Quo erat demonstrator

I am going to debunk some nonsense by a convservative sheep blogger called Gribbit. I have seen this story in several places while doing a google search, but I personally was introduced to it by a rant over at Scienceblogs. The basic idea is that illegal aliens kill more people through drunk driving and violent crime than the war in Iraq and Afghanistan kills soldiers. Therefore, illegal aliens are more lethal than the war. He uses a really sad attempt at statistical analysis to do prove his point.

According to statistics compiled by US Congressman Steve King (R-5th CD Iowa), 13 Americans are killed each day by uninsured drunk driving illegals 12 U.S. citizens die a violent death at the hands of murderous illegal aliens each day in crimes Do the math. [ (12 + 13) x 365 = 9125 ] That's more than nine thousand people killed every year in the United States by illegal aliens. By contrast, consider the death toll for US servicemen in Iraq and Afghanistan since the beginning of the war on terror, as reported last week by the Department of Defense: Total U.S. troop deaths in Iraq as of last week were reported at 2,863. Total U.S. troop deaths in Afghanistan, Pakistan and Uzbekistan during the five years of the Afghan campaign are currently at 289. Again, do the math. [ (2863 + 289) divided by five years of war = 630 ] So illegal aliens in the United States are more than fourteen times as lethal (14.48 actually) as a full scale armed conflict. Quod erat demonstrando.

First of all, he makes an error in determining the rate of soldier's deaths in Iraq and Afghanistan. The war in Iraq has been in effect for about 3.75 years rather than 5. It makes a difference. 2863 deaths over 3.75 years yields a resulting rate of about 763 deaths per year, rounding down. In Afghanistan, 289 deaths over 5 years yields a rate of about 58 deaths per year. Combined the average rate would be about 821 deaths per year rather than the 630 presented.

If we use percentages to level the playing field, we start to see the obvious fallacies. If we assume 13 drunk driving illegal alien fatalities per day is correct, we get an average of 4745 per year. In 1999 there were 187,000,000 licensed drivers in America. Although this number is most likely much larger now and doesn't include drivers who aren't licensed, we can use it as an estimate. 4745 out of a pool of 187,000,000 leaves us with a result of about .0025% of drivers are killed by drunk driving illegal aliens each year. 12 deaths per day due to violent crime is about 4380 out of a pool of about 300,000,000 potential victims (U.S. population) gives us an estimate that about .0014% of the American population is killed in violent crime in the U.S. each year. If one uses a more accurate average of 821 deaths per year in Iraq and Afghanistan out of a potential pool of 170,000 (150,000 in Iraq; 20,000 in Afghanistan) soldiers, that gives us a result of about .5% of the U.S. forces in Iraq and Afghanistan are killed each year. The risk, then, is about 100 times greater than the risk of being killed by an illegal alien drunk driving or a violent act by an illegal alien combined.

If you play with numbers, make sure you don't end up looking foolish. People who have poor reasoning skills should never cut and paste statistics or attempt to draw conclusions from raw data. One of two basic conclusions are almost always applicable whenever someone is misrepresenting statistical information to make a claim. They are either dishonest, or incompetent. And neither conclusion will lend legitimacy to any other argument you wish to make.

The Southern Fried Skeptic