Using social indicators to forecast partisan alignments in congressional election years

This paper presents five forecasting equations for political party alignments in the United States during the post-World War II period. Time series data published by various public and survey organizations are utilized to estimate for the years 1948–1974 for party identifications and results of elections for the House of Representatives. The observed equations explain from 86–93% of the variance for the years 1948–1974 and usually allow acceptance of the null hypothesis of no autocorrelation of disturbances. These equations are then employed to make forecasts of the 1976 election year alignments which are compared to observed values. The forecasting equations for 1976 are generally successful, for example, missing the congressional alignments by nine House seats. Some contingent forecasts are made for 1978 election year alignments, based upon certain alternative assumptions about economic and political climates. The equations do not intend to test new theories but rather to apply fairly simple ideas to forecasting trends in public opinion. The authors argue that forecasting equations are useful for anticipating future trends, for planning future research and for assessing whether an unusual event is manifesting itself or alternatively whether a certain trend in public opinion is to be expected from normal processes.