Yeah - the trick worked both ways>> "on the basis of the manipulated score, 10% of the subjects switched their voting intentions, from right to left wing or vice versa. Another 19% changed from firm support of their preferred coalition to undecided. A further 18% had been undecided before the survey, indicating that as many as 47% of the electorate were open to changing their minds, in sharp contrast to the 10% of voters identified as undecided in Swedish polls at the time."