Two things to keep in mind with polls.

[pnwmom]

Any "lead" within the margin of error -- which should always be stated, and depends on how many people were in the "sample," and is almost always at least 3%, and often higher -- is NOT A LEAD. Any results within the margin of error is a TIE.

The poll is only as good as its sample. With more and more people going to cell phones exclusively, it is going to be a challenge for poll takers to sample these people adequately. Particularly in this race, with so much of Obama's support coming from young people, it will be interesting to see how this all shakes out. His support may be underestimated in the polls, IF the young people who only use cell phones are somehow different from the young adults who don't.

[CK_John]

1. People lie an many change their minds. n/t

[pnwmom]

2. True enough. n/t

[mohc]

3. This is a common way of looking at polls

But it is not quite accurate to call a lead within the margin of error a tie. Most polls use a 95% confidence interval, so when you see a margin of error of 3% it means there is a 95% chance that the true results lie within 3% of the results of the poll. All the poll is really stating is a likelihood. With the same sample size and results, if the CI were dropped to 60% the margin of error would drop as well, so even if a lead is within the MoE of a 95% CI poll, it very well may not be within the MoE of a 60% CI poll. In that case there would still be at least a 60% chance that the candidate was actually leading (there would be some chance they were leading by more than the MoE as well as a chance that they were losing). So I guess it depends on what a "tie" is, but most of the time what a poll is showing you is the probability that a candidate has the lead, not how big the lead is necessarily.

[pnwmom]

4. Yes, I understand that. And I still say that a "lead" within the poll's

margin of error is a tie. Because if another sample was taken on the same day, it could just as likely have produced opposite results -- the candidate who was on top could be on the bottom (still within the margin of error).