« Latest Democracy Corps Survey Shows Kerry Up By Two Nationwide and Leading 52-45 in Battleground States. | Main | Kerry Ahead by 1 in New Harris Poll »

Latest LA Times Poll: Kerry/Edwards Leading By 1% in Horserace

In a poll released October 25 the LA Times has Kerry/Edwards leading Bush/Cheney 48% to 47% in a 2-way RV matchup.


I don't know where this comment belongs, but I have a general question. Is there data -- or common wisdom -- about how many of the voters who claim they are undecided at a late stage of a campaign actually end up voting?

I'm not sure where to post this comment, but I was wondering whether there was any data that could help predict how many voters who claim they are undecided at the late stages of a campaign actually end up voting.

On several websites I see that the election horse race summary gives the details of the state by state percentages according to each of the several polls. I notice that the "margin of error" is typically 4 points. I understand that this is approximately the expected 2 sigma spread from the sampling (random) error. However, I calculate that the 2 sigma spread among the various polls is more like 6 points. Isn't this a reasonable estimate of the systematic error of polling and a measure of the confidence limits for polls? Without estimates of the systematic as well as random error, how can I intelligently place my bets?