Wording and Survey Design

It doesn't take much to make a poll on the same topic give very different results. Case in point:

I don't have the "n's" for the 2 polls, but it doesn't take a rigorous test to see that these differences are very unlikely to be random sampling error.