This is a good example of misleading (presumably purposefully) use of statistical data. The data do not suggest that Hillary is the less electable of the two Democratic font runners. The question they neglected to ask is; how many percentages of the 40% anti-Hillary group would vote against Obama if he becomes the Democratic nominee? If most do, it doesn't matter which candidate is on the ballot. Another piece of important but missing information is how familiar the answerers were with each candidate. It is inevitable that candidates who are better known attract more anti-votes in this kind of survey.