Many questions central to political science, such as the issue of stability and change in the United States party system, revolve around the degree of persistence or memory in a political process. Fractional integration techniques, which allow researchers to investigate dynamic behavior that falls between the stationary and integrated alternatives, provide more precise ways to test hypotheses about the degree of persistence than current modeling strategies. Hypotheses: Choices about the treatment of the time series properties of the data and model specification may influence the substantive conclusions drawn about the dynamics of important political processes. Methods: Fractional integration methods are discussed and compared with common univariate diagnostic tests. A transfer function model of macropartisanship using fractional integration techniques is contrasted with traditional ARMA and ARIMA methods. Results: Fractional integration techniques offer a more flexible way to model a time series. Using fractional integration techniques, we find that macropartisanship is dominated by a strong permanent component, but also contains transitory dynamics in response to changes in economic evaluations and a measure of presidential approval. Our empirical work shows the importance of taking seriously the time series properties of data to ensure valid inferences about the dynamics of political processes.