I'm not arguing here but what is you theory as to why Newsweek would do this. It is a liberal rag or at least it was, when I canceled my subscription years ago.
They did the same thing to Dems after the DNC (oversampling them). Possibly more Republicans answered the phone, or else more Independents/Dems decided to call themselves Republicans.
but what is you theory as to why......
The only chance left is to become the underdog and hope that this will make the base turn out.
Of course the flip side is that the base will think all is lost and voting is not worth the effort.
So that when other polls tomorrow say that Kerry is 5 points behind (instead of 10) behind Bush, they can run the results with the headlines like:
“Kerry Gaining on Bush”,
“Bush Bounce Fading”,
“Independents Moving Back to Kery”,
“Bush Loosing Support as Kerry Connects with America”, etc…
After the Kitty Kelly book comes out, they will do another poll that now oversamples Democrats. This will show a wholly-bogus 10-point drop in Bush's numbers. The media will credit the Kelly book's "serious allegations" for this, which will give them an excuse to focus on the book for a few weeks, by which time Bush's real numbers will have dropped 10 points or more.
I believe that this occurs because they set out to make a certain number of "successful" phone calls. They ask the party affiliation of each. It just happens that there were more Republicans than Democrats in those "successful" calls.
Since the registration numbers are nearly equal between D's and R's, one would expect closer numbers in a random sample. They do not intentionally set out to oversample one party over another.
I'm not a statistician but I would assume that there is some reasonable way to "normalize" the results randomly (a la Zogby's "special sauce"). Whether or not they did this, I don't know.