400 Republicans, really?
That is a very strange sample size. Generally, they ask 1,000 registered voters, 1,000 likely voters and so on, then break down the data by demographic/political affiliation.
Although its crazy to see their results, I really have to question why they didn't ask the same question of Democrats, or have a larger sample size. Furthermore, I can't find any comparative data on the accuracy of PPP outside of their one quote from a WSJ article. RCP doesn't even have them listed as a pollister firm, which is very questionable.
*edit*
After reading up on PPP, they are a Democrat pollister firm which is usually employed by Democrats to provide biased, targeted polling. They've readily admitted to it, too.:
http://mediatrackers.org/2011/03/public-policy-poll-overrepresented-union-househoulds-in-poll/
http://www.politico.com/news/stories/0609/24062.html
According to PPP, they were paid to poll North Carolina's senatorial seat in 2010, and stated that Richard Burr was one of the most vunerable candidates in the nation. In reality, Mr. Burr won the seat by the largest majority in state history: http://en.wikipedia.org/wiki/Richard_Burr#2010_Senate_campaign
Back from the dead, I'm afraid.
