Despite veterans' ad, Kerry remains in the lead

[truthisfreedom]

http://msnbc.msn.com/id/5770431/

Despite veterans' ad, Kerry remains in the lead
MSNBC
Updated: 10:00 a.m. ET Aug. 20, 2004
<snip>

A new Pennsylvania poll from Franklin & Marshall College released Thursday showed Kerry with a six-point lead, 48 percent to 42 percent over Bush, but still within the margin of error, which was 3.8 percent.

The undecided amounted to eight percent. If one accepts the traditional pollster’s wisdom that two-thirds of the undecided ultimately vote against the incumbent, Kerry looks to be in fine shape in Pennsylvania.

Bush has made a remarkable 32 visits to the Keystone State since becoming president. This is puzzling in that he lost Pennsylvania by more than 200,000 votes last time. This could end up as one of the smartest — or most ill-advised — investments in modern campaign history.

But Pennsylvania illustrates the point that we don’t know for sure what the battleground states are and neither do the candidates. It is easier to say which states are not tossup states (Alabama, California, Utah, Rhode Island) than which ones are.

And this week's list of tossup states may not be the same as the week of Oct. 25.

[wtmusic]

1. Or maybe *because* of the SBV ad n/t

[joefree1]

2. Or maybe because of it

The more the swift boat liars talk about Vietnam the more apparent it is Bush never served.

[faithnotgreed]

3. did i read that right? kerry is "within the moe" of 3.8 yet leads by 6 ??

that may be stretching things even a bit more blatantly than usual

[THUNDER HANDS]

4. fuzzy math

[diablerie]

5. This could get ugly

If * were to get a bump coming out of the RNC...

[DCCyclone]

6. faithnotgreed, you're making a common mistake......

The "margin of error" in a poll doesn't have anything to do with the lead one candidate has over another, but rather is how much each candidate's percentage, INDEPENDENT OF EACH OTHER, could be off in either direction. This means that if a poll shows Kerry up 48-42 over Bush, with a 3.8% margin of error, then Kerry is anywhere from 44.2% to 51.8%, and Bush is anywhere from 45.8% to 38.2%. Because there is overlap between the two candidates' ranges, the race is, in fact, within the margin of error.

Further, the poll has a "second type" of "error" that's virtually never mentioned in media reports about poll results, referred to as the "confidence interval." This is the percentage of statistical confidence that the range for each candidate is accurate. The confidence interval for most of these polls is 95%. This means, in the example in above, that there is a 95% chance that Kerry's support among all voters on the day(s) the poll was conducted was between 44.2% and 51.8%. There is still a 5% chance that he could have pulled in either more than 51.8% or less than 44.2%. The same applies to Bush, i.e., there is a 5% chance he would have pulled in, on the day(s) the polling took place, either more than 45.8% or less than 38.2%.

What all this means is that for a candidate to be ahead BEYOND the margin of error, he has to lead by a percentage that is more than double the margin of error. So, in the example above, with a 3.8% margin of error, Kerry would have to lead by more than 7.6% to be said to lead outside the margin of error. If he leads by, say, 8%, at, say, 49-41, then there would be a 95% chance that he'd do no worse than 45.2%, and a 95% chance that Bush would do no better than 44.8%; thus, Kerry would be leading outside the margin of error.