As Primaries Affect Electoral Results. France and Italy in Comparative Perspec-
tive
Marino De Luca (University of Calabria) and Fulvio Venturino (University of Genoa)
Abstract.
Since some years the democratization of parties’ internal life is spreading in several
democracies. Among them, Italy and France have recently realized an impressive advancement.
Approaching the 2013 parliamentary election in Italy and the 2012 presidential
election in France, leftist coalitions have there launched open primary elections
to select their chief-executive candidates, Pier Luigi Bersani and François Hollande finally
resulting as nominees.
In this paper, we research whether the primaries’ levels of turnout and competition are
related with the following results of the presidential/parliamentary elections. We gather
the needed data referring to a comparable administrative levels: the Italian province and
the French départements. By so doing, we are able to reach the number of cases required
to implement an effective comparison.
General Conference
ECPR
Bordeaux, 4-7 September 2013
Panel 9H00-10H40, P034: Causes and Consequences of Primaries as Instruments for
Candidate Selection
Corresponding authors: deluca.marino@gmail.com, fventurino@unica.it
Contents
Primary elections and their critics................................................................................................1
Leftists go to primaries: the Democratic Party, the Socialist Party, and the selection of the
candidate chief-executive............................................................................................................3
Italy: towards the 2013 parliamentary election.........................................................................3
France: towards the 2012 presidential election .......................................................................4
Hypotheses and measurements..................................................................................................6
Results........................................................................................................................................8
A preliminary bivariate analysis ...............................................................................................8
Turnout, competition, and parliamentary elections: a (more) general picture......................... 10
Turnout and competition: what matters more? ...................................................................... 11
Discussion................................................................................................................................. 12
References................................................................................................................................ 13
1
As National Primaries Affect Electoral Results. France and Italy in Comparative
Perspective
Marino De Luca (University of Calabria) and Fulvio Venturino (University of Genoa)
Primary elections and their critics
Since a couple of decades a huge amount of comparative literature inform us that
parties are affected by a powerful crisis (Dalton and Wattenberg 2000; Dalton et al.
1984; Webb et al. 2002). Indeed, in democratic, Western-style countries they are shown
to be challenged by other kinds of political actors, such as social movements and new
forms of web-democracy, in representing the needs and interests of ordinary people. As
a consequence of these mounting antagonisms, researchers have detected, from the
1960s, a steady decrease in membership and turnout (Franklin et al. 1992; Franklin
2004).
Although declining, these “parties without partisans” are not inert when facing their
loss of legitimacy. Rather, they react to the challenges by adopting new forms of organization,
and promoting democratization in their internal life. These reforms have sometimes
been criticized as a manipulation of the members by the party leaders (Katz
2001). In any case, the role of the ordinary people has generally increased to affect all
the functions performed by the parties, including the selection of candidates and party
leaders.
Primary elections are one of the most noticeable reforms adopted by the parties to
expand the inclusiveness of their decision processes. In a nutshell, a party using primaries
accepts that its candidates and/or leaders are chosen by the members (closed primaries)
or by all the citizens (open primaries). A party is composed of several related
branches, a change – from oligarchy to democracy – regarding the political recruitment
predictably will affect the whole organization. The problem of the consequences brought
about by the use of primary elections in very complex, and not surprisingly it has been
initially researched in reference with the party politics in the United States. There the
earliest results have been quite pessimistic: parties adopting primaries are predestined
to be defeated in the following general elections, especially when primaries are competitive
and/or negative.
Today primaries are hardly a special characteristic of the American politics. At present,
this extraordinarily successful tool of democracy is also extensively used in Latin
America (Carey and Polga-Hecimovic 2006), Western Europe, Asia, and Africa (Ichino
and Nathan 2013). The expansion of the case studies has also made more problematic
the assessment of the political consequences produced by the primaries on the parties,
the party systems, and the political systems at large. In contrast with American researches,
primaries are now said to avoid major conflicts inside the parties, to improve
candidates’ image, and their fund raising capacities. Thus primaries became an asset
for electoral success, rather than the liability previously deplored.
This confused state of art is the outcome of plenty of empirical research sometimes
driven by approximate theoretical underpinnings. To obviate this undesirable condition,
recently Hazan and Rahat (2010) (HR) have put forward a framework aiming to order
2
and to standardize the research agenda on candidate selection methods. According to
their point of view, these methods differ in their levels of inclusiveness, and generate
noticeable consequences on four dependent variables, defined as participation, representation,
competition, and responsiveness.
