The model built for this analysis considers two types of member behaviors—voting behavior on roll-call votes on the House floor, and congressional activity on and off the House floor. This appendix describes the sources of the data and the methods used on the data to put House Republicans into factions.
The major source of voting statistics are those that have been compiled by the staff of the National Journal (NJ) since the beginning of the 97th Congress in January 1981. Each year since 1981, the National Journal staff has compiled ratings for every member of Congress on three ranges of policy—social, foreign, and economic policy. Annual House ratings in each category have been based on from thirty-nine to fifty-eight votes. The selection of votes is determined by journal staff in consultation with special interest groups and other organizations that rank members of Congress. A member's relative liberalism or conservatism in each area is compiled and compared against other members of Congress. Each member attains a score in each policy area on two zero-to-one-hundred scales, one measuring conservatism at zero and liberalism at 100, and the other with the opposite values. This study uses the scale with conservatism set at zero. National Journal scores are based on a member's comparison with other members; that is, a representative with a 28 score on the economic spectrum was comparably more liberal than only 28 percent of the House and more conservative than about 72 percent of the House. Often the two scores will not add to 100, since there may be many members