If the race between Bersani and Franceschini is a battle between the former Democratici di Sinistra and the Margherita (DL) respectively, then we should expect Bersani to do well in regions where the DS won a greater share of the combined vote of the two parties. Conversely, we would expect Franceschini to do well in regions where the Margherita did well relative to the DS.

It’s not that easy. The graph shows Bersani’s share of the two candidate vote against the DL share of the DL+DS vote in the Senate elections of 2006, the last elections for which we have separate vote totals for the DS and DL. If the DS-vs-DL story were true, we’d expect to see a downward sloping line.

Instead, we’ve got Bersani doing better than our (naive) expectations in Emilia Romagna and in Calabria, and poorer than expected in Lazio and Marche.

It’s not surprising that Bersani should do well in his home region (Emilia Romagna) — but then that’s Franceschini’s home region as well. I have no idea why he did so well in Calabria. One representative of Marino has denounced irregularities in the vote there [and here].

As for Lazio — well, Bersani did well in Rome, but Franceschini’s doing well in all those circoli outside the capital which add up, and his man Morassut went one better, winning more votes than the candidate for regional secretary affiliated to the Bersani motion, Alessandro Mazzoli.

For those of you who are playing along at home, take a look at the party statute and the electoral rules agreed by the party leadership in June.

The election of the national secretary happens in four stages [continues after the drop]/

In the first stage, registered party members in the several thousand party circles (circoli) vote for candidates to the provincial conventions. Each candidate will have a list or lists of candidates supporting him. This first stage has to happen by the 30th of September.

Each circle elects a number of delegates to the provincial convention. These delegates are elected using a largest-remainders electoral formula, using, as a quota, the Hare quota.

In the second stage, the delegates to the provincial conventions vote for delegates to the national convention. This means another round of voting, although here delegates’ preferences are well-known, so we can merely expect a rather mechanical (re-) application of the same largest-remainders electoral formula.

In the third stage, the party holds its national convention, composed of 1,000 delegates elected at the provincial conventions, plus a limited number of ex ufficio members.

In order to get to the third stage, a candidate needs to have at least 5% of all votes from the first stage and be one of the top three highest-polling candidates, or have at least 15% of votes.

At the national convention, delegates first vote on the motions of each candidate. This is essentially a show of strength — it has no formal bearing on anything that comes next. Candidates who have done well might, however, try and use this occasion to propose changes to the party’s statute, Ethical Code, or “Manifesto of Values”.

The national convention will be held on the 11th October.

Following this, the fourth stage involves primaries held across the country. These primaries elect the members of the National Assembly (1,000 delegates), who form an electoral college. Delegates are elected in constituencies which receive a number of delegates based half on population and half on the votes won by the party in the most recent general election.

This electoral college then goes on to vote for the candidate for secretary. If any candidate gets more than 50% + 1  of all votes, s/he is elected; failing this, there is a run-off. The primaries will be held on the 25th October.

To vote in the primaries, you need to give the party your name, address, date of birth, and, optionally, an e-mail address; agree that the party can use these details to contact you; and donate €2 to the party to cover costs. There will be polling stations all across Italy. Turnout may be slightly higher than the three million who turned out to elect Veltroni party secretary back in 2007.

The worst-case scenario, from an institutional engineering point of view, is this: all three candidates win more than 5% of the vote; one candidate wins the floor vote in the national convention; but the popular vote in the primaries goes the other way. The party’s national *Assembly* deadlocks, and the spoiler candidate directs his delegates to vote for the candidate who best accommodates his program.

What now looks increasingly likely: all three candidates (thus Marino) will make it through to the national convention, but Bersani will win at all stages.

	Bersani	Franceschini	Marino	Undeclared
DL	14	21	0	6
DS	40	19	2	23
FI	0	0	0	1
PSI	0	0	0	2
No prior experience	23	27	3	26

]]> http://chrishanretty.co.uk/blog/index.php/2009/09/22/bersani-winning-amongst-ex-ds-franceschini-with-ex-dl-and-newcomers/feed/ 0 http://chrishanretty.co.uk/blog/index.php/2009/09/15/getting-tables-out-of-pdfs-in-italy/ http://chrishanretty.co.uk/blog/index.php/2009/09/15/getting-tables-out-of-pdfs-in-italy/#comments Tue, 15 Sep 2009 12:42:12 +0000 Chris http://chrishanretty.co.uk/blog/index.php/2009/09/15/getting-tables-out-of-pdfs-in-italy/ The Italian Parliament annoys me tremendously. Not for substantial reasons (though it might also annoy me for that reason), but for technical reasons.

They have some nicely formatted XML files for the resoconti (minutes) of each parliamentary sitting.

But their voting information is stuck in crappy PDFs.

Grrr.

So, I have to

• download all the PDF files using a horrible bash script;
• convert them to XML (for file in *.pdf; do pdftohtml -xml "$file"; done)
• examine the XML file to find out where the column breaks are
• write a perl script to parse the files using this information

…and then merge them.

I’ve  now found one voting-intention poll, which shows Bersani in front (54%), Franceschini trailing (35%), with Marino putting in a respectable showing (11%). The sample size (N=651) is reasonable for a sub-sample, so it should be reasonably representative.

I doubt there’ll be many polls on the contest given the difficulties of reaching a representative sample of likely primary voters.