The conclusions reached by HR are quite pessimistic. At the very least, no method of
selection can maximize party democracy on the four mentioned dimensions at the same
time. At best, one may obviate the most damaging consequences adopting a mixed system,
so collecting the best of four worlds. But probably the growth of inclusiveness is
meant to generate major pathologies, therefore party democracy is not working.
HR have been criticized as based on quite thin and sometimes anecdotal evidence.
Whatever the case may be, it is well known that the internal life of parties continues to
be a “secret garden” of politics, where transparency is usually lacking, and data gathering
is a difficult task. What is certain, is that all scholars operating in this fresh field of
research adopt a theoretical perspective strongly based on their work. For instance, Indriðason
and Kristinsson (2013) (IK) have recently issued an empirical examination of
the four dimensions proposed by HR. Although inadequately known and rarely quoted,
Iceland is one of the European countries where the primary elections have been adopted
earlier, and are today still deep-rooted. Thus IK may make use of an impressive dataset
covering the four major Icelandic parties in the parliamentary elections for the
whole period 1971-2009. Contrasting HR, they reach more positive conclusions about
primaries and their consequences. For instance, primaries in Iceland did not restrain
female representation, nor they did damage the cohesiveness of the parliamentary parties.
Instead they promoted the renovation of the parliamentary political class and
boosted up party membership.
In this paper we propose an examination of the political consequences of the primary
elections in Italy and France. Our general approach is alike the theory of HR and the
methods of IK, but several differences should also be noted. First, while IK explicitly focus
on a single country, and HR draw evidence in prevalence from the Israeli primaries,
we research leftist coalitions promoting primaries in a couple of countries. Second, both
HR and IK make use of evidence gathered during several years, while we focus on a
single primary election for each involved country. Third, while both HR and IK research
extensively all the four dimensions mentioned above, we focus only on two dimensions,
namely participation and competition. Fourth, neither HR or IK deal with primaries related
to parliamentary elections, while we have to do with the 2013 parliamentary election
in Italy and the 2012 présidentielle in France. Fifth, as neither Iceland or Israel use a PR
electoral system to elect their parliaments, HR and IK examine single parties selecting
through primaries a slate of several candidates; instead in the case of Italy and France
several parties associated in a coalition struggle to select a single candidate for the
chief-executive, monocratic role. Sixth, HR make use of national data and information,
while IK refer to the six districts disposed by the Icelandic electoral law. In both cases,
data are made available by national authorities. Instead analyzing Italy and France we
refer to an administrative sub-national level – respectively called provincia and département
– which is a territorial division irrelevant either for primary or general elections,
but pertinent in analyzing our cases.
3
In the next sections we detail our approach, sketch the data we analyze, and finally
we present our own results and their theoretical implications. But first we start our analyses
with two brief narratives about the Italian and French electoral processes under in-
vestigation.
Leftists go to primaries: the Democratic Party, the Socialist Party, and the selection
of the candidate chief-executive
Italy: towards the 2013 parliamentary election
After the demise of the whole party system in the nineties, the Italian parties reacted
by adopting opposite strategies. On the one hand, major rightist parties – Forza Italia,
then Popolo delle Libertà, and Lega Nord – were created as personal parties promoting
the political successes of their self-enthroned leaders. On the other hand, the leftist parties
reformed the procedures for selecting candidates and leaders by injecting a greater
inclusiveness of the involved selectorates. As a consequence, the leftist Italian parties
have become the most advanced promoters of the open primaries in Western Europe.
At the national level, the first national primary election has been launched in 2005 by
the centre-left coalition then choosing the candidate Prime Minister. In that case 4.3 million
of selectors turned out, and Romano Prodi won both the primaries and the 2006
parliamentary election. Moreover, since its inception in 2007, the Democratic Party (PD)
relied on primaries to select its first leader, Walter Veltroni (Pasquino 2009; De Luca
and Venturino 2010). In 2009 it came back again to primaries to select the following
leader, Pierluigi Bersani (Pasquino and Venturino 2010). And finally, Bersani himself on
June 2012 announced a coalition primary election to be held next autumn to choose the
candidate chief-executive for the 2013 general elections.
Thus, in November and December 2012 a leftist coalition, named Italia. Bene Comune,
was created through the agreement of four parties: the Democratic Party (PD),
Left Ecology Freedom (SEL), the Democratic Centre (CD), and the Socialist Party (PSI).
