Beschreibung:
This chapter states that institutions and preferences are hopelessly confounded with the world of politics. Starting with data about institutional variation and variation in policy outcomes, it is impossible to make any hard and fast inferences about the impact of institutional features on policy outcomes. Do the states with a line-item veto spend less on certain budget categories because of the line-item veto, or because those states have electorates that prefer both the line-item veto and those expenditure patterns? Disentangling cause, effect, and spurious correlation can be virtually impossible with natural data. Experiments are uniquely suited for examining institutional effects. The experimenter can hold preferences constant and randomly assign subjects to treatments distinguished only by variations in institutional rules; resulting differences in behavior may be ascribed to the institutional differences with a degree of confidence that would be impossible in natural political settings.