The new electoral campaign for the Spanish Congressional elections starts tonight at 0.00 and... we are back! The 558project blog will provide projections of the electoral outcomes and updates when new information becomes available. We have improved our methodology although the basics are essentially unchanged. One thing has definitively improved: we have more data!!! You can read the steps of our methodology in the previous posts but here is a short summary.
Our methodology shares many features with Nate Silver’s method, which has been very sucessful in the prediction of US Presidential Elections. The method rests on two distinctive elements: what we call the fundamental model and a procedure to aggregate polls. The objective is to integrate all the available information (Censuses, surveys, polls, etc.) using a statistically sound methodology. The final estimation synthesizes the fundamental model and the information contain in all available polls. Both components are needed since polls by themselves, even if averaged, have biases and are less informative than one would hope. On the one hand the fundamental model is used to represent voting behavior and its evolution over time. On the other hand, the polls reflect voting intentions. They are aggregated with weights that depend on many factors and, especially, the forecast record of each pollster. Our approach is also characterized by its probabilistic nature. We then use Bayesian hierarchical models to synthesize data at provincial, regional and national level with polls. We use Bayesian updating to refresh the estimation over time when new information (for instance a new poll) becomes available.
However, the specific methodology we use to perform these tasks is quite different from Silver’s approach. We do not perform ad hoc corrections to the statistical model. In addition our method provides the weights to apply to the mixture of the forecast from the fundamental model and the aggregation of the results of polls (what Silver refers to as the “now-cast” or “snapshot”). Those weights change over time depending on the information content of the fundamental model and the polls updates.
By the way, we are not claiming that our method will provide perfect projections. What we claim is that you could get better projections than ours, by chance, using a particular poll or using your gut instinct, but you will not be able to do it systematically since our method uses all available information and blends it using sound statistical principles.
As of June 8, 2016 our projection for seats of the four largest parties is the following (the mark in the x axis corresponds to the median of the distribution)
Stay tuned for new analyses and updates.