A council composed by parties’ officials screened the candidates, allowing five of them
to run. Although its statute clearly asserts that the party leader was the only possible
nominee, PD fielded three candidates, namely the incumbent leader Pier Luigi Bersani,
the mayor of Florence Matteo Renzi, and the regional councilor Laura Puppato1
. The
other candidates were the SEL incumbent president Nichi Vendola and the MP Bruno
Tabacci from CD. All citizens could turnout after a pre-registration, a signature to support
the coalition manifesto, and a contribution of two Euros.
As shown in Table 1, as no candidate secured a majority among the 3 million of selectors
attending the first round, after a week the two most voted, Renzi e Bersani, fielded
at the second round, and the latter finally succeeded by gaining 60.9% of the votes.
1
In precedence the National Assembly of the PD had temporarily amended the statute to allow other
candidates to run besides Pier Luigi Bersani.
4
Table 1. Italy 2012, results of the center-left coalition primary to select the candidate prime minister
First round Second round
Candidates N Valid vote % N Valid vote %
Pier Luigi Bersani 1,395,096 44.9 1,706,457 60.9
Matteo Renzi 1,104,958 35.5 1,095,925 39.1
Nichi Vendola 485,689 15.6
Laura Puppato 80,628 2.6
Bruno Tabacci 43,840 1.4
Total valid votes 3,110,211 100 2,802,382 100
On December 30th
the primaries for the premiership were followed by semi-open primaries
promoted only by PD and SEL to select the candidates to the parliament. This
prolonged cycle gained great momentum for the leftist coalition, nevertheless the results
of the February parliamentary elections have been disappointing. The coalition led by
Bersani in fact has been unable to reach a majority in both Houses, therefore in the following
months the coalition broke up, Bersani was induced to leave the party leadership,
and the PD was disgruntledly induced to enter a great coalition with rightist and
center parties, respectively led by Silvio Berlusconi and the former premier Mario Monti.
France: towards the 2012 presidential election
The sequence of Italian style primaries seems to have produced appealing results for
other European parties. Inclusive selections are extraneous to the French parties, but
the Socialist Party (PS) in 2006 organized a closed presidential primary election – then
gained by Ségolène Royal – reserved only to enrolled members (Dolez and Laurent
2007; Le Gall 2007). Since then, several articles by intellectuels, practitioners and political
theorists appeared in the French press, and the debate urged the Socialist Party to
adopt the Italian primary system. Finally, in 2009, a detailed report called Pour des
primaires ouvertes et populaires was produced by the socialist renewal secretary, Arnaud
Montebourg. Then the party officially adopted a primary framework similar to the
Italian Democratic Party. And later, approaching the 2012 présidentielle, it ran an open
primary explicitly inspired by the Italian experiences.
Table 2. France 2011, results of the center-left coalition primary to select the candidate president
First round Second round
Candidates N Valid vote % N Valid vote %
François Hollande 1,036,767 39.2 1,607,268 56.6
Martine Aubry 805,936 30.4 1,233,899 43.4
Arnaud Montebourg 455,536 17.2
Ségolène Royal 183,343 6.9
Manuel Valls 149,077 5.6
Jean-Michel Baylet 17,030 0.6
Total valid votes 2,658,667 100 2,841,167 100
The so-called primaires citoyennes held in October 2011 (Lefebvre 2011) have been
a coalition open primary organized by the Socialist Party and the Radical Party of the
5
Left using a two-round electoral system. For prospective voters the requisites to participate
were a pre-registration in an electoral list, a contribution of one Euro, and the signature
of the charter party values. As listed in Table 2, six candidates ran in the first
round. As no candidate reached 50% of the votes, the two most voted, Hollande and
Aubry, contested a second round, and François Hollande gained a nomination as presidential
candidate reaching 56.6% of the total vote. Successively, Hollande defeated the
incumbent Nicolas Sarkozy in the runoff of the presidential election, and after few weeks
the leftist parties secured a large majority in the Assemblée Nationale to support the
government led by the prime minister Jean-Marc Ayrault.
In summary, Table 3,shows the main differences between the regulation of the two
primaries.
1. in both cases, a run-off is necessary when no candidate had secured more than
50% of the votes in the first round;
2. both primaries are open, as neither party members nor non formally enrolled citizens
are allowed to vote;
3. a pre-registration is requested to vote only in the Italian case;
4. people less than 18 years old cannot vote in Italy, while they are allowed in France;
5. in France migrants can vote if they are party members, while in Italy they have more
limitations;
6. the lesser fee amounts to two Euros in Italy, but only one in France;
7. the signature is mandatory in both primaries, but it is doubled in the Italian case;
8. candidatures are associated with the party deputies in France, but with the simple
voters in Italy.
