Thursday, February 19, 2004

Today's political note: More on polls

If a candidate's lead is less than a poll's margin of error, we can't say he's ahead. Most of us have that part down. But nor can we say that it's a "statistical dead heat" or "statistical tie." As the AP stylebook puts it, that's inaccurate if those numbers aren't the same, and "if the poll finds the candidates are tied, say they're tied."

How must the results be worded? AP gives guidelines.

Here are some examples:
1. Say Kerry has 43 percent and Edwards has 40 percent and the poll's margin of error is plus or minus 3 percentage points. The 3-point lead is within the margin of error; therefore, say the candidates are about even.

2. A poll in Kansas shows Bush with 52 percent and Kerry with 44 percent. The margin of error is plus or minus 4.5 percentage points. Bush's lead (8 points) is more than the margin of error but less than twice that number (9 points), so we can say Bush is slightly ahead or has a narrow lead. (Don't forget that AP calls for margins of error to be rounded to the nearest half-point or full point.*)

3. A CNN poll says that in a head-to-head match-up, Kerry and Edwards lead Bush. For Kerry, it's 55 percent to 43 percent. For Edwards, it's 54 percent to 44 percent. The margin of error is plus or minus 4 percentage points. Because Kerry and Edwards both lead by more than 8 points, it's safe to say that they're leading. (Note that there are other problems with this story, though: It doesn't point out that the 1,006 adults surveyed were registered voters. You just have to infer that.)

* Here's the technical stuff: This Survey USA poll lists the margin of error here as 4.3 percent. This number is determined by taking 1 over the number of respondents (549). You get 0.00182149. Then take the square root of that, 0.0427. Then, per AP, round it to the nearest half-point, for 4.5 percentage points. Since there is such a precise way of determining the margin of error, I'm not sure why AP requires this specific rounding. I've sent a note to stylebook editor Norm Goldstein; we'll see if he responds.)


Post a Comment

<< Home