Thursday, July 9, 2009

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

Parts I, II, III, and IV are at those links. By now I am sick of recapping them. Read them to get up to speed. Not to beat a dead horse, but I wasn't satisfied with the graph that showed all of the data points and no correlation. Hemenway's table that mikeb reproduced which I made less sensational (by putting it in a rate term and adding the DC data) compared the highest gun states with the lowest gun states. So maybe there is a general trend if we look at several states together.

Naturally, I created another table to show this. States are sorted by ownership rate and then 10 states are in each group (except the middle which has 11).

2006 Death Rate of Women (per 100,000)

Homicide Suicide Accident
Low Low 10 0.91 0.55 0.02
Low 10 1.24 1.73 0.04
Mid 11 1.42 1.73 0.04
High 10 1.61 2.04 0.08
High High 10 1.87 2.18 0.24

What we can see is that in general states that have more guns have more gun violence against women. One curious thing though is that the suicide rate jumps dramatically from the low-low states to the low states and then remains stable through the mid states and rising again for the high and high-high states. Although, it should be noted that the change from the low to the high-high is not near as dramatic as the change from the low-low to the low. If Hemenway and mikeb really wanted to prove their point, this is the table they should have used. Does this mean that guns are the cause of violence against women? Not at all. Remember guns are an inanimate object, if they were the cause of violence we would expect to see the same trend with men. Below is the same table except with men instead of women.

2006 Death Rate of Men (per 100,000)

Homicide Suicide Accident
Low Low 10 8.14 5.3 0.25
Low 10 7.35 12.02 0.31
Mid 11 7.97 6.4 0.24
High 10 7.20 13.33 0.7
High High 10 8.23 16.21 0.95

Well, the magnitude difference is readily apparent. Men get killed by guns 7 to 10 times more often than women in all categories. So, if we are going to say this data proves that guns are bad news for women, then we would have to say that guns are extra super duper bad news for men. More on that later. Notice how the low-low gun states have a nearly identical firearm homicide rate as the high-high gun states. The suicide rates are not consistent either. Let's rearrange the data one other way to see if we get the same result. This time, I grouped the states into each 10% of gun ownership (<10%,10-20%, etc).
2006 Death Rate of Women (per 100,000)

Homicide Suicide Accident
<10% 1.06 0.32 0
10-20% 0.60 0.28 0.01
20-30% 1.23 1.06 0.03
30-40% 1.28 1.65 0.03
40-50% 1.53 1.96 0.07
>50% 1.87 2.18 0.24

Except for the initial group (which only includes DC and HI) we see the same trend. So lets look at the men:

2006 Death Rate of Men (per 100,000)

Homicide Suicide Accident
<10% 13.12 3.04 0.54
10-20% 5.17 3.89 0.14
20-30% 9.43 7.7 0.3
30-40% 7.22 11.42 0.34
40-50% 7.36 12.57 0.55
>50% 8.23 16.21 0.95

Well, homicides and accidents don't play like the women's. I will admit, that if I hadn't looked at the information in Parts III and Part IV, then I might be inclined to believe the canard that more guns "causes" more gun violence to women. But, we have shown that as we add more datapoints that isn't the case. To recap:
2 datapoints - some evidence
5 datapoints - better evidence
6 datapoints - some contrary evidence
17 datapoints - a lot of contrary evidence
51 datapoints - clearly a lack of correlation (therefore causation is impossible)

So could there be any other factors that might give us more insight? There might be, remember what I was saying about regional differences before? Well it turns out that 7 of the 10 low-low states are in the Northeast. 6 of the 10 high-high states are in the West. 5 of the 10 high states are in the South.

In Part VI I'll look at regional data.

5 comments:

  1. Great Work Reputo,

    Absolutely fabulous, great deconstruction of canard.

    Were you planning on doing a post on things that have better correlation, such as education, poverty, race to firearm related deaths than ownership?

    ReplyDelete
  2. Reputo, You are the table master.

    ReplyDelete
  3. Bob S.

    Absolutely (some things that have better correlation like you mention and others that have no bearing like cheese production)

    ReplyDelete
  4. mikeb if you like these tables read my post on spreadsheets

    http://myreputo.blogspot.com/2009/02/spreadsheets-engineers-best-friend.html

    ReplyDelete
  5. MikeB,

    I'm sure that Reputo appreciates the compliment, but would - Like I, to hear your take on the INFORMATION and DATA presented....and how it contradicts the more guns more deaths meme you presented.

    Well?

    ReplyDelete