Regarding the GD post about maps debunking the NRA

[ManiacJoe]

After reading the study published in JAMA:
http://archinte.jamanetwork.com/data/Journals/INTEMED/926956/iiq130121_732_740.pdf

linked from here:
http://www.policymic.com/articles/74055/these-maps-debunk-everything-the-nra-has-told-us-about-guns

and previously posted to GD...

... I was looking at the data relating the legislative scores to the rates for suicide and homicide and combined. The correlations are not that surprising, when any exist...


Sorting by "overall/combined" death rate
- The states with the six lowest rates have very high legislative scores.
- No other correlation exists between the leg scores and the overall rates.


Sorting by suicides
- The nine lowest rates have very high leg scores.
- The five highest rates have low leg scores.
- No correlation exists for the middle rates and their scores.


Sorting by homicides
- The two highest rates have low leg scores.
- The three states with rates too low to be measured have low leg scores.
- No correlation exists for the rest of the homicide rates and scores.

[gcomeau]

1. Do you understand...

...that the graphs on page 6 of that report you just linked end any argument over whether there is correlation or not, seeing as those graphs are how statistical correlations are defined?

Now you can make the correlation vs. causation argument... but there is no argument over correlation. It's there. In black in white.

[gejohnston]

2. pretty misleading

reminds me of a Mark Twain quote
It lists Wyoming as having 348 gun deaths. In 2010 Wyoming had eight murders, five with guns. Accidents probably account for no more than ten, if that. That leaves the rest as suicide. All rural areas have higher suicide rates, that is equally true in gun free Japan and South Korea. The only change would be suicide by gun, but not the suicide rate. This one is more interesting when using the sort function.
http://en.wikipedia.org/wiki/Gun_violence_in_the_United_States_by_state

[gcomeau]

3. It breaks them down in Table 2

So no... not misleading. Unless you don't, you know, read the report.

[gejohnston]

6. that was before I read the report

the most important thing noted. The table doesn't really help since the numbers don't really match the other link.

Conclusions and Relevance: A higher number of
firearm laws in a state are associated with a lower rate
of firearm fatalities in the state, overall and for suicides
and homicides individually.

They did show that fewer firearms suicide, but not suicide in general.

As our study could not
determine cause-and-effect relationships, further
studies are necessary to define the nature of this
association.

Basically, he is admitting it is an exercise in post hoc ergo propter hoc. The only way to attempt do a such a study is to do a before and after study of the same place: Did murder and suicide rates drop in the same place after passage of the law? If so, can it be traced to the law itself? For example the US murder rate climbed after passage of the 1968 Gun Control Act. Was that the cause? No. Did gun ownership increase during that time? Yes. Did the crime rate rise because of the increase in gun sales? No, gun ownership increased because of fear of crime.

Is that relevant today? no, since sales and ownership is increasing while violent crimes are dropping.
Basically, it is an MD pretending to be scientist wanting to be published. I would like to read critiques of the study by criminologists and other social scientists.

[hack89]

4. Would the results be different if levels of poverty were compared to gun deaths?

levels of non-gun violent crime? Rural vs non-rural (for suicides)?

The point is that gun violence is a complex phenomena with numerous causes - until all are compared side by side and all the relationships between them fully explored, to simply declare that gun laws are the only reason is premature.

[gcomeau]

5. Probably. So?

But I refer you again to the difference between arguing causation and arguing correlation.

You were arguing about the existence of correlations in the data. The correlations are simply there, and there's and no two ways about that.

"to simply declare that gun laws are the only reason is premature."

That declaration was not made, so not sure why you're feeling compelled to point this out?

[hack89]

7. "maps debunking the NRA"

a little premature to say that considering we don't understand just how strong that correlation is compared to other possible causes.

I have no problem with the maps and the study. It was the politicized titled place on it that got my attention - it is fundamentally and intellectually dishonest in the context of gun control. Just another anti-gunner stirring shit in my opinion.

[ManiacJoe]

8. I do understand.

Do you understand... ...that the graphs on page 6 of that report you just linked end any argument over whether there is correlation or not, seeing as those graphs are how statistical correlations are defined?

Unfortunately for the researchers, their published data does not support their claims. Neither when viewed via Table 2, nor when viewed in the graphs in Figure 2.

The "household ownership" graph does show some correlation with the combined death rates, which makes sense. The two leg score graphs and their data table show no correlation.

[beevul]

9. Ahh yes, rates.

Where have we seen that tact before...

[ManiacJoe]

10. Care to be more verbose?

You might be on to something, or not.

[beevul]

11. On occasion, folks will post about "rates", including certain implications.

On occasion, folks will post about "rates", including certain implications.

Chief among them, is the implication that a place with a numerically higher "rate" of a given thing, has a bigger problem where that thing is concerned.

While that can be the case, there are numerous occasions where it is not.

[ManiacJoe]

12. Well, in this case the higher rates would be a bad thing.