We assess the importance of parties in Congress by comparing roll-call voti
ng behavior against the preferences of members of the House as expressed in
surveys conducted during the 1996 and 1998 elections. The surveys were con
ducted by Project Vote Smart. Our findings support two key conclusions. Fir
sts both party and preferences mattered in predicting roll-call behavior in
the 103d, 104th, and 105th Congresses. Second, the independent effects of
party were present in only about 40% of roll calls. The incidence of party
effects was highest on close votes, procedural votes, and key "party" issue
s. It was lowest on matters of conscience, such as abortion, and "off-the-f
irst-dimension" issues, such as affirmative action and gun control.