Peace&Freedom
Member
- Joined
- Nov 29, 2007
- Messages
- 5,123
On weighting by party ID, suppose you poll a representative sample and get 28-28 party ID, and apply some weight based on that result. Suppose you poll the same sample the next day and get 30-26, but the raw score for candidates A vs B is the same as the previous day's poll. If the same weight is applied to these results, it wouldn't be reflective of the population.
Change party ID to gender and you don't have this problem, as a person's gender (normally) doesn't change from day to day.
That's the issue, independents are mainly not being polled to begin with, to even be available to be weighed later, let alone to 42%. And recent polls have already been caught oversampling Democrats, without weight adjustments. And most polls to this point are still based on registered voters instead of likely voters, which also skews the poll outcome Democratic (as there are generally more registered Democrats who don't vote than registered Republicans who don't vote).
Last edited: