Extant research in judicial politics has documented disparities in adjudication and biases in judging. However, since most empirical studies to date have focused on courts that publish judges’ individual opinions, our understanding of collegial courts that report panel decisions without dissenting opinions remains limited. We propose a methodology that tests which of several decision-theoretic models of group deci- sion making best fit such per curiam decisions, infers judges’ individual preferences, and quantifies the proportion of cases that would be decided differently if the court’s consensus were consistently applied (an “inconsistency rate”). Applying this methodology to the Swiss appellate court handling asylum appeals, where cases are assigned conditionally at random to panels, we find a persistent inconsistency rate of about 5% due to variation in decision making between judges, and that judges’ preferences are correlated with partisanship in expected ways.