Wednesday, July 8, 2009

"Guns are Bad News for Women?" Huh ... - Part II

In Part I, I introduce mikeb's hypothesis that "guns are bad news for women" Let me now get into some statistics. Hemenway's table that mikeb provides has absolute numbers High Gun States and Low Gun States. Since the numbers are supposedly from 2003 and I don't know what criteria for high gun states or low gun states was used, I set about to recreate the table using rates (per 100000) so that numbers could be compared independent of population.

First, all homicide, suicide, and accident information was taken for the year 2006 using the CDC WISQARS database. Only fatalities were examined (I might do accidents another time). I included all homicide and legal intervention (cops shooting bad guys) in with homicide. Second, for the firearm ownership rates, I used the 2001 Behavioral Risk Factor Surveillance System in North Carolina which surveyed more than 200,000 people nationwide. One thing to note about gun ownership, there is not an accurate survey out there. All of them are based on people reporting, and people reporting have varying reasons why they would not answer truthfully. For example, some may say they have guns when they don't because they are suspicious of someone scoping out their place, some may say they don't have guns when they do because they are suspicious of someone scoping out their place, some may say they don't have them when they do because they are criminals, some may say they don't have them and they do because it is illegal to own what they have (i.e. cities with gun bans), and on and on and on. At best, the numbers from this survey are an indication of the legal firearms ownership only. So, from this we should realize that my data (and Hemenway) is tainted with inaccuracy from the start.

With this data I sorted it by gun ownership and came up with a list of High Gun States and Low Gun States that is similar to the one mikeb presents. So I know I am on the right track. The one difference is that District of Columbia is not included in the Low Gun States list. This is usually the case for many studies that are trying to show that guns are bad. DC has the lowest gun ownership rate in the nation, yet it is almost always on the high end with most crime statistics (obviously the criminals didn't realize they aren't suppose to have those guns). In this instance, omitting DC doesn't hurt the cause much since we are looking at guns and women (men by far commit more murders, have more suicides, and more accidents than women). The population of the samples is still about even (High - 10,866,675 women; Low - 11,032,622 women). Here is the table of the rates with and without DC.

2006 Death Rate of Women (per 100,000)

High Gun States Low Gun States Low Gun States (no DC)
Firearm homicide 1.67 0.64 0.60
Other homicide 1.46 1.04 1.00
Total 3.13 1.69 1.59

Firearm suicides 2.16 0.28 0.29
Other suicides 2.92 2.95 2.98
Total 5.08 3.24 3.27

Firearm accidents 0.93 0.03 0.03
Other accidents 37.87 23.18 23.11
Total 38.8 23.2 23.14

A couple of items for thought. First, in this instance why bother excluding DC? The rates come out virtually the same. Second, why did Hemenway/mikeb not list the Other Accident deaths? I have three theories on this. 1) Because accidental deaths by firearms are so small (especially when looking only at women) that statistically it is a non-issue (however, I would argue that we could reduce those accidents through more training - on the other hand, you can't eliminate them, people still have the right to be boneheads). Putting the non-firearm accidents would detract from the conclusion that Hemenway is trying to get you formulate. 2) Pulling this number from WISQARS requires a calculation and whoever gathered Hemenway's numbers didn't realize this. You can't do an autosort for non-firearm accidents, it is an undefined category. So I grabbed the total accidents and subtracted the firearm accidents and presto, you have the non-firearm accidents. 3) Whoever compiled the information was sloppy. This could have been a completely unintentional oversight. However, if it was, it was sloppy. Whenever you are comparing numbers or categories, you want to make sure that you have apples to apples comparison. In this case, ommitting the non accidental firearm deaths might lead someone to believe that you are hiding something.

So, how can these numbers be explained? Once again, if you believe that guns cause each of these (see Part I), then you're done. Go write a book and start getting grants from the Joyce Foundation, they are always looking for new shills. If, however, you actually care about reducing each of these, then we need to look at each individually. Remember, the hypothesis that mikeb put forward is that guns are bad for women and that guns in the home cause more violence towards women.

