Hostname: page-component-76d6cb85b7-lcgwf Total loading time: 0 Render date: 2026-07-26T12:02:28.182Z Has data issue: false hasContentIssue false
  • English
  • Français

Waving the same flag? Government legislation and obstructionism during the COVID-19 crisis in Italy

Published online by Cambridge University Press:  03 January 2025

Paolo Gambacciani Open the ORCID record for Paolo Gambacciani [Opens in a new window] ,
Andrea Pedrazzani Open the ORCID record for Andrea Pedrazzani [Opens in a new window]  and
Luca Pinto Open the ORCID record for Luca Pinto [Opens in a new window]
Paolo Gambacciani*
Affiliation:
Department of Political and Social Sciences, University of Bologna, Bologna, Italy
Andrea Pedrazzani
Affiliation:
Department of Social and Political Sciences, University of Milan, Milano, Italy
Luca Pinto
Affiliation:
Department of Political and Social Sciences, University of Bologna, Bologna, Italy
*
Corresponding author: Paolo Gambacciani; Email: paolo.gambacciani@unibo.it

Abstract

This study examines the amendatory activities of the majority and opposition parties in the Italian 18th legislature (2018–2022) in response to the COVID-19 pandemic crisis. Following the rally around the flag hypothesis, we test whether both sides exhibited similar legislative behaviour during emergencies. We exploit an original database covering amendments tabled by Italian legislators on bills converting decree-laws. Results reveal that the COVID-19 pandemic affected amendment activities without aligning majority and opposition behaviours. In other words, the opposition did not pull in the same direction of the government legislation. This can be explained by contingent factors and pre-existing party polarization.

Keywords

comparative politicsgovernmentItalyparliamentpolitical parties

Information

Type
Research Article
Information
Italian Political Science Review / Rivista Italiana di Scienza Politica , Volume 56 , Special Issue 1: The politics of polycrisis: institutions and political actors in turbulent times , March 2026 , pp. 9 - 23
Creative Commons
Creative Common License - CCCreative Common License - BYCreative Common License - NC
This is an Open Access article, distributed under the terms of the Creative Commons Attribution-NonCommercial licence (http://creativecommons.org/licenses/by-nc/4.0), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original article is properly cited. The written permission of Cambridge University Press must be obtained prior to any commercial use.
Copyright
© The Author(s), 2025. Published by Cambridge University Press on behalf of Società Italiana di Scienza Politica

Introduction

On 21 February 2020, at the break of the pandemic, Matteo Salvini, the leader of the League – the largest opposition party at the time – declared, ‘We must close everything’. However, just 6 days later, the same leader criticized the government's restrictive measures, stating, ‘The country is sinking […] it is necessary to reopen all activities and return to normalcy’.Footnote 1 Later, in March 2020, Salvini and his centre-right allies opposed government economic subsidies, and, in April 2020, they criticized Italy's participation in the Eurogroup's decision to use the European stability mechanism to assist European countries in dealing with the pandemic.Footnote 2 This prompted a strong reaction from Prime Minister Giuseppe Conte, who singled out Salvini and Giorgia Meloni – the leader of Brothers of Italy – arguing that the Italian government did not ‘operate under the favour of darkness’.Footnote 3

These quotations show that politicians faced uncertainty during the early stages of the COVID-19 pandemic, leading to changes in their communication and actions. They also demonstrate that the relationship between government and opposition during crises can be more complex than what the ‘rally around the flag’ argument suggests. This argument proposes that in times of crisis – typically, a war, a terrorist attack, a natural catastrophe or a global health emergency – public opinion and political actors unite against a common threat, setting aside their differences (Mueller, Reference Mueller1970). Consequently, a rally around the flag occurs, as government, majority and opposition will unite under the same flag to protect the lives of citizens best. However, when applying this framework to the COVID-19 pandemic, previous research offers only limited support for the rally around the flag hypothesis (Kritzinger et al., Reference Kritzinger, Foucault, Lachat, Partheymüller, Plescia and Brouard2021; Louwerse et al., Reference Louwerse, Sieberer, Tuttnauer and Andeweg2021; Christensen et al., Reference Christensen, Jensen, Kluth, Kristinsson, Lynggaard, Lægreid, Niemikari, Pierre, Raunio and Skúlason2023), emphasizing the need for a more nuanced understanding of its effects.

Drawing on existing research, this study examines whether the COVID-19 outbreak in Italy triggered a rally around the flag effect in the parliamentary arena. Deviating from previous studies that focused on public opinion and parliamentary speeches (Kritzinger et al., Reference Kritzinger, Foucault, Lachat, Partheymüller, Plescia and Brouard2021; Louwerse et al., Reference Louwerse, Sieberer, Tuttnauer and Andeweg2021), we analyse parties' amendment strategies to government decree-laws during the 18th legislature (2018–2022) to evaluate potential cooperation between the majority and opposition. Methodologically, we employ generalized additive models (GAMs) to dynamically examine the behaviour of majority and opposition parties, utilizing a novel database on parliamentary amendment activities.Footnote 4 Our findings suggest that while the pandemic influenced party behaviour, it did not foster a unified front between the majority and opposition. Despite the crisis, opposition parties persisted in obstructing the government's legislative initiatives through their amendments.

