ABSTRACT
People both inside and outside academia do election forecasting and this is an area where academics learn a lot from non-academics and vice versa. Sadly, much of the non-academic forecasting is not well documented. Election forecasts constitute data for analyzing current political situations, performance of political actors and the effect of substantive events. Rightly or wrongly, the demand for forecasts sets an academic agenda. Structural models are the archetypal political science forecasting models. They are traditionally based on fitting a parsimonious theoretically informed regression model to an historical set of elections and then using the resulting equation for prediction. The pendulum effect is often referred to as an election cycle, but that term is also used to refer to the period between two elections. No party has lost a British general election when it was ahead on both leadership and economic competence evaluations.