Table 3. Italy 2012 and France 2011, the rules of the primaries
Italy 2012 France 2011
1. Election
Two rounds, with a run-off in case no candidate
had secured more than 50% of the vote
Two rounds, with a run-off in case no candidate
had secured more than 50% of the vote
2. Participation All Italian citizens All French citizens
3. Registration
Be enrolled in the Italia. Bene Comune register
between 4
th
and 25
th
November 2012
Be registered in the electoral lists before 31
st
December 2010
4. Under 18 Not admitted
Admitted if 18 yo at the time of the 2012
presidential election; alternatively, admitted if
member of a party
5. Foreigners
EU citizens resident in Italy and citizens of
other countries in possession of a valid residence
permit
Admitted if member of a party
6. Contribution 2 Euros 1 Euro
7. Signature
Support for the call of the coalition Italia.
Bene Comune and for the Charter of Intents
Support for the values of freedom, equality,
fraternity, laity, justice, solidarity and pro-
gress
8. Candidature
Before 25
th
October 2012, 20,000 signatures
of voters of the center-left (no more than
2,000 in each region)
Either, endorsement from 5% of the Socialist
MPs (17 endorsements), 5% of members of
the National Council (16 endorsements), 5%
of regional councilors (100 endorsements)
from at least 10 département and 4 regions,
or 5% of socialist mayors of cities with more
than 10,000 inhabitants (16 endorsements)
from at least 4 regions
6
Hypotheses and measurements
Our general approach aims to shed light on the eventual relationships linking primary
to general elections. Of course, a number of provisions are still needed to actually implement
such a broad statement. Because we adopt quantitative techniques of research,
we sketch these provisions using the language of statistics.
First of all, as independent variables we focus on two characteristics of the primary
elections: turnout and competition. In our view, these variables are related to the electoral
performances of the promoting parties in an overall intuitive way. Those parties will
reach the best results in the general elections when the primaries have first featured a
high turnout and a low competition. In both cases the rational is easy to understand.
High levels of turnout at the primaries show that involved parties are fielding appealing
candidates. Furthermore, a high number of activists is important to fight a “post-modern”
centralized campaign (Norris 2000), and to have a positive image in the public opinion.
So that the parties could even start a “spiral of silence” (Noelle-Neumann 1993) pushing
them towards an effortless electoral success. In a nutshell, crowded primaries mobilize
a great number of citizens who can be trusted to participate in the following general
elections.
While high levels of turnout are alleged to boost parties, primaries featuring high levels
of competition may damage them. This hypothesis is well entrenched in individuallevel
theory (Anderson et al. 2005). Exceedingly competitive primaries entail the selection
of a nominee gaining a small margin of votes on his challengers. In turn, this involves
that a great number of selectors have seen their favorite candidate eliminated,
leaving them two alternatives when coming close to the general election. Disgruntled
but loyal voters will support the nominee from their party as a second best. But those
completely disappointed will desert the polling station, and sometimes may vote against
their own supposed candidate. Thus, the greater the competition at the primary elections,
the greater the number of potential deserters, the lesser the amount of votes
gained by the nominee at the general election.
Then turnout and competition are our main independent variables, however both may
be measured according to different procedures. Our choices are as follows. For turnout
at the primary elections, we calculate the percentage as a ratio between the number of
selectors and the number of people enfranchised for the general elections2
. Usual practice
is to refer to the number of votes gained by the nominee at the following general
elections. As will be clear below, we disregard this solution in order to avoid problems of
reciprocal causality in our analyses.
Also for competition, we make use of several different indexes (Kenig 2008). To assess
here the levels of competition in the Italian and French primaries we prefer to use
the classical effective number of candidates proposed by Laakso-Taagepera (1979).
The reason is that we have to deal with multicandidate primaries, and this index takes in
account the electoral performances of all candidates. Thus it seems to be more practical
2
The results of the Italian primaries have been released without the information concerning blank and
invalid ballots. Thus our calculations here are based on valid votes rather than the number of selectors.
7
in comparison with other indexes using only the result of the front runner, or the difference
between the two best competitors, i.e. the closeness3
.
Besides these two main predictors of the performances at the general elections, we
make use of a couple of control variables to detect the possible existence of spurious
relationships. As the result of a party in a given election is necessarily influenced by the
result of that party in the previous election, we consider the percentage of votes gained
by the main leftist party in 2007 and 2008 for the French and Italian cases respectively.
By so doing, we aim to control for autoregressive effects created by the stability of the
vote in a particular territory. Finally, we consider GDP per capita as a proxy to tap the
level of economic development.
