The voter samples for this study were drawn from lists of all registered voters in Arizona, New Mexico, Colorado, and Utah. Sampling and mailing procedures were conducted in two waves--first in Arizona and New Mexico and later in Colorado and Utah. Sample size varied from state to state, as voters were selected on the basis of state senatorial districts. There are thirty such districts in Arizona, forty-two in New Mexico, thirty- five in Colorado, and twenty-nine in Utah. Fifty voters per district were chosen, using a stratified random selection process by which voters were selected in proportion to the registration of Democrats, Republicans, and Independents in each district. Spanish language copies of the questionnaire were included in the mailing to respondents with Spanish surnames in Arizona and New Mexico.
We attempted to interview the total populations of senators in all four states. In Arizona, data were obtained from 28 of the 30 senators, while in New Mexico 36 of the 42 senators participated in the study. In Colorado 28 of the 35 senators were interviewed and in Utah 26 of the 29 senators participated. Senators were given a self-administered questionnaire similar to that given the voters in their states. In addition, subsequent oral interviews were conducted to allow legislators to expand their answers as well as cover new topics. Copies of all of these questionnaires follow. Some changes (additions and deletions) were made in the interview schedules between the two waves of the survey.
For the voter samples a mailing procedure similar to that described by Don A. Dillman ( "Increasing Mail Questionnaire Response in Large Samples of the General Public", Public Opinion Quarterly vol. 36 ( 1972) pp. 254-257) was employed. Mailings were highly personalized and involved four major steps. In the first step, a first class letter addressed to the sampled voter, a questionnaire, and return envelope were sent out. One week later this was followed by a reminder postcard to all potential