Starting with accidents, we can pretty much discount the second part of the hypothesis since accidents cannot be considered violence. Its unintentional. No one was trying to harm anyone. As for the first part, well, this is actually expected. More guns without the proper training means more accidents. Not only that, but more handling of guns (without training) means more accidents. But it won't always be documented as such. For instance, I would be willing to bet that if you armed every single person in DC with guns (that were illegal to own previous to Heller) you wouldn't have their firearm accident rate change one bit (0 per 100,000). Why? Because if someone got shot, how would it be reported? If it is reported as an accident then the person who had the gun is now in trouble with the law for having an "illegal" gun. If it is reported as an assault, it becomes another unsolved crime in a city full of violence. Meanwhile in North Dakota, where they have lots and lots of guns and handle them regularly (i.e. they are not hidden away in a drawer in case someone breaks into their inner city apartment - never seeing any range time), the accidental firearm death rate is also 0.

Next we have homicides. Relatively few guns bought at pawn shops, retail stores, or gun shows are used in crimes. The Bureau of Justice Statistics estimates that 80% of guns used in crimes are from family, friends, a street buy, or illegal sources. This jives nicely with the FBI statement that 80% of crime is gang related. All of these guns couldn't be guns in the home (and the statistics above would indicate that as much as 80% weren't). So the second premise is shakey again, up to 80% of the women killed are not killed by legal guns in the home (which is what the high and low states are categorized by). I'll discuss the first premise later.

Finally, suicides. This can be an emotional subject for some. Most people know someone who has committed suicide. However, whether a person shoots themself in the head or slits their own throat, the suicide is no less horrific. As with homicides, the CDC does not record where the firearm came from that was used. However, we can probably safely assume that people with access to firearms with choose firearms in a higher proportion than people who don't have access to firearms (that is common sense). Some people really want to kill themselves. Some people just want to cry out for help. Those who want to kill themselves will choose the most lethal method (guns, poison, jumping off of high buildings), those who want to cry out for help choose less lethal means (overdose, self asphyxiation, slitting wrists). To examine suicide and firearms we will have to look at our dataset further.

So up to this point we have two parts of the hypothesis and three types of death - a total of six items to evaluate. I have already shown that it is of no use to evaluate accidents for either one of the premises and that the second premise is faulty for homicides. That leaves three items to explore:
1) Are guns bad for women with respect to homicides?
2) Are guns bad for women with respect to suicides?
3) Do guns in the home cause more lethal suicides?

Tune in to Part III for more analysis.


  1. Dear Reputo, After reading your analysis thoroughly, I answer yes to all three questions. Whose side are you on anyway?

  2. Bet you won't Dare answer WHY you think that way, Mikey!

  3. Mikeb,

    I haven't answered them yet. Your answer is simple in that you assume the cause of homicide and suicide is guns - I have already given enough evidence that the cause is primarily gangs and depression respectively. In my next part I will show that your hypothesis more guns = more death is a false correlation.

  4. It will never cease to amaze me how supposedly grown, adult individuals can supposedly believe that an inanimate, non-sentient lump of metal can "cause" anything. That someone believes such a thing is strange, to say the least. That someone believes it, and thinks that it should be a basis for abridging and infringing on the rights of the law-abiding... that is simply intolerable.

    Oh well. As is usual, the pro-rights side of the argument has provided all manner of information, facts, and figures to support its stance... while the anti-rights side of the argument relies on sweeping generalities and attempted guilt trips.

    Keep up the good work, Reputo. It will not matter to MikeB, but it might help other people.

  5. All,

    In going through my dataset again as part of writing Part III, I found that I had messed up some of the numbers for Wyoming and Wisconsin (I accidentally copied the male numbers to the female column). Since males cause far more death than females (I can blame males since they are not inanimate - anyone for banning men in general?) and Wyoming was part of the High Gun States, I had to update the table some.