Because the attitude research approach to determining public opinion typically yields data irrelevant to public policy formation, researchers need to consider nontraditional approaches to public opinion measures. Attitude research assumes that people actually have readily measurable opinions and that each attitude can be measured singly. Measurement is directed toward uncovering a person's most preferred position. To be politically relevant, however, survey data must indicate the intelligibility of public responses; require respondents to set priorities and make trade-offs; depict what costs would be tolerated for a given policy; and distinguish degrees of political acceptability. Alternative measures include the "budget pie" technique in which participants are asked to allocate a fixed amount among different programs. A second technique involves providing participants with chips that they allocate to slots which represent government programs. Disadvantages of these methods include the amount of time required and the capacity of citizens to make resource allocations. A third approach allows respondents to select ranges of responses categorized according to acceptance, indifference, and rejection. Aggregation of data in all of these approaches is a major problem. Nonetheless, the goal should not be to measure public attitudes, but rather, political preference. (Author/KC)