|
Pollsters like Gallup and Pew used a random sample and did not weight their polls based on party ID of that random sample. You can't oversample a group if the sample is random.
Many people, including myself, before the election saw a shift toward the GOP in terms of party ID before the election, and this shift did not come as much of a surprise.
There's no problem to future polls, regardless of problems in the 2004 exit polls, if they use a random sample instead of pegging their data to the political situation of four years ago.
|