Sullivan's innumeracy:

Background: On October 22, Dana Milbank wrote an article in the Washington Post about Bush's misrepresentation of the facts: One item he mentioned was this: (emphasis added)

Other times, the president's assertions simply outpace the facts. In New Hampshire earlier this month, he said his education legislation made "the biggest increase in education spending in a long, long time."

In fact, the 15.8 percent increase in Department of Education discretionary spending for fiscal year 2002 (the figures the White House supplied when asked about Bush's statement) was below the 18.5 percent increase under Clinton the previous year -- and Bush had wanted a much smaller increase than Congress approved.

The following day (Wed, Oct 23) Sullivan publishes an email he received from a reader. The key sections are: (emphasis [yellow] added)

A FEW DECENT POINTS?? An email provides some balance to my link to Dana Milbank's Washington Post article yesterday about president Bush's sloppiness with facts: Good dish . . . with the notable exception of touting Milbank's article. Although he accuses the President of lying (6 times by my count), his evidence is less than compelling. ... Milbank's statements are, if anything, more dubious than the President's. Then he goes on to make some basic logical errors. ... Finally, he closes with a basic error in arithmetic: "Other times, the president's assertions simply outpace the facts. In New Hampshire earlier this month, he said his education legislation made "the biggest increase in education spending in a long, long time." "In fact, the 15.8 percent increase in Department of Education discretionary spending for fiscal year 2002 (the figures the White House supplied when asked about Bush's statement) was below the 18.5 percent increase under Clinton the previous year. . ." In fact, a 15.8 % increase is "bigger" than a previous year's 18.5 % increase (115.8 x 118.5 = 137.22; 137.22 - 118.5 = 18.72% ). Dana might wish to define an "increase" as a multiplicative factor, but dictionaries, math books, and common usage all refer to addition. ... [Milbank] has failed abysmally in his attempt to make a cogent case. And if this is the best available argument that Bush is a liar, he must be pretty darn truthful.

Sounds persuasive! At least it does if you're not paying close attention. But it's wrong. It's the result of mathematical ignorance or it's a deliberate trick. Here's why.

The first set of numbers (15.8 and 18.5) are percentages. But then the numbers are applied to real units (in this case dollars) which are subtracted to produce another real unit number (again, in dollars) - which is automagically treated as a "percentage" and (incorrectly) compared to a number in the first set. Let's look at some examples.

By applying the spending increases to a base amount of one hundred dollars, you get a number that, when incorrectly treated as a percentage, seems reasonable (say, anything between 5% and 30%), but it's purely an accident. (Or is it?)

Spending numbers should be adjusted for inflation and made per-capita if they are to have any meaning. Absent that kind of analysis, we speak of year-to-year differences in terms of percentage because it allows us to see if real spending is keeping up with, or exceeding previous levels (by subtracting inflation and population growth rates). And here is where the writer deceives us: By taking a (made-up) spending change and treating it as a rate of change.

Another possibility: Now it's true that the year-to-year increase in dollars is slightly greater under Bush. That's what the writer's 18.7% really measures: dollars, not growth rates. He (or she) has inadvertently arrived at a percentage increase in funding by Bush compared to year 0. But then it's used to gauge Clinton's increase also compared to year 0. Thus, we're really evaluating dollars vs. dollars - even though it's expressed in percentages. Bush's spending is measured by comparing it to a budget two years prior, but Clinton's by one year. (If you want to properly engage in this sort of oddball analysis, then Clinton's spending increase should be compared to the budget for year -1.)

Actually, we're not sure what approach the writer is making: using a base of $100 and then converting spending amounts to percentages - which was our first impression because of the stated emphasis on addition and disdain for multiplication, or, comparing apples to oranges (Bush-$-increase:spending-2-years-earlier vs. Clinton-$-increase:spending-1-year-earlier). Either way is wrong.

For Sullivan to be introducing this kind of "reasoning" into the debate shows how desperate (or ignorant) he really is. Where did this guy go to college?

Sullivan owes Milbank an apology for promoting this nonsense.

posted by Quiddity at 10/27/2002 04:52:00 AM