TESTING COMMITTEE COMPOSITION HYPOTHESES FOR THE UNITED-STATES-CONGRESS

Tests of theories of the composition of congressional committees are often problematic. Among the problems are presumed normality of the data, presumed cardinal meaning of the data, and the use of a mean score of committee preference when the corresponding theory only contains implications for the median score. This essay introduces a testing method that avoids these problems. It tests five hypotheses: (1) committees are preference outliers, (2) Democrats stack their slots on committees with liberal outliers, (3) Republicans stack their slots with conservative outliers, (4) committees are representative of the whole House, and (5) Democratic contingents on committees are representative of the whole Democratic caucus. Strong support, however, cannot be found for any of the hypotheses. The preference outlier hypothesis receives the most support, but even here the evidence is mixed.