This paper is organized as follows. The second section reviews existing literature on the rally around the flag argument. The third section outlines the characteristics of the Italian case study and formalizes the research hypotheses. The fourth section presents the data used in the analysis and explains how the variables were operationalized. The fifth section reports and discusses the findings of the analysis of the amendatory activities of the Italian majority and opposition parties. The final section concludes the paper.

The ‘rally around the flag’ argument

The ‘rally around the flag’ effect is a phenomenon wherein citizens tend to increase their support for their government during times of crisis. According to this, the country's government and political leaders tend to experience a temporal boost in approval ratings at the beginning of a war or during an economic and health crisis (Costello, Reference Costello2021). The rally around the flag effect occurs when a country faces a ‘specific, dramatic, and sharply focused’ international event in which the country is directly implicated (Mueller, Reference Mueller1970). During such events, government officials must implement policies to safeguard the population. Consequently, citizens' anxieties and their imperative need for protection led to a transient increase in the government's popularity. This psychological effect typically ends when citizens feel less vulnerable. After the peak of the emergency, citizens begin to evaluate the consequences and continue to express support for the government only if they agree with its decisions. For example, in a war context, the presidential approval rating usually diminishes when citizens become aware of the human cost of the conflict (Oneal and Bryan, Reference Oneal and Bryan1995).

The rally around the flag literature revealed that other variables, such as media coverage and the number of deaths, mediate the effect on the governmental approval rating (Burk, Reference Burk1999). Remarkably, one strand of research, the opinion leadership school, later hypothesized that the rally around the flag effect occurs only when opposition leaders refrain from openly criticizing the government (Brody and Shapiro, Reference Brody and Shapiro1989).

Specifically, part of this literature has also demonstrated that the rally around the flag effect induces a shift in the behaviour of opposition leaders, as a cross-partisan consensus becomes more likely during the peak of a crisis. For example, according to Gallup, after 11 September, George W. Bush's approval rating rose from nearly 51% in early September 2001 to 90% in the weeks after 9/11.Footnote 5 Concomitantly, for almost 2 months, Democrat legislators supported nearly every decision taken by the Republican Administration in Congress (Chowanietz, Reference Chowanietz2011).

The rationale is that, during a crisis, opposition leaders tend to refrain from extensive interference with government activities, either due to a shared sense of national patriotism (patriotic reflex) (Mueller, Reference Mueller1970) or a lack of resources to criticize the government effectively (Brody, Reference Brody1991). According to the first hypothesis (patriotic reflex), oppositions would unite behind their ‘commander-in-chief’ to maximize their nation's prospects in the current crisis (Mueller, Reference Mueller1970). Conversely, according to the second, the opposition would support the country's government only out of necessity, lacking the necessary information to critique governmental action (Brody, Reference Brody1991). Regardless of the reason, akin to the government's initial increase in approval rating, the rally around the flag effect among the opposition is transient. After some time, the opposition resumes obstructionist activities, especially in cases of a repeated attack involving few causalities (Chowanietz, Reference Chowanietz2011).

Subsequently, when the world faced the consequences of COVID-19, scholars leveraged the rally around the flag hypothesis to frame the implications of the COVID-19 pandemic on government approval ratings. Following Mueller's (Reference Mueller1970) definition, COVID-19 constituted a rally around the flag event. It represented an unprecedented international crisis that was ‘specific, dramatic, and sharply focused’ directly involving the country and its government. Like a military crisis, COVID-19 impacted everyone's living conditions, exerting unparalleled pressure on governments to find a policy response. Secondly, like a war, the pandemic had high visibility in media coverage, catching many governments by surprise. Scholars have argued that the rally around the flag effect occurred primarily when governments responded decisively to citizens' imperative need for protection (Bol et al., Reference Bol, Giani, Blais and Loewen2021). Consequently, at the onset of the pandemic, most European Union (EU) governments experienced a brief rally around the flag effect (Mazza and Scipioni, Reference Mazza and Scipioni2022).

Scholars have explained the decrease in governmental approval ratings by examining how COVID-19 gradually evolved into a politicized issue. In particular, COVID-19 became politicized when it went from a health emergency to an economic one (Lynggaard et al., Reference Lynggaard, Exadaktylos, Jensen and Kluth2024). While at first (i.e. February–March 2020), the public broadly agreed on the measures taken to save human lives, later on (i.e. June–July 2020), it evaluated differently the economic, social and health repercussions of the decision to reopen (Jensen et al., Reference Jensen, Lyngg and Kluth2022). Concomitantly, political parties, media and organized interests advocated for differing preferences regarding lockdown measures. As a result, the previously unanimous consensus surrounding government policies disappeared, leading to a decline in government approval ratings (Kritzinger et al., Reference Kritzinger, Foucault, Lachat, Partheymüller, Plescia and Brouard2021).

