[simonh]

For example in the one where the participant's own answer was disguised as that of another person we can't discount the result so easily. That's also true of the studies where participants downgraded their confidence when asked to give an account of it.

On the invented studies, bear in mind that the point wasn't to measure changing the participant's mind, only for them to rate the value of a study that either supported or contradicted their initial position. Their only basis for evaluating the value of either study was their own pre-existing bias, so objectively they had no reason to evaluate them differently.

That's quite different from expecting them to change their minds, as the reasons for them holding their position might not even have been addressed by the study. For example someone who disagrees with capital punishment on moral grounds may not care whether it is an effective deterrent or not so may no have any reason to doubt a study that it is an effective deterrent.