It is not uncommon to set quotas for certain demographic groups that don't necessarily represent the poplulation as a whole. This is done, so the groups can be compared to one another. The reason this is done is so the numbers in each comparison group are sufficiently large enough to allow statistically significant comparisons.
In order to extrapolate the data to the population as a whole, each record (respondent) is weighted up or down to reflect the respondents actual proportion of the entire population, using census data.
For example, if I set a quota of 100 women and 100 men, and the actual proportion of the population is not 50/50, but 60/40, women respondents would be weighted 1.2 and men would be weighted .8.
When analyzing the data, and comparing women to men, the data is left unweighted. When looking at the entire population, the weights are applied.
That being said, I don't trust polls by Newsweak. Polls are controlled by the people that pay for them.
Their methodology, if applied correctly, appears to be sound, however.
Interesting Strider. And I assume you are saying that Newsweek did not apply any weightings?
-T