The literature shows that party politics has generally followed the trend described above. For instance, Christensen et al. (Reference Christensen, Jensen, Kluth, Kristinsson, Lynggaard, Lægreid, Niemikari, Pierre, Raunio and Skúlason2023) showed that opposition parties rallied around the flag only at the outset of the pandemic in all five Scandinavian countries. Later, centre-right parties generally assumed a pro-business stance, calling for a faster reopening, while centre-left parties called for more restrictive measures. Furthermore, at the beginning of the pandemic, in certain countries, the sense of national unity was reflected even in the opposition's engagement in COVID-19 decision-making. To tackle a health crisis, Belgium's minority government joined forces with opposition parties to support new measures. Meanwhile, in the Netherlands, a member of the opposition became the new health minister following the resignation of their predecessor (Louwerse et al., Reference Louwerse, Sieberer, Tuttnauer and Andeweg2021).

Although arguments exist for the opposition cooperating with the government during crises, studies have shown that the rally around the flag effect is often temporary. Soon after the initial emergency phase, public support for the government may decline, and opposition parties may resume challenging the government in the legislative arena. Regarding public support, Kritzinger et al. (Reference Kritzinger, Foucault, Lachat, Partheymüller, Plescia and Brouard2021) illustrated that following the initial peak of the pandemic in the EU, majority and opposition voters exhibited varying levels of trust in the national government (see also Colloca et al., Reference Colloca, Roccato and Russo2024 on Italy). As to legislative activities, Louwerse et al.'s (Reference Louwerse, Sieberer, Tuttnauer and Andeweg2021) analysis of parliamentary speeches revealed that opposition MPs shifted towards openly contesting government measures as the situation progressed. Notably, the public's opposition to the executive's anti-pandemic policies mirrored the established positions of the opposition parties (Chiru, Reference Chiru2024).

Case-specific hypotheses

In the previous section, we outlined the ‘rally around the flag’ argument, with a particular focus on recent literature applying this framework to the COVID-19 crisis. In this section, we draw upon this literature to formulate research hypotheses for the Italian context. Italy exhibits several characteristics potentially indicative of a rally around the flag phenomenon. First, in the initial phase of the pandemic, there was a sense of national unity and fear of contagion in Italy, which, according to the rally around the flag hypothesis, facilitated the rise in Prime Minister Giuseppe Conte's approval ratings. Italy was the first Western country to be affected by COVID-19 and implement lockdown policies (Maggini and Pedrazzani, Reference Maggini and Pedrazzani2021), and, during the initial months of the pandemic (March–May 2020), Prime Minister Conte monopolized access to all news broadcasts (Bromo et al., Reference Bromo, Gambacciani and Improta2023). As a result, in line with the criteria identified in the literature, Prime Minister Conte witnessed an increase in popularity, which declined over time. Specifically, at the beginning of the pandemic (March 2020), Conte surged from 45 to 71% of the national consensus (February–March 2020). Then, in October 2020, his approval rating returned to 55%.

Secondly, during the 2020–2022 period, the majority and opposition parties shared government responsibilities between the national and regional levels, with opposition parties – FI, League,/ and FDI – leading the majority of regional governments (for a complete list of party acronyms, see Table A1 in the online appendix). As suggested by Louwerse et al. (Reference Louwerse, Sieberer, Tuttnauer and Andeweg2021), under such circumstances, opposition leaders cannot credibly fault the effectiveness of government actions because both actors share a similar responsibility to protect the lives of citizens in a multilevel setting (Figure 1).

Figure 1.

Government Conte II's approval ratings.

Source: Demos & PI, Atlante Politico 82–91.

A bar graph showing approval rating by month and year. See long description.
Figure 1 Long description

The bar graph shows approval rating left parenthesis percent right parenthesis by Month slash Year. The x axis label is Month slash Year. The y axis label is Approval rating left parenthesis percent right parenthesis. The y axis shows labeled ticks at 0, 20, 40 and 60. Bars and values by x axis category, left to right: Sep dash 19: 44. Oct dash 19: 40. Dec dash 19: 42. Feb dash 20: 44. Mar dash 20: 70. Apr dash 20: 63. Jun dash 20: 60. Aug dash 20: 57. Oct dash 20: 55. Dec dash 20: 57.

To explore the broader implications of crises on opposition legislative behaviour in Italy, we propose examining amendment activities to government decree-laws. An amendment is a change to a bill that is to be discussed and voted on in parliament, in the same way that bills are. In the Chamber of Deputies and the Senate, any parliamentarian can propose an amendment to a bill. Amendments can take various forms, from the suggestion of deleting or changing existing articles to adding new ones. The action of tabling many amendments in the plenary session is primarily made to delay government legislation, as according to the internal rule of procedure, every amendment must be previously discussed (Mattson, Reference Mattson and Döring1995). Specifically, obstructionism is particularly prominent during the examination of decree-laws, especially in the Chamber of Deputies, as its standing orders do not currently allow for time limitations on the discussion of decree-law conversion bills (Article 154 of the Rules of Procedure of the Chamber of Deputies; see Giannetti and Pedrazzani, Reference Giannetti and Pedrazzani2016). Furthermore, decree-laws are measures issued by the government in urgent and necessary situations. However, for them to have full effect, they must be converted into law by the parliament within 60 days from the publication of the decree's text in the Official Gazette. This time constraint gives oppositions a strategic opportunity to table many amendments, as their discussion slows down the conversion process, forcing the government to resort to the confidence procedure.Footnote 6

