[SALChamps03]

The poll interviewed 1,216 adults in the U.S. from Jan 9-12, and has a 2.9 percent margin of error.

US population 300 million.

1,216 = .000004% of the population. These polls claiming to represent the opinion of the American public do not survey enough people for the poll to be taken seriously. A minimum of 10%(30,000,000) would be needed for that purpose.

[Young Scholar]

Actually, statistically, about 1,000 people is enough if a truly random sample is used. But if the sample is manipulated, you could make the poll say whatever you wanted.

[Theo]

You only need perhaps 700 people to have a statistically sound poll, assuming the sample is representative of the population. 1,216 is plent of people.