small for an analysis of 50 cases, and partly because the aggregation by state paints with too broad a brush (many here have argued convincingly on this basis). There are just too many differences within states to make this a compelling analysis, and their own results show that urbanization is a significant factor in the model of homicide rates. What if you have an urban population with low ownership rates and high firearm-related homicide rates, and a rural population with high ownership rates and low homicide rates? Combing this into one number for a state is not going to tell you anything useful.

Anyway, I don't think there's much more I can say on this one. Sorry to be beating a dead horse so long but I thought as long as I have the article I might as well address some of the issues that came up.

John

 

[Sharps-shooter]

Guns Bad!

Quite right that the survey may be the only available means of getting the data. And great point about the differences within states.

To me, this looks like the sort of study that someone would use to get the ball rolling on a larger project. Funding agencies don't like to give out money to start projects-- they much prefer to contribute to already-successful ongoing research in an area. THis study looks like it could have been done without a great deal of yime, effort, or money (as compared to what would be required for testing the effectiveness of a treatment, for instance).

So, it generates a high level of publicity for a low level of initial input. I haven't read anything by these scientists previously, but It will be interesting to see what they do next.

 

[Grey54956]

Hello, what's this? I believe I have a portion of the data that was used in this article. To say that the article is misleading is an understatement. The authors are either completely incompetent in statistics or are shamelessly misrepresenting any analysis in their conclusions.

I can find no significant correlation between the rate of gun ownership and the homicide rate. If anything, there is some evidence that the rate of gun ownership actually has a slight benefit to the homicide rate, but again, this would not be statistically significant, i.e. valid.

More on this after dinner...

 

[SDC]

John Lott has also taken this "study" apart on his website ( http://johnrlott.tripod.com/ ), and the only way these weasels were able to arrive at these laughable conclusions were by completely ELIMINATING the District of Columbia from their data, and by picking and choosing different years that best fit their "desired result". When you use a consistent data point/year, OR you put DC back into the mix, it completely REVERSES their findings.

 

[Gene_WI]

Counter study

The NRA should come out with counter study.

The correlation of NRA membership and gun crime through the states. It would be interesting to see the results.

IS NRA membership data public? If it is the above is a very simple task.

 

[Justin]

John-

Wow! Thank you for your posts!

 

[Grey54956]

Sorry about the delay. Hamburgers were tasty and required sufficient time to prepare and enjoy. Back to my findings.

The authors of the study want to draw a correlation between the rate of gun ownership and the homicide rate. Regardless of correlation not being equal to causation, correlation can reveal a potential cause. The authors do this by presenting one set of data and falsely equating it with something else.

To get the study to come out the way they wanted, they used three pieces of information: state population, gun ownership rate, and homicide rate. Then they engaged in a little slight of hand: they multiplied the population with the gun ownership rate. This should create an rough number of armed citizens in a state. Then the authors grouped the states by quartiles based on the calculated number of armed citizens. The subgroup averages were then compared to one another and they show that states with fewer guns overall have a lower homicide rate that those with more guns.

The problem with this is that they aren't actually correlating the gun ownership rate with homicides, but rather a calculated number of guns to the homicide rate. The problem is that the effect of population level is wrapped up in that. California which has a moderate gun ownership rate has 35 million people, so the number of guns in the state is high. Montana which has a high gun ownership rate has less than a million people, has far fewer guns than California. Because of this, California would be in the highest quartile in their analysis, while Montana would be in the lowest. In this case, population has far more bearing on the homicide rate than gun ownership rates.

If you were to take the gun ownership rates, and divide them into quartiles and compare them, you find that each subgroup has roughly the same average homicide rate, and roughly the same deviation from average. You can knock DC and Loiusiana from the mix as outliers, as these probably have some extenuating ciircumstances.

The article is horribly flawed and misleading. If this were found in an engineering trade magazine, the authors would be shunned for life.

 

[YK]

There are lies, damn lies and statistics.

John Rogers, thanks for your posts. Regarding confidence intervals, I thought that you can get a pretty tight ones even with a small sample size if the absolute difference between the groups is large. I may be wrong though, a study of statistics is a lost war for me.

It would be interesting to know if data analysts were blinded.

