When personalization meets direct democracy An analysis

When personalization meets
direct democracy
An analysis of the determinants of voting in
the Italian Constitutional referendum
Diego Garzia (University of Lucerne)
Paolo Bellucci (University of Siena)
Andrea De Angelis (University of Lucerne & EUI)
Università di Roma “Sapienza”
10 Febbraio 2017
Point of departure
• Opinion formation in direct democratic
decisions (Kriesi, 2005)
• Systematic path (based on arguments)
• Heuristic path (parties, governments, etc.)
The legacy of the 2nd Italian Republic
• Pervasive personalization of politics
– Political communication in old and new media
– Personal parties and personal leaders
– Growing leader effects on political attitudes
(partisanship) and voting behaviour
– The “primacy of the political supply” hypothesis
Research Question
• What is the factor(s) that best explain the
reason why Italian voters voted in favor (or
against) the reform?
Qual è la principale ragione per cui ha votato SI?
perche' temevo le conseguenza economiche di una
vittoria del No
perche' ho seguito le indicazione del partito a cui
mi sento piu' vicino
per esprimere sostegno al Presidente del consiglio
Renzi e al suo governo
perche' approvo il contenuto della riforma
costituzionale
0,0
0,1
0,2
0,3
0,4
Fonte: Inchiesta Post-Elettorale ITANES, Dicembre 2016
0,5
0,6
0,7
Qual è la principale ragione per cui ha votato NO?
perche' ho seguito le indicazione del partito a cui
mi sento piu' vicino
per oppormi al Presidente del consiglio Renzi e al
suo governo
perche' non sono d'accordo con la nuova legge
elettorale (Italicum)
perche' non approvo il contenuto della riforma
costituzionale
0,0
0,1
0,2
0,3
0,4
Fonte: Inchiesta Post-Elettorale ITANES, Dicembre 2016
0,5
0,6
0,7
End of the story?
• No… we suspect that social desirability bias is
playing a role
End of the story?
• Indeed… social desirability bias is playing a
role
• We move from self-assessments to actual
relationships between explanatory factors
Research Hypotheses
• H1: Voting choice is better explained by reliance
on decisional shortcuts as compared to (cognitive
more demanding) evaluations of the actual
contents of the Constitutional reform.
• H2: Based on the “personalization of politics”
theory, we expect that – among all decisional
shortcuts – voters’ approval of the prime
minister is the best predictor of their vote choice
in the referendum
Operational Measures
• Systematic path
– Evaluation of Italicum
– Evaluation of Constitutional Reform’s main features (reduction
number senators, end of symmetric bicameralism,
centralization, reduced quorum for referendums)
• Heuristic path
– Evaluation of national economic conditions (retro/socio)
– Evaluation of government’s action on selected issues (economy,
Jobs Act reform, immigration, school system's reform, tax
reductions)
– Strength of (PD) partisanship
– Evaluation of Matteo Renzi (Thermometer)
From bivariate to multivariate evidence
• Every single measured taken into account is
significantly related to vote choice
• Strong patterns of covariance, to be ruled out
in a multivariate setting
• What is the model that best explains vote
choice in the 2016 referendum?
Modeling Strategy
• Multinomial logistic regression: base category
abstention Vs. YES / NO in turn
• Coefficients are odds ratios
• All models (including baseline model M0)
control for age, gender and education level
Model fit statistics
85%
80%
75%
70%
65%
60%
55%
50%
baseline systematic heuristic composite
economy partisanshipgovernment leader
% of correctly predicted cases by each model
Model fit statistics
85%
H1 CONFIRMED
H2 CONFIRMED
80%
75%
70%
65%
60%
55%
50%
baseline systematic heuristic composite
economy partisanshipgovernment leader
% of correctly predicted cases by each model
Le Ragioni del NO
(0)
Evaluation of Italicum
Evaluation of the Constitutional Reform
Reduction number senators
End symmetric bicameralism
Centralization
Reduced quorum for referendums
(1)
0.91
(3)
(4)
(5)
(6)
1.02
0.90*
0.90*
1.04
.04
55%
3027
.51
77%
2313
(7)
0.92
1.02
0.91
0.94
1.07
0.92
Retro/Socio Economic Evaluation
Assessment of Renzi Government
Economy of R's local area
Jobs Act reform
Immigration
School system's reform
Tax reductions
Closeness to PD
Thermometer Evaluation: Renzi
Pseudo R-Squared (Nagelkerke)
% Correctly Predicted
N
(2)
0.76**
0.92
1.07
1.05
1.04
0.95
0.94
0.98
1.00
0.94
0.97
0.81
0.89*
.52
78%
2666
0.97
0.91
0.95
1.00
0.98
0.99
0.75
0.85**
.62
81%
2151
0.62**
.24
65%
2988
.43
75%
2725
.27
69%
3027
0.84**
.47
76%
2918
Le Ragioni del SI
(0)
Evaluation of Italicum
Evaluation of the Constitutional Reform
Reduction number senators
End symmetric bicameralism
Centralization
Reduced quorum for referendums
(1)
1.69**
(3)
(4)
(5)
(6)
1.05
1.31**
1.18**
1.09
.04
55%
3027
.51
77%
2313
(7)
1.46*
1.07
1.24**
1.09
1.05
1.09
Retro/Socio Economic Evaluation
Assessment of Renzi Government
Economy of R's local area
Jobs Act reform
Immigration
School system's reform
Tax reductions
Closeness to PD
Thermometer Evaluation: Renzi
Pseudo R-Squared (Nagelkerke)
% Correctly Predicted
N
(2)
2.74**
1.03
1.05
1.10
0.99
1.22**
1.33**
1.13
0.91
0.97
1.03
1.83**
1.26**
.52
78%
2666
0.94
1.03
1.11
0.95
0.94
1.08
1.62*
1.15*
.62
81%
2151
3.18**
.24
65%
2988
.43
75%
2725
.27
69%
3027
1.57**
.47
76%
2918
Conclusions
• Heuristics mattered more than content
– Inner complexity of the reform or response to the
personalization imposed by the supply side?
• NO voters just disliked Renzi
• YES voters liked Renzi, but also Italicum, certain
features of the reform (and PD)
• Can we single out a single factor? And if yes, how???