While it is widely accepted that in the United States, political party labels serve as brand names which cue voters about the beliefs and ideologies of members, I explore the possibility that the signals sent by these labels are contingent upon the party membership of the individual voter. More specifically, I draw on social identity theory and hypothesize that individuals will be more likely to perceive heterogeneity among members of their own party. I find support for this hypothesis in perceptions of both the overall ideologies and voting records of US senators. Additionally, I compare respondent perceptions back to actual voting records and find that inpartisans are (1) only more likely to be correct when senators do in fact vote differently and (2) significantly less likely to be correct when senators vote the same way. These results suggest that the partisan differences uncovered are due to psychological bias and not just informational asymmetries and that biases stemming from group membership may lead to distorted opinions of senators and the representation they provide.