I confess that getting my hands on polling data and seeing the names of political parties and leaders means for me to fight with a vague sense of nausea: we are in the final stages before the general elections, with declarations and promises worthy of a souk negotiation. Each side uses any weapon in order to gain voters and dismantle their opponents: there are those who enjoy themselves in this context, but I find the show not very edifying.
However, armed with pragmatism, I update the analysis of the political scenarios that emerge by aggregating, with the help of a little Bayesian statistics, the results of the public polls present on the site http://www.sondaggipoliticoelettorali.it and carrying out a real and its own electoral simulation.
We are now in the "third episode" of this analysis, so I won't dwell on the obvious importance of these elections, nor on the methodology, but I will limit myself to referring you to the previous posts: analysis of 23 January and analysis of 7 February.
This time, however, I am proposing the results of the electoral simulations immediately, updated with all the publicly available polls until the entry into force of the "level playing field".
The graph shows the probability distribution of the votes of the various political subjects estimated with an electoral simulation, using official polls as an information basis. The horizontal axis shows the possible results in terms of preferences granted and, for a given value, the higher the height of the "bell", the more likely it is. The three bells correspond to the simulation carried out on January 22nd, to the most recent one, on February 6th and finally to today's one, on February 20th.
In the upper part of the table you will find, for each coalition, the interval which, in 90 cases out of 100, includes the estimated electoral result, with its average value. For example: the first line “PD+SEL+…” tells us that in 90% of cases the centre-left coalition will get between 34,5% and 35,1% of the votes, that on average the value will be 34,8 % and therefore only in 5% of cases will there be results worse than 34,5% and better than 35,1%.
As you can see, compared to the last time, the situation has changed again. In one month we have witnessed an interesting evolution. The most probable political scenario has remained the same: victory for the centre-left, with a low (but not zero) probability of having a majority in both the Chamber and the Senate. Hence the need for a post-electoral coalition, probably with Monti. The estimated probability for this scenario remained around 95%.
What has changed?
As already mentioned last time, the number of surveys has grown, so the information should be considered more accurate. This is reflected in a "bell" with the narrowest range of possible outcomes. In other words, there is less uncertainty about the outcome, assuming the polls are reliable.
The question of questions: what will be the weight of Beppe Grillo's 5-star Movement?
The 5 Star Movement continues its advance. The great unknown is the undecided voters (about one in three): if their behavior should be very different from the average that emerges from the polls, there may be some nice surprises. Assuming that the "last second voter" leans more towards outsiders than major coalitions, the Movement could be the electoral surprise.
How plausible is Berlusconi's comeback?
Berlusconi's centre-right advances a little compared to the previous time. However, the PdL + Lega coalition stands 7-8 percentage points below the PD + SEL coalition, at least according to polls before the entry into force of the level playing field. If there hasn't been a rollover in recent days, it's hard to find a different situation after the polls close. Probabilistically, the comeback is unlikely.
Is Monti's coalition likely to finish fourth?
Indeed, the analysis shows that the coalition of the outgoing President is in fourth position, after the M5S. In any case, it seems that the result remains above the 10% threshold, a limit that would allow the coalition to access the Chamber of Deputies.
Is the scenario of a Bersani victory confirmed?
Bersani's centre-left is basically stable, a bit below 35%. Again according to the data from the surveys analysed, this is the coalition firmly in the lead in the preferences of Italians.
Quintum datur?
Ingroia's Civil Revolution falls below 5%, the other parties unfortunately have negligible numbers for them (again according to the aggregate of polls).
Conclusions
Undecided aside, the number of surveys I analyzed is several dozen. The conclusions, therefore, should be quite reliable. I know, I know: there are those who believe that the companies that carry out surveys are not all equally reliable, nor independent. Frankly, I think so too... however, statistics in these cases prove to be a good friend: by aggregating a multitude of polls carried out by different subjects, the "noise" should be smoothed out (the intuition is that a positive bias towards a coalition in one poll is compensated for by an opposite bias in another).
How stable will the government that emerges be, probably a centre-left + center coalition?
Impossible to give an answer now. If indeed the solution were a broad centre-left coalition + centre, the balance between Monti and Vendola would remain to be seen. But with the Italian economy which, in the second quarter of 2010 - third quarter of 2012, was the worst among those of the G7 member countries, it will be good for whoever wins to "do the right things" from the point of view of economic policy.
If there is someone with an insane passion for statistics, this is the link to the technical document that better explains the methodology used; the document is posted on my personal page of the Advise Only Community, just register (free of charge).
