Over at Science Blogs, Mike the Mad Biologist complains bitterly about innumeracy among political reporters. Which reminded me of a recent breathless report on NPR concerning the dramatic tightening of the presidential race: McCain Draws Even! It seems the same polling organization that two months ago found Obama to be leading McCain by 48-40 percent now shows the Democrat's lead to have narrowed to 46-43. Thus a long interview with the director of the polling organization, who held forth on the various factors that had allowed McCain to surge to his current 3-point deficit--a margin, we were told, that is "inside the poll's margin of error, thereby indicating a possible dead heat."
Apparently the reporter felt no need to question whether, if the poll indeed has a margin of error greater than 3 percent, the whole premise of the story might be mistaken. Is it possible that Obama was and is ahead by, say, 47 to 41 percent? And that in the earlier poll this difference registered 48-40 and in the most recent one at 46-43? If 46-43 means the race could be a dead heat, does it not also indicate that it could be something else? Mike's complaint is to the point: if the race were tied, there is, practically speaking, no chance that virtually every poll would indicate Obama is ahead by a small (or medium) smidgen. Or, to put it differently, the fact that no poll shows McCain ahead disproves the reporter's suggestion that the race might be a dead heat.
It appears the race has tightened. But that is not the same as "Presidential Race Draws Even."
I'm a worrier, but it is too early to worry too much, and, anyway, some of these news stories probably reveal less about the state of the presidential horse race than they do about the state of the reporter's mathematical education.
Comments