Some additional information about our research design is still needed. To begin with,
neither in Italy nor in France the primaries to select the candidates for the role of chief
executive have been held with a two-round system. In both cases, our analyses make
use of data pertaining to the first round. Then, as anticipated in the first section, our territorial
unity of analysis are province for Italy and départements for France. We have
chosen this level because primary promoters used these levels to spread over the results
of their primaries4
, because it makes it easy to match the results of the primary
and the general elections, and finally because it permits us to analyze one-hundred
cases both in France and in Italy, so escaping the statistical problem of “small N”.
Finally, we have to describe our dependent variable(s). The starting point here is that
we are analyzing two multiparty systems, where the leftist wings are highly fragmented,
and also arranged in pre-electoral coalitions largely spurred by the electoral laws5
.
Thus, once decided that we are interested in examining the electoral results of the parties
after their respective primaries, we are left undetermined which parties we are talking
about. And we have no theoretical guide to pick out in advance the parties most interesting
for us. So, we start the next section by a preliminary examination encompassing
all the above mentioned parties. In practice, this means that we consider the coalitions
operating during the 2013 and 2012 parliamentary elections. Then in the case of
Italy we investigate the three main parties composing the coalition Italia. Bene Comune,
all of which had presented their own candidates at the primary elections: the Democratic
Party (PD), Left Ecology and Freedom (SEL), and the Democratic Center (CD)6
. For
France, we will carry out a preliminary analysis of the leftist coalition and of the following
parties: the Socialist Party (PS), the Radical Party of the Left (PRG), and the Greens
3
Kenig (2008) has proposed a new index where the effective number of candidates is weighted by the
sheer number of competitors. Although promising, this index would be redundant to be used here. This is
because we compare several districts in a single primary election where the (sheer) number of candidates
is a constant rather a variable.
4
The reader should be aware that so far Italy and France did not have adopted any national regulation,
and therefore in both countries the primaries are a voluntary and self-regulated activity.
5
Since 2006 Italy uses a mixed electoral system having a proportional structure; for parliamentary
elections France employs a classical two-round system since 1958, with a single exception dating back to
1986.
6
The whole coalition was formed by other minor partners. At the election for the Lower House (Camera
dei Deputati) the Sud Tiroler Volkspartei (SVP, the regional party of the Deutsch-speaking minority)
gained .43% of the total vote and 5 parliamentary seats. In the case of the Upper House (Senato della
Repubblica) Il Megafono-Lista Crocetta, the Socialist Party (PS) and i Moderati won together .67% of the
votes and no parliamentary seats.
8
(Europe Écologie-Les Verts, EELV)7
. In all these cases, the results for the Lower Houses
are considered.
Results
A preliminary bivariate analysis
As mentioned above, this work aims to test whether participation and competition in
primaries elections have influenced the performance of the left-leaning coalition in the
following general elections. For this purpose, the first step is to identify whether some
correlations actually exist among our variables.
Let us start with the Italian case. As shown in the first row of Table 4, the primary
turnout is strongly related to the votes of the coalition-left and the Democratic Party in
the 2013 parliamentary election, and the value of statistical significance is very high. On
the contrary, it is not related with the SEL and the Democratic Centre performances.
Table 4. Italy: correlation matrix between primary turnout and vote the center-left parties
Primary
turnout
Vote to centerleft
coalition
Vote to Democratic
Party
Vote to Left
Ecology Free-
dom
Vote to Democratic
Centre
Primary turnout 1 .740 (.000)** .868 (.000)** .174 (.070) -.037 (.701)
Vote to center-left
coalition
1 .746 (.000)** .331 (.000)** -.033 (.732)
Vote to Democratic
Party
1 .002 (.982) -.172 (.073)
Vote to Left Ecology
Freedom
1 .454 (.000)**
Vote to Democratic
Centre
1
Note: significance level is presented in brackets.
** Correlation is significant at the 0.01 level (2-tailed)
Table 5 shows the same correlations for the French case. Here, primary turnout has
a positive correlation to the votes obtained by the coalition and the Socialist Party in the
2012 parliamentary election. On the contrary, it is not related to the PRG and Green
Party vote. In comparison with Italy, the scores are slightly lower, and the correlation of
the coalition is larger than that of the main party.
7
PS and PRG presented respectively five and one out of the six candidates which contented the primaries.
We do not deal with few leftist independent candidates fielded in several uninominal districts.
They are indicated as “miscellaneous left” (DVG, Divers de Gauche), gained 3.40% of the votes in the
first round, and a final amount of 22 parliamentary seats.
