Toward stability in presidential forecasting: The development of a multiple indicator model

Most election forecasting models rely on single indicators for such complex variables as the health of the economy, yet each employs different indicators for the identical concept. This article argues that single indicator models are inappropriately dependent upon past correlations, leading to inflated goodness of fit measures and unstable future forecasting success. We develop a multiple indicator model whose performance favorably compares with that of other models and provides additional forecasting stability by addressing the single indicator problem.