Playing the synthesizer with Canadian data: Adding polls to a structural forecasting model

Election forecasting has become a fixture of election campaigns in number democracies. Structural modeling, the major approach to results, relies on ‘fundamental’ economic and political variables predict incumbent’s vote share usually few months advance. Some scientists contend that adding intention polls these models—i.e., synthesizing polling information—can lead important accuracy gains. In this paper, we look at efficiency different model specifications predicting Canadian federal elections from 1953 2015. We find only allow modest gains late campaign. With backdrop mind, then use make ex ante forecasts 2019 election. Our findings have implications for discipline Canada as they address benefits combining results; varying lag structures; issue translating votes into seats.