Elections analysis versus prediction.
The nomination of Donald Trump as the Republican Party's candidate for the US presidency raises questions for how analysts ought to respond to unlikely scenarios and how 'analysis' differs in its construction from prediction. His victory raises the question whether the low probability assigned by many observers to his chance of winning was accurate, and he nevertheless won due to the inherent volatility of primary politics, or if it was the result of fundamentally flawed modelling from the outset.
Past elections suggest Republicans will face hurdles in the Midwest and Northeast against Clinton, but Trump argues he has unique appeal.
Gauging analytical quality on the basis of a single high-profile event may encourage misleading deference to previously correct analysts.
Including uncertainty levels is a key, but often neglected, part of creating sound predictive models.
Threshold events, such as a 'winner-take-all' primary or 'first-past-the-post' election, can see minor shifts lead to outsized outcomes.