Considering the Italian context and drawing on the rally around the flag literature outlined above, we expect first oppositions to exhibit some degree of sensitivity to the crisis, leading them to adjust their amendment activities compared to non-crisis periods. This change could be attributed to the reduction in parliamentary activity during the pandemic (Bromo et al., Reference Bromo, Gambacciani and Improta2023) and constitutes a precondition for the rally around the flag effect. Second, during the initial stages of a crisis, the rally around the flag effect might lead to a temporary decrease in amendment activities as the opposition prioritizes cooperation or avoids actions that could be perceived as undermining national unity. Third, as the crisis progresses and its long-term consequences become more evident, opposition parties might gradually return to their usual level of amendment proposals, potentially even escalating it in response to government policies or perceived mishandling of the situation. The first hypothesis establishes a general expectation of change, whereas the remaining two explore the potential trajectories of this change across different phases of the crisis. In a more formal way, we expect that:

RH1: The pandemic crisis prompts an adjustment in opposition amendatory activities.

RH2: During the initial wave of the pandemic crisis, there is no significant difference between the amendatory activities of majority and opposition parties.

RH3: As the crisis progresses and its long-term consequences become more evident, the majority and opposition parties exhibit significant differences in amendatory activities.

Data, coding and methods

Data and coding

This study focuses on how governing and opposing parties behave during the COVID-19 crisis, using amendment tactics to decree-law conversion bills as a key indicator. Decree-laws were extensively used to manage the COVID-19 pandemic. These decrees addressed various aspects of the crisis, such as restricting citizens' rights and planning post-pandemic reopening. Building on this rationale, in this article, we examine the number of amendments presented by parliamentary parties and admitted to the plenary vote during the decree-law conversion process.Footnote 7

To maximize the potential for observing amendatory activities, we focus our analysis exclusively on the Chamber of Deputies. As mentioned above, the Chamber's rules do not impose time limitations on discussions surrounding decree-law conversion bills, creating a more conducive environment for amendatory practices to occur. To conduct our analysis, we leverage the Chamber of Deputies' linked-data. This allows us to extract information on all 125 decree-law conversion bills voted on during the 18th legislature (2018–2022). Of the 125 decree-law conversion bills analysed, 69 originated in the Chamber of Deputies and underwent their first reading there. The remaining bills originated in the Senate and were received for a second reading in the Chamber of Deputies. Generally, the Minister for relations with parliament decides the initial chamber to present the bills following contingent factors. For example, to make the bill approval faster in the first reading, they consider the amount of parliamentary work already scheduled for the discussion in each chamber. In the Italian context, governments can avoid voting amendments using the confidence procedure. According to the Italian laws and chambers' rules of procedure, if the government attaches a question of confidence to an article of a bill, and if the house votes in favour, the article is approved, and all amendments are considered rejected (Art. 116, Rule of Procedure of the Chamber; Art. 161, Rule of Procedure of the Senate).Footnote 8 Figure 2 depicts the monthly distribution of these bills, highlighting whether the government attached a confidence vote.

Figure 2.

Monthly distribution of decree-law conversion bills in the Italian Chamber of Deputies (2018–2022).

Source: Authors' elaboration on Chamber of Deputies linked data.

A stacked bar graph showing the monthly number of decree dash law conversion bills, split into with confidence and without confidence. See long description.
Figure 2 Long description