As for many above, cause/effect issues, aggregated vs. nested design, and various potential recall/report biases and flaws (see post above) don't sit well with me.

Finally, there is no attempt to look at the risk/benefit ratio, say rates of criminal firearm homicide vs. rates of crime prevention/self-defense using firearms. Would've been easy, just add one question to a survey...

 

[Desertdog]

Finally, there is no attempt to look at the risk/benefit ratio, say rates of criminal firearm homicide vs. rates of crime prevention/self-defense using firearms. Would've been easy, just add one question to a survey...

That was not the reason for the survey. If they had done a true survey they probably would come up with the truth. When they are looking for data to support their conclusion, they make it come out they want it.

It reminds me of the scienst that taught a frog to jump on demand. Then he cut off a front leg and said JUMP. The frog jumped.

Then he cut off the other front leg and said JUMP. The frog jumped.

Then he cut off a hind leg and said JUMP. The frog jumped.

Then he cut the second hind leg and said JUMP. The frog just sat there.

Conclusion: Cut off all 4 legs of a frog and they go DEAF.

 

[v65magnafan]

Part of Response by John Lott

Problems with the latest Gun study [Lott]

--------------------------------------------------------------------------------

http://johnrlott.tripod.com/

Problems with the latest Miller, Hemenway, Azrael
study on guns

The New York Times reports yesterday that a new study
from Miller, Hemenway, and Azrael claims: "States with
the greatest number of guns in the home also have the
highest rates of homicide, a new study finds. . . . "

Well, I have just spent a short time looking at the
study, but there are some of things that are pretty
obvious: 1) They excluded the District of Columbia
without any explanation, 2) they use other crime rates
to explain the homicide rate (by the way, they don’t
use anything like an arrest or conviction rate,
nothing to do with law enforcement), 3) they use
purely cross-sectional data that never allows one to
properly control for what may cause differences in
crime rates, and 4) data from different years is used
without any explanation (for the sake of argument I
will use what they did, but it is weird to have the
unemployment rate from 2000 to explain the homicide
rate from 2001 to 2003, etc.). The data for a panel
test on this is readily available from the sources
used in their paper, though I have only collected the
data to redo the estimates for 2001 that they use (why
is it that these papers where one can put together the
data in an afternoon get any serious attention). Why
they only looked at the CDC data for 2001 when it is
available for many other years is a bit of a puzzle.
Since Miller and Hemenway have refused in the past to
let me look at their data, I didn't bother this time
and simply put the data together myself.

The bottom line is that their results comes from two
factors: the exclusion of DC and the use of other
crime rates to explain the murder rate. Changing these
two facts causes their result to go from positive and
significant to negative and significant. I also
decided to run these regressions on the robbery rate
and doing so produced a statistically significant
negative effect whether or not DC was excluded. Using
arrest rate data, not shown, also caused the results
to be more significantly negative. If I had the
necessary panel data handy, my strong presumption is
that would also reverse with their result whether or
not DC was included.

It is problematic to include the other crime rates in
these regressions, particularly since they must
believe that guns cause robbery as well as homicide.
The results below indicate that more guns mean fewer
robberies (again this is using their flawed set up,
though I believe that this would continue to be
observed with panel data).

In any case, noting that this is purely
cross-sectional data and not very useful, here is an
attempt to redo their estimates looking at the
homicide rate from 2001 to 2003 on the gun ownership
rate from the CDC and the other variables that they
use (I wasn't able to find their gini coefficient, but
that is the only variable that they used that wasn't
included):

[lots of statistical stuff snipped]

What it means. Again, this uses purely cross-sectional
data, but accepting that: their result depends on
excluding DC and including other crime rates to
explain the murder rate. This would mean that more
guns, less homicide. Even when DC is excluded, the
simple correlation using cross-sectional data is
negative, though not at all statistically significant.

[more statistical stuff]

For Robbery whether you included DC or not there is a
statistically significant negative relationship
between the CDC's measure of gun ownership in 2001 and
robbery rates in that year.

posted by John Lott 1/24/2007 at 3:10 AM

 

[langenc]

I did not read all posts.

That statement is FALSE:

Compare Virginia with DC (not a state)-then with any other states nearby.
I know VA vs DC the statement is FALSE.
In fact it is more dangerous in DC than Baghdad.