Today PunditFact gave us another example, though this one they fixed pretty quickly when they learned something was amiss. Therefore our focus will fall not on the fact that a liberal, student Jalen Ross, gained the benefit from PunditFact's temporary misstatement of fact, but on the PolitiMath (the relationship, or lack thereof, between percentage error and PolitiFact's ratings) involved in keeping a "Mostly True" rating for Ross after fixing the mistake.
Ross attempted to use evidence from a scientific experiment to support the idea that racism remains a prominent problem in the United States. But he did not properly use the study's findings, as PunditFact belatedly pointed out:
Ross erred slightly in his exact wording. While white-sounding names spurred 50 percent more callbacks than the ones with black-sounding names, black-sounding names were 33 percent less likely to get responded to.The slight error in Ross' exact wording ended up exaggerating the discrimination against so-called "black-sounding names" by about 50 percent ((50-33)/33).
So Ross used bad math and ended up exaggerating the key figure by 50 percent.
Let's compare what happened to Republican Mitch McConnell when he used questionable math and exaggerated his key figure by 37 percent or more (bold emphasis added):
In any event, the lowest estimate of Bush's war spending through 2008 that is even remotely defensible is $808 billion. Tack that onto the $132 billion cost of Katrina and you get $940 billion for the wars and Katrina.PolitiFact's ruling on McConnell's claim? "False."
That's well over the expenditures expected from the Democrats' stimulus and children's health insurance bills, which total $686 billion once tax cuts are subtracted. Even if we included the cost of the tax cuts (for a total of $818 billion), he would still be wrong because that's less than the $940 billion that uses a more accurate cost of the war spending.
Hmmm. Apparently McConnell had no valid underlying point that the stimulus bill cost a great deal of money.