A stacked bar graph with the title Monthly distribution of decree dash law conversion bills in the Italian Chamber of Deputies left parenthesis 2018 dash 2022 right parenthesis. The x axis label is Year slash Month. The x axis shows 2018 dash 04, 2018 dash 05, 2018 dash 06, 2018 dash 07, 2018 dash 08, 2018 dash 09, 2018 dash 11, 2018 dash 12, 2019 dash 01, 2019 dash 02, 2019 dash 03, 2019 dash 04, 2019 dash 06, 2019 dash 07, 2019 dash 09, 2019 dash 10, 2019 dash 11, 2019 dash 12, 2020 dash 02, 2020 dash 03, 2020 dash 04, 2020 dash 05, 2020 dash 06, 2020 dash 07, 2020 dash 09, 2020 dash 10, 2020 dash 11, 2020 dash 12, 2021 dash 02, 2021 dash 03, 2021 dash 04, 2021 dash 05, 2021 dash 06, 2021 dash 07, 2021 dash 08, 2021 dash 09, 2021 dash 10, 2021 dash 11, 2021 dash 12, 2022 dash 01, 2022 dash 02, 2022 dash 03, 2022 dash 05, 2022 dash 06, 2022 dash 07, 2022 dash 09. The y axis label is N dot of Decree dash Law Conversion Bills. The y axis shows tick labels 0, 2, 4, 6. Legend: With Confidence; Without Confidence. Stacked bar values by Year slash Month, listed as Without Confidence plus With Confidence equals total: 2018 dash 04: 0 plus 1 equals 1. 2018 dash 05: 1 plus 1 equals 2. 2018 dash 06: 1 plus 1 equals 2. 2018 dash 07: 2 plus 1 equals 3. 2018 dash 08: 2 plus 0 equals 2. 2018 dash 09: 0 plus 2 equals 2. 2018 dash 11: 1 plus 0 equals 1. 2018 dash 12: 0 plus 1 equals 1. 2019 dash 01: 0 plus 2 equals 2. 2019 dash 02: 0 plus 1 equals 1. 2019 dash 03: 0 plus 1 equals 1. 2019 dash 04: 1 plus 2 equals 3. 2019 dash 06: 0 plus 2 equals 2. 2019 dash 07: 1 plus 1 equals 2. 2019 dash 09: 1 plus 1 equals 2. 2019 dash 10: 2 plus 2 equals 4. 2019 dash 11: 2 plus 1 equals 3. 2019 dash 12: 2 plus 2 equals 4. 2020 dash 02: 1 plus 2 equals 3. 2020 dash 03: 2 plus 2 equals 4. 2020 dash 04: 1 plus 2 equals 3. 2020 dash 05: 1 plus 1 equals 2. 2020 dash 06: 3 plus 1 equals 4. 2020 dash 07: 0 plus 2 equals 2. 2020 dash 09: 0 plus 1 equals 1. 2020 dash 10: 1 plus 1 equals 2. 2020 dash 11: 1 plus 1 equals 2. 2020 dash 12: 3 plus 2 equals 5. 2021 dash 02: 1 plus 1 equals 2. 2021 dash 03: 0 plus 0 equals 0. 2021 dash 04: 5 plus 1 equals 6. 2021 dash 05: 2 plus 2 equals 4. 2021 dash 06: 3 plus 2 equals 5. 2021 dash 07: 3 plus 1 equals 4. 2021 dash 08: 1 plus 2 equals 3. 2021 dash 09: 2 plus 2 equals 4. 2021 dash 10: 1 plus 2 equals 3. 2021 dash 11: 2 plus 2 equals 4. 2021 dash 12: 2 plus 1 equals 3. 2022 dash 01: 1 plus 1 equals 2. 2022 dash 02: 2 plus 2 equals 4. 2022 dash 03: 2 plus 2 equals 4. 2022 dash 05: 1 plus 2 equals 3. 2022 dash 06: 1 plus 2 equals 3. 2022 dash 07: 0 plus 1 equals 1. 2022 dash 09: 0 plus 0 equals 0.

According to Figure 2, out of the 125 conversion bills analysed, confidence votes were attached to 70. Interestingly, the figure does not show any time pattern in terms of the number of bills converting decree-laws or votes of confidence. Most importantly, we have not observed any significant difference since the COVID-19 pandemic outbreak (February 2020). This suggests that parliamentary votes and discussions on bills continued without major interruptions during the state of health emergency. Notably, the number of bills approved with the confidence procedure did not have any peak during the 18th legislative term (2018–2022). This suggests that the overall number of amendments tabled by parliamentarians has not changed significantly during the period of investigation, as governments resort to this procedure when a massive number of amendments would prevent the bills from being passed (i.e. within the timeframe of 60 days stipulated by Art. 77 of the Italian Constitution).

For each of the 125 decree-law conversion bills in our selection, we compiled data on the amendments admitted for a vote in the plenary session. More precisely, we counted the amendments proposed by each party in the Chamber of Deputies at the time of the vote. The party/conversion bill is our unit of analysis, while the number of amendments per party/conversion bill constitutes our dependent variable. Our analysis examines a dataset encompassing 850 observations. This number reflects the combined behaviour of all parties across every decree-law conversion bill voted on during the 18th legislature.Footnote 9

On average, each party submitted 2.38 amendments per bill. Notably, the minimum number of amendments proposed by a party was 0, while the maximum reached 102. Remarkably, no amendments were permitted for discussion in the Chamber of Deputies for any decree-law conversion bill linked to a confidence vote. This finding highlights the strategic use of confidence votes by the government, as it effectively discourages amendments and expedites the bill's passage. This observation is not necessarily a concern, as confidence votes do not exhibit a clear pattern over time. Table A1 in the online Appendix summarizes the key descriptive statistics for the dependent variable, offering preliminary insights into variations in amendment behaviour between governing and opposition parties.

We focus on two independent variables to test the previous section's hypotheses. The first is majority/opposition status, which allows us to examine how belonging to the governmental majority or the opposition affects parties' amendatory activities. Majority/opposition status is operationalized as a dummy variable, with 1 indicating opposition and 0 indicating majority at the time of the plenary vote. The second variable is a time trend, which tracks changes over time and identifies potential health crisis periods. We use a time trend variable to identify crisis periods that tracks the number of days since the legislature began its term (starting on 23 March 2018). This provides a more accurate monitoring of changes in activity during crisis periods compared to non-crisis periods, rather than using a dummy variable to differentiate between ‘regular’ and ‘crisis’ periods.

