|
"It may be the case that a poll conducted entirely over the weekend picked up a few more Republicans than it should of."
ok, first, it "should HAVE", not "should OF". fire the reporter. and the editor. i try to cut du posters some slack, but newspapers should get this stuff right.
anyway, as for polling, party identification is something you normally start each survey with, along with age and other demographic information. this way, you can easily control for reaching a skewed population.
in other words, from the many polls taken over the summer, the percentage of the population that identifies itself as republican, independent, democrat, and other is well known. so if, in a labor day poll, you reach more republicans, you simply adjust the numbers to match the known population. very strightforward algebra does this.
yes, people sometimes lie in polls, but the main point is that people don't lie any differently over labor day weekend versus any other polling period.
|