...results that are taken exclusively since Kerry's speech. I agree with that totally. Anyone see any polls that have been conducted over this past weekend? Also, to get results that are +/-3% or better as margin of error, isn't a sample size of only 1,000 a bit short?
Here is a link to a sample size calculator that I use which suggests that the sample should be around 1,068 completed interviews. If they are basing the results on just 1,000 completed interviews their margin of error is actually 3.1%, according to the calculator.
http://www.surveysystem.com/sscalc.htmAgain, these findings are still based on popular vote, which as we know from the 2000 election, meant nothing for Gore.
Now, if a polling firm was to do a survey with say 2,500 completed interviews, the margin of error would drop to 1.96% at the 95% level, or 2.58% at the 99% level of confidence.
Why are these polling companies using such small samples in a race that is as close as this one? It can't be money, because both political organizations have enough money to cover the costs of doing such polls. I'm thinking it's to maintain the drama and theater of the presidential campaign.