en

Caution, rigor, luck and succesful predictions

30 Jun, 2016 - - @politikon_es

Six months ago, as the results from the election were made public, I published a piece that contained several explicit predictions. As time stands, two of them have already been clearly confirmed, namely, that Podemos would not accept to be part of any government in which it would not be the senior partner and that, as a consequence of this, the probability of new elections was rather high. This analysis was further elaborated in a subsequent piece in which I explained how the room for agreement being small, the fluidity of votes being high, key actors had advantages for refusing any deal; an analysis that I further elaborated for the case of Podemos here.

Beyond being right about the outcome, which is objectively clear, I believe that the reasons I provided to support my arguments have been rather confirmed than disconfirmed, that is, I was right for the right reasons -although, to be honest, I expected the issue of Catalan secession to play a bigger role in the agenda and Ciudadanos a smaller one. I would like to enumerate some conclusions.

Applying social science in a cautious manner can actually be effective. I made my diagnosis, which remained accurate over time, in the presence of the information that was available on the election night based on public information. Anyone with a modest understanding of the Spanish political context and access to Google Scholar could have come to the same conclusions. Using my knowledge of the literature on multiparty governments, I proceeded to locate events, assign probabilities to them, combine them into scenarios and define outcomes.

While most forecasters like to take credit when they are right as a reflection of their superior skills, either for profesional or ego reasons, I won’t engage in such activity here- I don’t think I am particularly smart or insightful. In fact, I would like to argue the opposite. As it is clear from my the way I formulated my predictions -assigning probabilities, using words like ‘likely’, ‘rather’ or ‘perhaps’ often, words that are too often absent from newspaper headings-, while being informative, I was uncertain about them – that is, luck played a big role. This is mainly due to unobservable factors (we know, for example, little about the internal structure of parties) about which ‘experts’ can say little.

Related to the above, while news cycle analyst and journalists seem to be interested in real time analysis in which everything can change at any point in time following meeting actors, declarations, it sometimes makes substantial sense to ignore this information and it definitely did in the case of this election. The reason is that it is hard to discern to which extent this factors are signal or noise. In the case of bargaining the formation of a government, most parties have the incentive to send the signal that they are open to be part of it in order to avoid being blamed by voters. Ciudadanos, for example, reached a deal the socialist party that clearly did not have the support to become the basis for a government. Perhaps they honestly hoped that this would become the coalition agreement, but if they did not, their dominant strategy was to appear as favoring the agreement so they could avoid fading into irrelevance. The same goes for Podemos, who rejected to be part of the agreement and suggested counter-proposals that were clearly beyond the red line for the socialist party. Each time one of these came up, it could be interpreted as a real willingness to be part of an agreement -but in my view reflected an attempt shift the blame for no agreement and go again to the ballot, as strategy that, in retrospect, has paid off. From this point of view, focusing on stable and more fundamental factors whose importance is well understood, arguably makes more sense.

An implication of the above is the role that social scientists can play in public discussion. Political science looks at empirical regularities and theoretical puzzles and tries to solve them, which equips scholars with the capacity to deliver information on trends and frame the discussion. They can, in fact, do this each time something unexpected occurs. It is hard, however, to see how they can interpret micro-events better than the average person beyond suggesting, as I did here, that they are perhaps ignorable or that not much can be said about them (I conclusion that is rarely welcome).