The crux:

My favorite study (pdf) in this space was by Yale’s Geoffrey Cohen. His experiment found the position of an individual’s preferred political party overwhelmed both the objective policy content and the individual’s preexisting beliefs. Cohen had a control group of liberals and conservatives look at a generous welfare reform proposal and a harsh welfare reform proposal. As expected, liberals preferred the generous plan and conservatives favored the more stringent option. Then he had another group of liberals and conservatives look at the same plans – but this time, the plans were associated with parties.

Both liberals and conservatives followed their parties, even when their parties disagreed with them. So when Democrats were said to favor the stringent welfare reform, for example, liberals went right along. Three scary sentences from the piece: “when reference group information was available, participants gave no weight to objective policy content, and instead assumed the position of their group as their own. This effect was as strong among people who were knowledgeable about welfare as it was among people who were not. Finally, participants persisted in the belief that they had formed their attitude autonomously even in the two group information conditions where they had not.”