Building on existing research, we recognize that other factors beyond parties' majority/opposition status and crisis periods can influence variations in how actively parties propose amendments to decree-law conversion bills. We categorize these additional factors into three main groups: (1) characteristics of parties within parliament and their relationship with the government, (2) attributes of the government itself, and (3) features of the bill under consideration. Starting with the first category, previous research suggests that a wider ideological gap between a party and the government leads to less cooperation in voting or in expressed sentiment during parliamentary speeches (Tuttnauer, Reference Tuttnauer2018; Hohendorf et al., Reference Hohendorf, Saalfeld and Sieberer2021; Louwerse et al., Reference Louwerse, Sieberer, Tuttnauer and Andeweg2021). We believe this may equally apply to the proposed amendments to the legislation sponsored by the government. Even in contexts characterized by low levels of politicization, such as significant public health emergencies, we can anticipate that parties whose ideological stances diverge more significantly from the government will, on average, propose a greater number of amendments (Louwerse et al., Reference Louwerse, Sieberer, Tuttnauer and Andeweg2021). We also control for the size of parliamentary party group, as larger parties benefit from two key advantages: more resources to dedicate to amendments and a larger pool of members who can individually submit proposals.

Louwerse et al. (Reference Louwerse, Sieberer, Tuttnauer and Andeweg2021) highlight a further party characteristic that may contribute to explaining opposition behaviour: parties' prior experience in government. Parties with a history in government are more likely to be responsible and cooperate with the government on legislation and propose fewer amendments (Tuttnauer, Reference Tuttnauer2018). This ‘responsible’ behaviour may be even more pronounced during crises, as shown by De Giorgi and Moury (Reference De Giorgi and Moury2015: 118). As De Vries and Hobolt (Reference De Vries and Hobolt2020) argue, government experience can also differentiate between challenger and mainstream parties, which, according to country-specific case studies, are likely to exhibit distinct behaviours during crises like COVID-19 (Vande Walle et al., Reference Vande Walle, Wolfs and Van Hecke2021; Hájek, Reference Hájek2023). Challengers may distance themselves from the government, while mainstream parties often emphasize cooperation. Recent research suggests that similar dynamics can be observed between populist and non-populist parties, as populist parties often exploit crises to criticize governments for making decisions that they perceive as detrimental to the people (Lehmann and Zehnter, Reference Lehmann and Zehnter2022).

We use expert survey data from Giannetti et al. (Reference Giannetti, Pedrazzani and Pinto2022) collected during the 2018 general elections to measure the ideological distance between a party and the government. First, we calculate the government's average position on a left-right scale based on the scores of parties in the cabinet. Then, we define ideological distance as the absolute difference between a party's position and the government's average position.Footnote 10 Party size is computed as the seat share at the time a decree-law conversion bill is voted on. Next, we created a dummy variable for a party's prior government experience. This variable is coded as 1 if the party held a government position before 2018 and 0 otherwise. Among the parties in the 18th legislature, the M5S, FDI and IV had no prior government experience (coded 0). To classify parties as populist (1) or non-populist (0), we incorporate a further dummy variable. The data for this variable were obtained from the PopuList database (Rooduijn et al., Reference Rooduijn, Pirro, Halikiopoulou, Froio, Van Kessel, Lange, de Lange, Mudde and Taggart2024).Footnote 11

Moving to the attributes of the governments, previous research (Martin and Vanberg, Reference Martin and Vanberg2011; Pedrazzani and Zucchini, Reference Pedrazzani and Zucchini2013) highlights that in multiparty governments, delegating power to cabinet ministers can become problematic when the coalition itself is deeply divided on policy issues. Intra-coalition divergence incentivizes ministers to avoid agreed-upon compromises and take more extreme positions. This, in turn, triggers efforts by other coalition members to challenge and amend draft legislation to bring it back in line with the original compromise. For this reason, we expect a higher number of amendments as policy divisions in the government increase. To measure policy divisions in the government, we use the absolute distance between the two most extreme governing parties on the left-right scale. Party positions are derived from the same expert survey data mentioned above.

We include several control variables related to the decree-law conversion bills to account for additional factors potentially influencing amendments. First, we consider whether the bill originated in the Chamber of Deputies. We expect bills starting there to receive more amendments on average than those first examined by the Senate. This is because, according to recent practice in the Italian parliament, bills are mainly discussed and scrutinized in the legislative branch where they were introduced (Massa Pinto, Reference Massa Pinto2022). Second, we control for the number of committees assigned to examine the bill, as more committees generally translate into a higher likelihood of amendments being proposed. Finally, given our focus on legislative behaviour during a health crisis, we specifically isolate conversion bills of which the Ministry of Health is among the proposers. To incorporate the characteristics of the decree-law conversion bills, we employ several dummy variables. Bills that originated in the Chamber of Deputies were examined by more than one committee, and were proposed by the Ministry of Health are all coded as 1.Footnote 12

Methods

Our hypotheses state that during the initial stages of the COVID-19 health crisis, opposition parties will adjust their usual amendatory activities as they will exhibit a temporary decrease in amendments challenging the government. However, amendatory activities will return to pre-crisis levels as the crisis progresses. To empirically test these hypotheses, we need to interact our key independent variable, majority vs. opposition status, with a time function. This approach allows us to investigate whether amendment activity from the two groups changes significantly throughout the crisis (before, during and after). The challenge lies in selecting the appropriate time function. Traditional modelling methods typically employ parametric functions of time (linear, quadratic, square root, logarithmic) to represent the relationship between our independent variable (time), the dependent variable (number of amendments proposed by each party) and the majority/opposition status. While these functions are convenient, they impose a specific structure on the relationship, which might not accurately reflect the real data generation process.

