[Arthurio]

Many pollsters weight their samples based on party, age, sex, race, etc. others do not. If a pollster is obtaining samples that are not representative of the population at large, the results become questionable. Some argue that since party affiliation is not set in stone (like race or sex is), it should not be weighted

[muawiyah]

Alas, what is the methodology ~ can you describe it in detail. If you do a random sample of 1000 and you get an expected number of self-identifications, as long as you don't have a long list of sample cells (age, weight, sexual preference, handicapped, blind, etc.) ~ if it was random, and you've controlled for all externalities so that you got as random a sample as you can get, then you are not weighting anything.

The more detailed samplings that give you a normal distribution of characteristics of the population in general usually involve THOUSANDS of samples ~ else you'll have sample cells out there that don't have enough responses to fit within the expected confidence interval.

We really should always expect a national sample to report back more Democrats than Republicans.