More Newsweek poll sampling weirdness

[HFishbine]

37.2% GOP
31.6% Dems
31.2% Ind


0.0% Age 18-29
60.0% Age 30-49
40.0% Age 50-64
0.0% Age 65 & over

33.3% Military Household
66.7% Non-Military


As we know, the sample greatly overweights pollees with GOP affiliation. But the "weirdness" is farther down. In the breakdown by age, the 60-40 split for young-middle-age and old-middle-age is plausible; I don't know the real figures. But what about the young and old? Apparently, the 18-29 year olds and the over-65s were not included in the aggregate at all! Were they sampled?

Similarly, in the military/non-military split, it is apparent that some intern just assumed that 1/3 of the sample was in a military household. In the population, the fraction of military households cannot exceed 5%.

http://www.pollkatz.homestead.com/#anchor_117

[Walt Starr]

1. The Internals on the Polls tell me one thing

The Corporate Media is TERRIFIED.

[Lex]

2. I can't wait to hear the hemming and hawing by the media on Nov. 3rd.

The pollsters will rush forth to proclaim that they will have to "fundamentally change the way" they do polling to reflect cell phones and cable-provided phones, etc.

Just a guess.