Another Study PROVES guns are a problem.

Uncle Buck

New member
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2443681

Not sure how many have seen this study. I am still reading through it and looking to find holes and problems with it.

The first glaring problem I see is it uses data from a period of high crack use. In other words, I am wondering how many of the user of firearms were involved in illegal activity.

More to come. :)

As always, I would love your input on this also. A gun grabber friend sent it to me as proof that guns are bad.
 

44 AMP

Staff
First off, I'm not going to download and read the study. I did read the linked page.

Coming from people at Stanford and Johns Hopkins already taints it in my view, and the statements about how they used "plausible" models, leads me to think they made their conclusions, and did a study to back them up.

They do tell us right out that they are essentially using new parameters looking at previous studies, which also makes their conclusions suspect to me.

And in particular the line about how Right to Carry laws INCREASE rape and assault (but only by .1)seems in particular contrary to logic.

To my way of thinking, if legal, permitted CCW was in ANY way responsible for increasing crime, the increase would have to be those legal, permitted CCW holders, committing those crimes. And IF that were the case, those people would be getting arrested, tried and convicted, and that data would be available for "study".

Nowhere have I heard of that being the case. As a group, legal permitted CCW holders have a lower rate of committing crime than police officers do! (or so some study told me once...;))

In my not so humble opinion, the study is worth squat....
 
There's an old saying that may apply: "Figures never lie, but liars always figure."

Any time you see someone using old studies to justify new (and different) results, you know right away that they are manipulating the statistics to the point of torturing them. John Lott is a statistician. He generally doesn't torture statistics to make his points. Look at some of his work.
 

Uncle Buck

New member
From the "report" -
Table 2 shows far fewer statistically significant effects, but every one of which suggests RTC laws increase crime -- for rape, aggravated assault, robbery, auto theft, burglary, and larceny. There is not even a hint of any crime declines.

I kind of think you guys are right about this study, it is trash.

The FBI has shown DECREASES in the seven categories listed above. They have used county data, not the state data that is usually reported to the FBI.
I am only on page 28 of this report and they are still explaining why John Lotts models are no good. Torturous? An understatement to say the least.
 

rkittine

New member
The anti gun people read these things and want to believe them as it verifies there stand in their minds. We need more pro-active positive press and just pointing out to ourselves all the holes in these studies is not helping us in my opinion. What can we do to get our own information published and out to the masses? These people for sure are not going to read about it here or in Rifleman etc. (Are they changing that to Rifle Person now?)

Bob
 
The anti gun people read these things and want to believe them as it verifies there stand in their minds.

And the pro-gun folks do the same darned thing. The expression "Figures never lie, but liars always figure" applies to both sides of the gun debate. Both sides use the same expression to describe the results of the other. Funny how that works.

The biggest problem on BOTH sides is understanding the differences between correlation and causation.

Of interesting note of recent studies is the CDC's study that pro-gun people complained was a huge waste of money and worked to stop such funding of research. http://www.nydailynews.com/news/national/funds-studies-gun-violence-article-1.1809263

Actually, the results were not what was expected...
http://www.gunsandammo.com/politics/cdc-gun-research-backfires-on-obama/
 

Glenn E. Meyer

New member
To my way of thinking, if legal, permitted CCW was in ANY way responsible for increasing crime, the increase would have to be those legal, permitted CCW holders, committing those crimes. And IF that were the case, those people would be getting arrested, tried and convicted, and that data would be available for "study".

44 is on the money. These kind of studies do not look at causality in a manner acceptable to experimental studies.

If the permit holders committed the crimes, then you have case - do they? :rolleyes:
 
But ... but ... but ... EVERYONE knows that if more people are carrying GUNZ!!! there will be more GUNZ!!! around for bad guys to steal and to use in the commission of crimes.

It's just commonsense ™...
 

Evan Thomas

New member
And in particular the line about how Right to Carry laws INCREASE rape and assault (but only by .1)seems in particular contrary to logic.

Just for the record, a .10 "confidence level" doesn't mean that there was a .1 increase in rape and and assault. All it means is that the difference they measured had (according to the analysis they used) a 10% probability of being due to random variation. It says nothing about the actual size of the difference, except that it wasn't very large: in my day, anything less than a .05 confidence level wouldn't have been accepted as meaningful. (Note that a level of .10, while numerically larger than .05, is "less than" the latter in the sense that it reflects a smaller degree of confidence in the data.)

Just another reason to be dubious about this. If the confidence level had been only .15, I'm sure they would have reported that as showing support for their position. :rolleyes:
 

Skans

Active member
I've read enough in this thread to know the study is total crap. I have no intention of reading it or clicking on the website that posted it.
 

Uncle Buck

New member
C'mon Skans, you have to at least get through the first 48 pages with me. Then when you wake up, maybe you could finish the other half of the report.
 
To my way of thinking, if legal, permitted CCW was in ANY way responsible for increasing crime, the increase would have to be those legal, permitted CCW holders, committing those crimes.
And if that was happening, we'd be hearing all about it, all the time.

As it is, every time a CCW holder gets involved in a sketchy shoot or has so much as an ND in public, HuffPo and Mother Jones are all over it. If CCW holders were actually committing forcible felonies? That would be a gold mine for them.

The absence of any such incidents says enough. They can massage the numbers however they like. Heck, they've been doing it since CCW became common in the early 1990's. They've still yet to find that vein.
 