9
Table 5. France: correlation matrix between primary turnout and vote the center-left parties
Primary
turnout
Vote to centerleft
coalition
Vote to Socialist
Party
Vote to Radical
Party of the
Left
Vote to the
Greens
Primary turnout 1 .675 (.000)** .498 (.000)** .006 (.953) .072 (.488)
Vote to center-left
coalition
1 .734 (.000)** .062 (.546) -.073 (.480)
Vote to Socialist
Party
1 -.462 (.000)** -.143 (.164)
Vote to Radical
Party of the Left
1 -.209 (.041)*
Vote to the Greens 1
Note: Significance level is presented in brackets
** Correlation is significant at the 0.01 level (2-tailed)
* Correlation is significant at the 0.05 level (2-tailed).
Matrices presented in Tables 6 and 7 show the correlations among the primary competition
and the electoral performances of the leftist parties. As expected, negative correlations
prevail. In Italy (see Table 6, first row), they may be detected in three out of
four cases, and the only exception (SEL) features a non-significant score.
Table 6. Italy: correlation matrix between primary competition and the vote for center-left parties
Primary
competition
Vote to centerleft
coalition
Vote to Democratic
Party
Vote to Left
Ecology Free-
dom
Vote to Democratic
Centre
Primary competi-
tion
1 -.412 (.000)** -.495 (.000)** .046 (.636) -.215 (.024)*
Vote to center-left
coalition
1 .746 (.000)** .331 (.000)** -.033 (.732)
Vote to Democratic
Party
1 .002 (.982) -.172 (.073)
Vote to Left Ecology
Freedom
1 .454 (.000)**
Vote to Democratic
Centre
1
Note: significance level is presented in brackets.
** Correlation is significant at the 0.01 level (2-tailed).
* Correlation is significant at the 0.05 level (2-tailed).
For France (see Table 7, first row) the table is slightly less favorable to our hypotheses.
There is a negative relationship between the primary competition and the vote in
the parliamentary election, which is valid only in two out of four cases with an intermediate
level of significance. Moreover, the two deviating cases (PRG and Greens) also feature
a medium level of significance, which was lacking for the Italian deviant case. Both
in France and in Italy, the negative consequences brought about by high levels of primary
competition are stronger for the largest party rather than the coalition.
10
Table 7. France: correlation matrix between primary competition and the vote for center-left parties
Primary
competition
Vote to centerleft
coalition
Vote to Socialist
Party
Vote to Radical
Party of the
Left
Vote to the
Greens
Primary competi-
tion
1 -.250 (.014)* -.348 (.001)* .253 (.013)* .229 (.025)*
Vote to center-left
coalition
1 .734 (.000)** .062 (.546) -.073 (.480)
Vote to Socialist
Party
1 -.462 (.000)** -.143 (.164)
Vote to Radical
Party of the Left
1 -.209 (.041)*
Vote to the Greens 1
Note: Significance level is presented in brackets
** Correlation is significant at the 0.01 level (2-tailed)
* Correlation is significant at the 0.05 level (2-tailed)
Let us briefly compare the above information regarding turnout and competition. The
scores referred to turnout are larger than those pertaining to competition. Thus, the
electoral advantages produced by (high levels of) turnout seem to be larger than inconveniences
caused by (high levels of) competition. However, in order to fully understand
the above tables, it is necessary to remember that those bivariate correlations are powerless
when facing possible spurious relationships. Therefore, in the next sections, we
review these preliminary results through a multivariate analysis incorporating other factors
supposedly related to electoral performances. By so doing, for a sake of clearness
we focus only on the Italian and French center-left coalitions.
Turnout, competition, and parliamentary elections: a (more) general picture
In this section, we discuss the short-term electoral effects of the primaries also considering
the long-term parties’ strength and the economic level of the involved territory.
To this aim, building on previous analyses, our models now present two additional control
variables:
 the vote for the Democratic Party and for the Socialist Party in the parliamentary
election preceding those under examination;
 the GDP per capita8
.
Everywhere the electoral performance of a party is subjected to a path dependency,
and in several Italian and French provinces the vote features small swings. For this reason,
we make use of the vote obtained by the PD and PS in the previous parliamentary
elections as a control9
. Next, it is well known that turnout – at general and primary elections
– is higher where economic levels are elevated. Moreover, leftist parties today
8
Related data have been collected by API (2010) for Italy and by INSEE (2010) for France.
9
By so doing, we control the electoral result of a coalition using the previous result of the main party of
that coalition. This is not the best strategy, but the most obvious alternative – to use the previous votes
gained by the coalition – is made unviable by the dramatic changes intervened between the two consecutive
elections.
