I think it's important that they examined at least the correlation between the thought-experiment answers and the actual behavior.
If there's one thing that makes me immediately suspect a study, it's when the data includes in significant part anything self-reported by the study subjects, without accompanying very strong evidence (ideally from someone else's much larger study) that such self-reporting corresponds to something more objectively measurable.
If there's one thing that makes me immediately suspect a study, it's when the data includes in significant part anything self-reported by the study subjects, without accompanying very strong evidence (ideally from someone else's much larger study) that such self-reporting corresponds to something more objectively measurable.