Bersani is also still leading in the “invisible primary” , the endorsements contest. From my running tally of Deputies who’ve declared a preference one way or the other, I find:

• 72 deputies in favour of Bersani
• 60 deputies in favour of Franceschini
• 4 deputies in favour of Marino
• 73 deputies undeclared

which, although it shows a closer race than last time, still shows Bersani out in front.

The leader will be chosen by a primary on the 25th October, preceded by a debate at the party’s congress two weeks prior. The congress will involve a vote on each candidate’s motion, but I have no idea whether the primaries will crown the winner of the vote at the party congress.

Although members of the Chamber of Deputies are only a part of the congressional delegates, they’re certainly an important group, especially in the “invisible primary”. So, in preparation for an article I’m writing together with Alex Wilson, I’ve decided to keep an eye on who they’re supporting. You can see the spreadsheet I’m using; if you have better information, add it in the comments below.

So far, the tally looks like this:

• 65 deputies supporting Bersani, with question marks surrounding Michele Ventura and Carlo Trappolino
• 47 deputies supporting Franceschini, with question marks surrounding Renzo Carella and Marco Causi (both are supporting Franceschini’s backed candidates in the elections for regional secretary, but I can’t find anything about support for Franceschini)
• 4 deputies supporting Marino
• 3 deputies who have declared that they have no preference
• 93 deputies for whom no info available/no declared preference

The electoral law of 2004 had a majority bonus, and differing electoral thresholds for electoral lists depending on the success of each list’s candidate for the presidency of the region. Ironically, the law was cited by the centre-right government when it embarked upon the reform of 2005 — citing a reform of a very centre-left region was good cover for the reform.

Now the major parties, plus some hangers-on — Pd, Fi-Pdl, An-Pdl, Alleanza Federalista and Partito Socialista — have decided that they’ll capitalize upon the reduction of party system fragmentation at national level, and increase the thresholds further. The threshold for those who run alone (or without a successful candidate for regional president) increases from 1,5% to 5%, and there’s a 4% threshold for all other parties.

Somewhere there’s a really nice paper to be written on the cross-level diffusion of electoral systems. It would require collecting a mountain of data however…

Working out the result of the 2001 Italian general election ‘as if’ the  2005 electoral system had been used is tricky for two reasons.

First, had the 2005 electoral system actually been in place, parties would have changed their strategies, and voters might have (re)acted differently. This holds for any exercise of this kind, in Italy or elsewhere.

Second, there were a number of changes of coalition and changes of party identity over this period. This is a particularly Italian problem.

Nevertheless, let’s assume that we can take the coalitions as they were in 2001, and the parties’ share of the proportional vote in 2001, and calculate a distribution of seats on the basis of the electoral law adopted in 2005.

Let’s start with the Chamber of Deputies. The coalition of the centre-right would have won the election, taking 49.5% of the vote and thus winning the majority bonus of 340 seats.

All five parties in this coalition — Forza Italia, Alleanza Nazionale, the Lega Nord, the UDC, and the Nuovo PSI — would have won seats, since the first four would have passed the 2% threshold for parties in a coalition, and the Nuovo PSI would have been admitted as the `largest loser’ in the coalition.

Within this coalition, the seats would have been divided proportionally (largest remainder, IIRC), and so Forza Italia would have had 202 seats (29.4 / 49.5 * 340), Alleanza Nazionale 82,  the Lega Nord 27, the UDC 22, and the Nuovo PSI 7.

Two other lists or coalitions would have won seats — the coalition of the centre-left (35%), and Rifondazione Comunista (5%). Rifondazione Comunista would have got a proportional share of the remaining 290 seats (in 2001 there were not the 12 seats for Italians living abroad), or 5 / 40 * 290, or around 36 seats. The centre-left would have taken the remaining seats.

The Radicals and Italia dei Valori would have (narrowly in the case of the latter) failed to reach the 4% threshold for lists not belonging to a coalition. They would not have won any seats.

For the Senate, it’s more difficult to say, since the parties in the two coalitions ran under a single label. From eye-balling the figures for each coalition, the centre-right would still have a majority in the Senate, since it would be roughly thirty seats short of a majority just on the basis of the majority bonuses in each region alone. Rifondazione Comunista would not have surpassed the 8% threshold in any region, but Italia dei Valori would have met that target in Molise.

I’m tremendously pleased that sites like this are starting to happen in Italy. The Camera actually has quite well-formated XML files for each of the plenary sessions, and all it needed was for someone to put that to good use.

So wonderful is the site that I feel churlish to pick on them. But just note that their graph of parliamentarians’ positions is based on multidimensional scaling, and as such is going to fall prey to the same problems as optimal classification analysis — extreme deputies in the governing coalition who rebel are going to be depicted as centrists, when they’re not.

Still, nice visualization.

In December 2005, Italy’s mixed-member electoral system was replaced with a system of bonus-adjusted proportional representation. The reform conformed with rational-choice models in that it was imposed by the ruling coalition, which sought to bolster its own power interests. But the case illustrates the impossibility of reducing such power-based motivation to a single goal, such as seat maximization. Power is shaped by many factors, and electoral systems influence many of these. This article develops a theoretical framework for understanding the various power-oriented considerations that may operate in electoral reform. It then analyses the role these played in Italy. It argues, in particular, for the need to take account of coalition dynamics when studying such processes.