11
may prevail in the wealthier zones. Thus, using GDP as a control variable we aim to
check for effective, not spurious, relationships between our main variables.
Table 8. Primary turnout and other predictors of coalition voting: Italy and France
Italy France
b s.e. b s.e.
Constant 9.338 3.037 23.727 1.933
Primary turnout .863 .190 2.179 .324
Vote to PD 2008/PS 2007 .253 .088 .363 .067
GDP per capita .000 .000 .000 .000
R square .603 .680
N 110 96
To start with, we regress the vote for the coalition in Italy and France against primary
turnout and the two above mentioned control variables. Results reported in Table 8
show that in both cases control variables explain a substantial amount of the coalition
support10
. More importantly, primary turnout continues to be a key predictor even after
controls are considered, and the R squares high levels prove the fitness of the model
overall.
Table 9. Primary competition and other predictors of coalition voting: Italy and France
Italy France
b s.e. b s.e.
Constant 17.029 6.114 24.892 5.752
Primary competition -4.986 1.772 -.124 1.461
Vote to PD 2008/PS 2007 .491 .063 .639 .068
GDP per capita .000 .000 -.000 .000
R square .559 .522
N 110 96
Similarly, in Table 9 the coalition performances are regressed against competition
and control variables. For Italy, the table maintains the same rationales: all variables are
significant, and contribute to a substantial fitness of the model. Not so for France. In this
case, an autoregressive effect prevails, and the 2012 vote is mostly explicated by the
2007 vote and the other variables – primary competition included – being irrelevant.
Summing up: after the control variables are included, turnout continues to boost the
vote of the leftist coalition in both Italy and France. Not so for competition. It damages
the coalition in Italy, while in France its effects disappear in face of the stability of the
vote. Once Italians were strong identifiers, while French parties were unable to develop
strong ties to their supporters. Today things seem to have changed noticeably.
Turnout and competition: what matters more?
So far we have discovered that primary turnout improves parties’ fortune, while primary
competition plays a damaging role. We have also incidentally noted that the positive
effects of the turnout seem to be greater than the negative effects of the competition,
but on this point we rely hitherto on cursory and unsatisfactory evidence. Now we
make use of a new analysis to address this crucial point with greater accuracy.
10
As shown by the little score of the standard error in comparison with the score of the regression co-
efficient.
12
Table 10 details the results of two regressions in Italy and France. In both cases, the
variables are the same as the previous analyses, the major enhancement being the
concurrent use of primary turnout and competition as predictors. In order to assess the
relative weight of each independent variable, scores reported in the table are standardized
regression coefficients.
Table 10. Turnout vs. competition as predictors of coalition voting: Italy and France
Italy France
Primary turnout .417 .549
Primary competition -.131 .021
Vote to PD 2008/PS 2007 .272 .415
GDP per capita .206 -.247
R square .616 .680
N 110 96
Note: the values are standardized regression coefficients (beta).
In both cases, primary turnout has the “right” sign, and is the best predictor of the coalitions’
electoral performances. This appears to be a trustworthy substantiation of the
positive role of the open primaries when robustly attended by the citizens. The case for
the competition in less clear. In Italy, it presents a negative sign – i.e., it depresses the
vote for the coalition, as expected – but is the least relevant among all the considered
variables. In France instead it is not relevant at all. In any way, parties should not be too
anxious about the presence of several viable candidates. The votes gained in the previous
elections in both cases are very important predictors, being the second after primary
turnout. At a glance, what is striking is the lesser role played by the path dependency
in Italy. But one should be aware of the astonishing results of the critical 2013 parliamentary
election, when the newcomer Movimento 5 Stelle won 25% of the votes. We
did not have clear expectations about the role of the GDP. Now, we note that in both
cases it is quite an important predictor, and that it shows a different sign for Italy and
France. Thus, we are informed that leftist Italian parties gain more votes in the richer
provinces, while in France la Gauche plurielle is more supported in the underprivileged
dèpartements.
Discussion
As the primary elections have been only recently brought into play outside the United
States, the cases for research in Europe are not many. There are several research
strategies to overcome the ensuing constraints. For large countries, it is possible to divide
the national territory in some partitions – based on poll stations, electoral districts,
or administrative precincts – in order to create a sufficient number of cases to avoid the
“small N” problem. In this paper we have compared two large countries, Italy and
France, so the most straightforward approach entailed the use of data collected at the
level of province and départements. After evaluating a sufficient number of cases to apply
statistical techniques, we have reached some intriguing results.