To address the limitations of linear and generalized linear models in capturing complex relationships, we turn to GAMs (Hastie and Tibshirani, Reference Hastie and Tibshirani1990; Wood, Reference Wood2017; for applications in the field of political science, see Beck and Jackman, Reference Beck and Jackman1998). GAMs offer greater flexibility by accommodating non-linear patterns in the data. This flexibility is achieved through the use of smoothers. These are essentially flexible functions that adapt to the underlying trends in the data, regardless of whether they are linear or non-linear. This allows GAMs to identify potential turning points, periods of rapid change or cyclical patterns over time that traditional linear models might miss.

In a GAM, the relationship between the response variable and predictors is expressed as a combination of a parametric sub-model and the sum of smooth functions of one or more predictor variables. This flexibility allows GAMs to capture non-linear patterns while maintaining more structure than a completely unrestricted model. Unlike strictly parametric linear models or GLMs, GAMs balance flexibility and interpretability by incorporating both parametric and non-parametric components (Beck and Jackman, Reference Beck and Jackman1998; Wood, 2017). GAMs can accommodate dependent variables such as discrete counts and can incorporate various distributions to properly connect the combination of (parametric and smoothed) predictors to the expected values of the response variable.Footnote 13

Results

Table 1 presents four GAM analyses of the number of amendments in the Chamber of Deputies, with varying data subsets to check the robustness of our results. The first model (M1) includes all the observations (850). The second one (M2) excludes 280 observations of decree-law conversion bills with attached confidence, resulting in 570 observations. The third one (M3) focuses solely on 472 observations related to decree-law conversion bills assigned to the Chamber for the first reading. Finally, the fourth model (M4) combines the filters of models 2 and 3, analysing 347 observations of bills excluding those with confidence and including those assigned for the first reading in the Chamber. As mentioned earlier, each GAM contains two parts: parametric coefficients and smooth terms. In our analysis, the parametric component comprises all the covariates associated with parties, governments and bills introduced in the previous section. Additionally, we include the dummy variable that identifies the majority/opposition status. This dummy variable captures the mean of the dependent variable across the two levels within the two groups. Furthermore, we incorporate a smoother for the time variable (days since the start of the legislature). Specifically, we estimate a separate smoother for each level of the dummy variable that identifies majority/opposition status, which is equivalent to including a factor–smooth interaction term. Distinct smoothness parameters are estimated for each of these smoothers as well.Footnote 14 Because of additivity, the influence of each variable in a GAM can be analysed independently, as in linear regression.

Table 1.

GAMs of the number of amendments in the Italian Chamber of Deputies (2018–2022)

Table showing regression results for models M1 to M4, including parametric coefficients, smooth terms and model-fit statistics like adjusted R squared and deviance explained. See long description.
Table 1 Long description

The table contains regression results for models M1, M2, M3 and M4, with parametric coefficients and smooth term effective degrees of freedom. Columns are labeled M1, M2, M3 and M4. There are 15 rows. Row 1: Opposition coefficients are consistently positive across models, with values 4.399, 4.501, 3.638 and 3.809, indicating a strong effect. Row 2: Ideological distance shows small positive coefficients, 0.028, 0.027, 0.034 and 0.027. Row 3: Seat share has coefficients of 0.052, 0.052, 0.042 and 0.039. Row 4: Prior experience in government shows negative coefficients, -0.897, -0.900, -0.926 and -0.922. Row 5: Populist party coefficients are negative, -0.351, -0.356, -0.379 and -0.415. Row 6: Government policy divisions have positive coefficients, 0.231, 0.185, 0.178 and 0.161. Row 7: First reading Chamber is significant in M1 with 0.707, but not in other models. Row 8: More than one committee shows increasing coefficients, 0.490, 0.957, 0.893 and 1.240. Row 9: Ministry of Health coefficients are positive, 0.683, 0.574, 0.664 and 0.620. Row 10: Intercept values are negative, -6.594, -5.809, -5.550 and -5.389. Smooth term effective degrees of freedom: Row 11: s(days): opposition0 values are 3.446, 3.975, 3.677 and 4.157. Row 12: s(days): opposition1 values are 7.268, 6.443, 6.505 and 5.877. Row 13: N values are 850, 570, 472 and 347. Row 14: Adjusted R squared values increase from 0.398 to 0.520 across models. Row 15: Deviance explained increases from 0.534 to 0.648. The table highlights consistent positive effects of opposition and increasing model fit from M1 to M4. Coefficients are likely log-odds and 'days' in smooth terms may refer to time-related variables. The table provides a comprehensive view of model performance and variable significance.

