I've written on polls before [here] and today I see the need to do it again. You gotta watch'em closely -- How are the questions phrased? Who are the people being polled?, etc. etc. Here is Kathryn Jean Lopez making an important point:
I've never been a math whiz, so I usually leave poll analysis to others, but take a look at "Newsweek poll puts Democrats far ahead in Congress " — should anyone be shocked that Dems come out ahead when only 28 percent of your poll is made up of Republicans, and 35 percent Democrat, is it surprising that dems fare better? (Of those sampled, 282 identified themselves as Republicans, 349 as Democrats and 330 as Independents.)