First of all, a preliminary analysis demonstrated that the impact of the primaries is
relevant only for major parties. Although we lack a satisfying account, niche parties appear
to be perceived by voters in such a way that their involvement in the primaries
13
simply does not influence their success in the general election. In addition, the wideranging
analyses on major parties and coalitions have confirmed simple intuitions. The
citizens’ participation in the primaries is a great starting point to be successful in general
elections. Also the theoretical insight questioning the commitment of the followers of the
losing primary candidates is supported by our data. Finally, as people’s involvement
matters more than the underdogs’ disappointment, the votes for those parties embracing
the democratization of the candidates’ selection will be enhanced.
Often party officials are afraid because they believe primaries damage their parties.
Thus, if they cannot avoid them, they look at least for a low number of candidates and a
restricted competition. If our conclusions are right, they should reverse their conservative
strategies: more primary candidates, more competition, more turnout, and finally
more ballots in the polls. Perhaps they are rather afraid for their own careers. But this is
another story.
References
Anderson, C.J., Blais A., Bowler S., Donovan T., Listhaug O. 2005. Losers’ Consent. Elections
and Democratic Legitimacy. New York: Oxford U.P.
Carey, J.M., Polga-Hecimovic J. 2006. Primary Elections and Candidate Strength in Latin America.
The Journal of Politics 68 (3), 530-43.
Dalton, R.J., Wattenberg M.P. (eds.) 2002. Parties without Partisans. Political Change in Advanced
Industrial Democracies. Oxford: Oxford U.P.
Dalton, R.J., Beck, P.A., Flanagan, S.C. (eds.) 1984. Electoral Change in Advanced Industrial
Democracies. Realignment or Dealignment?. Princeton: Princeton U.P.
De Luca, R., Venturino, F. (eds.) 2010. Mobilitazione e partecipazione elettorale. Un’analisi
delle “primarie” per l’elezione del primo segretario del Pd. Rome: Aracne.
Dolez, B., Laurent, A. 2007. Une primaire à la française: la désignation de Ségolène Royal par
le Parti Socialiste. Revue Française de Science Politique 57 (2), 133-61.
Franklin, M.N. 2004. Voter Turnout and the Dynamics of Electoral Competition in Established
Democracies since 1945. Cambridge: Cambridge U.P.
Franklin, M.N., Mackie, T., Valen, H. (eds.) 1992. Electoral Change. Responses to Evolving Social
and Attitudinal Structure in Western Countries. Cambridge: Cambridge U.P.
Hazan, R.Y., Rahat, G. 2010. Democracy Within Parties. Candidate Selection Methods and
their Political Consequences. Oxford: Oxford U.P.
Ichino, N., Nathan, N. 2013. Do Primaries Improve Electoral Performance? Clientelism and Intra-Party
Conflict in Ghana. American Journal of Political Science (57) 2, 428-41.
Indriðason, I.H., Kristinsson, G.H. 2013. Primary Consequences: The Effects of Candidate Selection
through Party Primaries in Iceland. Party Politics, forthcoming.
Katz, R.S. 2001. The Problem of Candidate Selection and Models of Party Democracy. Party
Politics 7 (3), 277-96.
Kenig, O. 2008. Democratization of Party Leadership Selection: Do Wide Selectorates Produce
More Competitive Contests?. Electoral Studies (28) 2, 240-47.
Laakso, M., Taagepera, R. 1979. “Effective” Number of Parties. A Measure with Application to
Western Europe. Comparative Political Studies (12) 1, 3-27.
Lefebvre, R. 2011. Les primaires socialistes. La fin du parti militant. Paris: Raison d’Agir.
14
Le Gall, G. 2007. Le primaire au PS: l’irrésistible ascension présidentielle de Ségolène Royal.
Etat de l’Opinion, Septembre 2005-Décembre 2006, 47-74.
Norris, P. 2000. A Virtuous Circle: Political Communications in Postindustrial Societies. Cambridge:
Cambridge U.P.
Noelle-Neumann, E. 1993. The Spiral of Silence: Public Opinion – Our Social Skin. Chicago:
University of Chicago Press.
Pasquino, G. (ed.) 2009. Il Partito Democratico. Elezione del segretario, organizzazione e potere.
Bologna: Bononia U.P.
Pasquino, G., Venturino, F. (eds.) 2010. Il Partito Democratico di Bersani. Persone, profilo e
prospettive. Bologna: Bononia U.P.
Webb, P., Farrell, D.M., Holiday, I. (eds.) 2002. Political Parties in Advanced Industrial Democracies.
Oxford: Oxford U.P.