Presidential approval, party brands, and candidate quality in U.S. national elections

This article considers (1) presidential approval, (2) party brand differentials, as measured by the generic ballot, and (3) the presidential candidate polling differentials during the general election campaign to forecast the 2024 U.S. presidential and congressional elections. While all these three mass public opinion variables are leveraged to forecast collective partisan election outcomes, we consider the variables together as theoretically distinct determinants of partisan fortunes at both the executive and legislative levels and make the following contributions. First, using novel time-series data of mass opinion since 1937, we show that all three variables are weakly correlated and thus distinct conceptual and empirical measures of mass public assessments of partisan stimuli. Second, we use these three mass opinion variables to specify a unified model of U.S. national elections which better predicts variation in electoral outcomes compared to the standard forecasting approaches, finding that congressional election outcomes are predicted by party brands while presidential elections are predicted by presidential approval and the presidential candidate polling differentials heading into election day. Lastly, we validate our forecasting model using out-of-sample and 2024 forecasting predictions against other standard forecasting approaches.