Very misleading headline. If the interviewees had been asked "do you support Bush's plan to..." the 'rat respondents would have been 2-1 against it. Any question that begins "should people have the choice to..." gets over 50% of the vote. This has nothing to do with views on retirement funding and everything to do with how the question is phrased.
Just those 2 words alone will cause many to say no when they would say yes to the actual plan.