Search

We address the question of whether class bias in the American electorate has increased since 1964. We analyze the Census Bureau's Current Population Survey and the National Election Studies for seven consecutive presidential elections, 1964-88. Our results show that conclusions regarding changes in class bias are sensitive to which measure of socioeconomic class is used - income, education, or occupation. We argue that income is the appropriate measure since government policies that discriminate based on socioeconomic class are most likely to do so based on income and there are measurement problems associated with using either education or occupation over time. Our analysis shows that there has been almost no change in class bias in the electorate since 1964.

I show that restrictive registration laws do not dissuade individuals with lower levels of education from voting any more than individuals with higher levels of education. This finding contradicts the result reported in Wolfinger and Rosenstone's classic analysis of turnout. I show that their conclusion was actually an artifact of the methodology they employed. Examining predicted probabilities generated by a nonlinear model such as probit or logit may produce spurious results when used to determine interactive effects between two independent variables. By respecifying the model of turnout to explicitly include terms to test interactive hypotheses and reanalyzing the data from the 1972 Current Population Survey (as well as data from the 1984 survey), I show that in fact, no such substantive interactive effect between registration laws and individuals' level of education exists at the micro level.

We offer a model of multiparty, elections that combines voters' retrospective economic evaluations with consideration of parties' issue positions and the issue preferences of voters. We show that both policy [...]