Merad

New member
All it means is that the difference they measured had (according to the analysis they used) a 10% probability of being due to random variation. It says nothing about the actual size of the difference, except that it wasn't very large: in my day, anything less than a .05 confidence level wouldn't have been accepted as meaningful.

.05 is still taught as the threshold for a meaningful result, at least in the last stats class I took (2012).

I skimmed the paper a bit a few days ago but couldn't motivate myself to read it in detail. Did you notice if they offer any justification for claiming conclusive results with a larger than normal confidence level?
 

Evan Thomas

New member
.05 is still taught as the threshold for a meaningful result, at least in the last stats class I took (2012).
That's reassuring. I don't recall that they offered any justification for using that confidence level; but I've seen similar "analysis" in other articles in which it seemed that the authors were, um, invested in a particular result.

It's also a trend which I suspect is made easier by changes in statistical analysis (and its increasing computerization): the software spits out a probability, instead of giving a yes/no answer as to whether a pre-selected significance level has been reached. In the good old days, you had to go to the bother of increasing your sample size to force a significant result. :rolleyes:
 

FlyFish

New member
There's no "bright line" that defines statistical significance. If something is "significant" at an alpha of .10, it means (really oversimplifying here) that there's a 10% probability the observed results could have occurred just by chance, which is the same as saying that if you accept that the results are "real" you have a 10% chance of making a Type 1 error, which is an incorrect rejection of a true null hypothesis (i.e., there's a 10% chance you're wrong). If the results are significant at .05 and you accept them, your probability of a Type 1 error is only 5%.

So far so good, but as you "insist" on a smaller and smaller alpha before accepting results to be real, your probability of a Type 2 error - incorrectly accepting a false null hypothesis, which means incorrectly rejecting "real" results, becomes larger and larger. The commonly accepted alpha of .05 is really just a convention that many researchers (my understanding is that it originally came out of drug trial studies) agree represents a reasonable balance between the chances of making a Type 1 vs. Type 2 error.

The problem is that, depending on the study in question, the consequences of making one type of error vs. the other are not the same. If the nature of the study is such that making a Type 1 error would be very problematical but a Type 2 would not, then it's appropriate to set a very small alpha, but the reverse is also true. In my work (biostatistics), I've conducted studies in which we stipulated an alpha of .01 before accepting results as "significant" and also studies where the alpha was more appropriately set a .40.

To return to the case in question, the authors of the study are, in effect, saying "We're pretty sure these results are real ["real" meaning they reflect an actual effect and didn't occur by chance], but there's a 10% chance we're wrong." Based on that, readers are free to agree with the authors or not, as they choose.
 

44 AMP

Staff
Not being a statistician, or anything remotely close, I'll just go with my gut feeling on this, that when they make statements like this...

RTC laws increase crime -- for rape, aggravated assault, robbery, auto theft, burglary, and larceny.

I think there is more than a 10% chance they are wrong....
;)
 

Ike666

New member
Just a charming historical note. An alpha level of .05 came about courtesy of Sir Ronald Fisher. At the time he was the QC statistician for the Guinness brewery and determined that 1 bad bottle in 20 meant the process was nominal.

These days real scientist use Bayesian analyses and dump all that NHST dreck.
 

barnbwt

New member
"Coming from people at Stanford and Johns Hopkins already taints it in my view, and the statements about how they used "plausible" models, leads me to think they made their conclusions, and did a study to back them up."
I really hate that we so often utilize source bias in this area, but not nearly so much as I hate that said rejection is practically always clearly justified. It ticks me off that almost all discussion outside the reviled "echo chamber" really is riddled with all manner of misinformation, to the point of being nearly useless for the exchange of ideas. Like a "last sane man" scenario :mad:

Because guns have been purged from education for generations, they have basically become something of a lost technology for large swathes of Americans. Whatever moral implications carried with the guns became forgotten covenants. The result is that the public discourse rapidly devolves into how many guns equal the weight of a duck, and whether that determines if they should float. Sometimes very technical and erudite discussions, from accredited academic institutions, no less.

How do you rebut that? How do you even engage it? I must say, the most disturbing and damning aspect of these types of studies is that they show how the role of law has become distorted in the modern era; no longer rooted in moral authority (i.e. punishing that which is wrong or harmful to others) it is guided almost solely by the ethereal faith of "utility." That's how you end up with gems like this from supposed intellectuals;

"Across the basic seven Index I crime categories, the strongest evidence of a statistically significant effect would be for aggravated assault, with 11 of 28 estimates suggesting that RTC laws increase this crime at the .10 confidence level"
"Our analysis of admittedly imperfect gun aggravated assaults provides suggestive evidence that RTC laws may be associated with large increases in this crime, perhaps increasing such gun assaults by almost 33 percent."

*cough* What? One third of your estimates suggest AA increases at a 10% confidence (even in High school AP, it was understood that anything over 5% was wild speculation, with 1% being preferred for supporting anything close to 'incredible claims' like that mere passage of laws will increase aggravated assaults with a gun 33%). I would also argue that the fact 100 pages of analysis was required for such broad and simple conclusions is also strong evidence of a very convoluted analysis concealing misrepresentation or outright manipulation of data. There should be a number of very clear and simple data correlations if carry laws are increasing assaults with guns by 33% :rolleyes:

No mention either of the attempt to untangle correlation/causation; just the assumption that laws would have an effect on the crime rates. That's a pretty big assumption if you think about it (why do specific laws get enacted when they do, again...when a relevant class of crime has spiked up and gotten attention)

TCB
 
Last edited:
Top