For example I've noticed that the Kos polling always seems to be 5-10 points higher on Obama approval than virtually every other pollster. Then I checked the cross tabs and found that they are weighting their samples about 10% more Democrat and 10% less Republican than most other polls.

Basically they looked for the best possible party ID survey they could find and used it, even if 2006 and 2008 turnout or more recent surveys were different.

I'm sure we could find some right wing web sites who did the opposite, so take it all with a grain of salt.

But to prove my point…

“Do you think our current health care system should be better?”

“Would you say that you're happy with your current health care?”

Both questions would likely get a majority affirmative answer.

Notice that the second question does not use the word “satisfied” or ask about improvements to their health care. It also doesn't specify any portion of care, so the person could even be led to believe that they're discussing the performance of their doctor. Depending on what answer you want, you ask either question 1 or question 2.