I make no secret of the fact that I find Ann Coulter amusing. I sometimes think her sarcastic tone may create more problems than a more reasoned argument would, but then again that is her style, and, let's face it, over the top, extreme arguments sell books and talk show air time, and allows her to enjoy fame and fortune a more reasonable writer would not. So I can't fault her for her choices, they obviously worked out very well for her.

However, looking at her most recent article, I have to say I am glad the LSAT doesn't have a math section or Ms. Coulter would probably be in a different field.

How so? Let me reproduce all the relevant paragraphs:

Let's compare murders by veterans to murders by other 18- to 35-year-olds in the U.S. population at large. From 1976 to 2005, 18- to 24-year-olds -- both male and more gentle females -- committed homicide at a rate of 29.9 per 100,000. Twenty-five- to 35-year-olds committed homicides at a rate of 15.8 per 100,000.

Since 9/11, about 1.6 million troops have served in either Iraq or Afghanistan. That makes the homicide rate among veterans of these wars 7.6 per 100,000 -- or about one-third the homicide rate for their age group (18 to 35) in the general population of both sexes.

But fewer than 200,000 of the 1.6 million troops who served in Iraq and Afghanistan have been women, and the murder rate for the general population includes both males and females. Inasmuch as males commit nearly 90 percent of all murders, the rate for males in those age groups is probably nearly double the male/female combined rates, which translates to about 30 to 55 murderers per 100,000 males aged 18 to 35.

So comparing the veterans' rate of murder to only their male counterparts in the general population, we see that Iraq and Afghanistan veterans are about 10 times less likely to commit a murder than non-veterans of those wars.

Does anyone else get a sneaking suspicion the math is a little... well, impressionist?

Just to point out one example, the says that the military rate of 7.6 is "about one third" of the civilian rate of 15.8. As anyone can tell, it is a whole lot closer to one half than one third. The again the rest of her numbers also seem to be inflated in her favor. But before anyone conclude that she is puffing up numbers to help support her argument, let me point out that the inflation is pretty mild, and generally is relatively close, just not close enough to be accurate. In other words, it doesn't sound like someone trying to jigger numbers in their favor, but more like the vague approximations favored by those who can't quite make the numbers add up.

For those who lack my compulsion to always check calculations in my head, allow me to demonstrate what I am saying.

The single sentence which holds the key to my objection is this one.

Inasmuch as males commit nearly 90 percent of all murders, the rate for males in those age groups is probably nearly double the male/female combined rates, which translates to about 30 to 55 murderers per 100,000 males aged 18 to 35.

Now think about that. We know the mixed-sex rate for the age group is 15.8 per 100,000. If males commit 90% of murders, that means out of 100,000 people, males will commit 14.2, while making up slightly less than half of the group. So the male murder rate would actually be about 28.4 or slightly more per 100,000. And while this is close to "about 30", it is a far cry from 55.

And so, the military homicide rate of  7.6 per 100,000 is something short of one quarter of the civilian male rate (~28.4), not one tenth as Ms. Coulter argues. Even if there were 55 murders as her bad math alleges, 7.6 is not one tenth of 55, but very close to one seventh.

There is another problem with her argument, that may actually make the numbers work in her favor. The military is skewed much younger than the general population in the 18 to 35 group. While the general population is distributed pretty evenly, the military has a huge preponderance of people from 18 to 22, as many who go into the military serve only one four year stint, making it more fair to compare those in the military with 18 to 22 year olds, rather than 18 to 35 year olds.

In addition, one should also consider the racial, economic and educational makeup of the military. While the military does not fit the stereotype the liberals promote of poor, uneducated minority kids, it is true that most enlisted men do not have college degrees, which means the bulk of those in the  military lack college degrees, which likely would increase the number of murders among comparable civilians, murders becoming less frequent in any demographic group as education increases.

Then again, these are quibbling points. The fact is, Ms. Coulter's argument is correct, and, if we looked into it in more detail, the numbers may point even more strongly in her favor, should we consider, for example, education, income, and a more detailed breakdown of age distributions. So I am not arguing that Ms. Coulter is wrong, nor am I claiming that she intentionally massaged the figures to help support her argument.

My point is much more basic. You see, I am sure some of my readers are asking "so what?" To many my quibbling over numbers seems petty. However, I do it to make a point. If you are going to include precise numbers, as Ms. Coulter did, and use calculations performed on those numbers to make your point, then you better be precise. As you can see, it was a trivial matter to show how Ms. Coulter had made a complete mess of the calculations to figure out how many murders were committed by men. Were I inclined to try to debunk her argument, that would have provided me with valuable ammunition, even though the numbers themselves still supported her argument, just not as strongly as she claimed.

So, please, in the future, if we are going to use precise numbers and, even more importantly, if we are going to perform calculation, make sure you know what you are doing. Math isn't that hard, especially not at this level, so it takes very little to get it right. However I am always surprised at how many fail to make even that minimal effort.

POSTSCRIPT

This argument may not apply so well to Ann Coulter as it does to others, but as she provided the best example, I chose to use her. However, as I know too well, most people who dismiss Ms. Coulter do not do so with reasoned arguments, but simply by calling her a "hater" or a "harpy", making allusions to her being a transvestite, or by bringing up some prior inflammatory remark. So I doubt anyone will be picking apart any calculations in her writing. Still, it is a good example of how bad math can make an otherwise valid argument appear weaker, so I used it anyway, even if this specific argument may never face such criticism.