Predictive Estimation in Canadian Federal Elections
Various estimation methods have been employed to predict election outcomes. This project was motivated by discrepancies between the real outcomes of recent Canadian federal elections and the predictions by the existing approaches such as Grenier (www.threehundredeight.com) and Rosenthal (Canadian Journal of Statistics, 39(4):721-733 (2011)). In fact, each of the existing procedures requires a set of assumptions. The assumptions are not explicitly listed in the accessible references. It is unclear how the prediction performs with violated assumptions. We formulate the required assumptions by the two prediction procedures proposed in Rosenthal (2011), and provide variance estimation for the two estimators. Departures from the assumptions are explored with the real data of the federal elections from 2006 to 2015. An extensive simulation study is conducted to examine potential impacts of various deviations from the assumptions. We find that misleading polling results may cause the most damage to the prediction.