The latest Field Poll from California holds good news for Republicans: Carly Fiorina trails the incumbent Democrat, Sen. Barbara Boxer, by just 47 to 44 percent. Our Jim Geraghty calls this “one of the least reassuring three-point leads ever.” But at least he admits what the results plainly show — that Boxer is leading, however tenuously. Other writers, including some of our astutest allies, have gone a step too far by calling the results a “statistical tie” or a “dead heat.”
For the record, there is no such thing as a “statistical tie,” unless the results show an actual tie. The kernel of truth in that phrase is that surveys like this have a margin of error — in this case, plus or minus 3.2 points. Roughly speaking, this means that if an actual election had been held at the time the poll was taken, and people had voted the way they told the pollsters, there’s a 95 percent probability that the election’s margin would have been within 3.2 points of the poll’s margin.
So far, so good. But some writers take this to mean that any result within that plus-or-minus-3.2-point range is just as likely as any other. That’s not true. In fact, the probability distribution is clustered around the middle, and it’s more than 50 percent likely that the actual result would be within 1 point of the reported figure. More important, perhaps, if being within the margin of error creates a “statistical tie” for Fiorina, it would be just as correct to call the results a “statistical 6-point lead” for Boxer.
It’s time to retire the phrase “statistical tie” and give readers credit for a little intelligence. A three-point lead four months before the election is like a three-point lead at halftime of a basketball game: It’s better than being behind, but it can be overcome fairly quickly. The message of the Field Poll results is that it’s a close race in California, and Boxer probably has a small lead. Calling this a tie, however well-intentioned one may be, is closer to spinning than reporting.