Note: Standard errors in parentheses for parametric coefficients. Effective degrees of freedom are reported for smooth terms.

Significance codes: ***P < 0.001; **P < 0.01; *P < 0.05; +P < 0.10.

Table 1 reveals a significant difference in the expected amendment frequency between opposition and government-supporting parties. The positive and statistically significant coefficient associated with opposition status indicates that, on average, opposition parties present more amendments than governing parties. Furthermore, the statistically significant effective degrees of freedom values in the second part of the table suggest that the relationship between majority/opposition status and the response variable is not constant over time. This implies the presence of non-linear interactions between the smoother and our dummy variable, as the effect of opposition status on amendment frequency varies throughout the legislative session. For a deeper understanding of these non-linear interactions, Figure 3 visually represents the marginal effect of opposition status across different time points, along with 95% confidence intervals (CIs). The figure allows us to trace potential variations in how opposition status influences amendment frequency throughout the legislative session for each model estimated in Table 1. Figure 3 additionally highlights the three distinct waves characterizing the evolution of the COVID-19 health crisis in Italy over time.

Figure 3.

Marginal impact of opposition status on the number of amendments.

Note: Marginal effects are computed while other covariates in the model are held at their observed values.

Different types of data visualizations such as four line graphs showing slope over days for M1, M2, M3 and M4. See long description.
Figure 3 Long description

The image A showing M1. Text: M1. The horizontal axis label: days. The horizontal axis range: 0 to 1500. The vertical axis label: Slope. The vertical axis range: 0.0 to 10.0. Vertical axis tick labels: 0.0, 2.5, 5.0, 7.5, 10.0. Text inside plot: 1st wave, 2nd wave, 3rd wave. A single curved line starts near 5.0 at day 0, drops to near 2.5 around day 450, rises to near 5.5 around day 1000, drops to near 4.0 around day 1350, then rises to near 5.5 by day 1500. Three vertical shaded bands appear around days near 700, near 1000 and near 1100. A row of small tick marks appears along the bottom of the plot. The image B showing M2. Text: M2. The horizontal axis label: days. The horizontal axis range: 0 to 1500. The vertical axis label: Slope. The vertical axis range: 0.0 to 10.0. Vertical axis tick labels: 0.0, 2.5, 5.0, 7.5, 10.0. Text inside plot: 1st wave, 2nd wave, 3rd wave. A single curved line starts near 5.5 at day 0, drops to near 2.5 around day 450, rises to near 5.5 around day 1000, drops to near 4.0 around day 1350, then rises to near 4.8 by day 1500. Three vertical shaded bands appear around days near 750, near 1050 and near 1150. A row of small tick marks appears along the bottom of the plot. The image C showing M3. Text: M3. The horizontal axis label: days. The horizontal axis range: 0 to 1500. The vertical axis label: Slope. The vertical axis range: 0.0 to 10.0. Vertical axis tick labels: 0.0, 2.5, 5.0, 7.5, 10.0. Text inside plot: 1st wave, 2nd wave, 3rd wave. A single curved line starts near 4.5 at day 0, drops to near 2.2 around day 450, rises to near 5.0 around day 1050, drops to near 2.7 around day 1300, then rises to near 3.0 by day 1500. Three vertical shaded bands appear around days near 750, near 1050 and near 1150. A row of small tick marks appears along the bottom of the plot. The image D showing M4. Text: M4. The horizontal axis label: days. The horizontal axis range: 0 to 1500. The vertical axis label: Slope. The vertical axis range: 0.0 to 10.0. Vertical axis tick labels: 0.0, 2.5, 5.0, 7.5, 10.0. Text inside plot: 1st wave, 2nd wave, 3rd wave. A single curved line starts near 5.2 at day 0, drops to near 2.2 around day 450, rises to near 5.2 around day 1000, then declines to near 2.7 by day 1500. Three vertical shaded bands appear around days near 750, near 1050 and near 1150. A row of small tick marks appears along the bottom of the plot.

Figure 3 reveals several key insights. First, the marginal effect of being an opposition party significantly increases the number of amendments for most of the legislative session. This is evident as the CIs never cross the zero-line (indicating no difference between majority and opposition). Second, the magnitude of the opposition effect is not constant over time. It is initially higher, then decreases to its lowest point near the end of the Conte I government (20 August 2019). Interestingly, a turning point emerges, with the opposition impact increasing again throughout the first wave of the pandemic crisis (March–April 2020) until reaching a peak during the second wave (October–December 2020). This peak is followed by a decrease during the third wave (March–April 2021) and the eventual plateauing of the effect. This pattern holds true across all models, indicating robust results regardless of selection criteria.

Figure 3 reveals an unexpected pattern in amendment activity. While the figure confirms potential adaptation by opposition parties during the crisis (H1), the observed trend diverges significantly from what the rally around the flag hypothesis would predict. Instead of a decrease in opposition amendments during the initial phases, as expected by H2, the figure shows an acceleration throughout both the first and second waves of the crisis. Notably, the decrease in amendatory activity for both opposition and majority parties only occurs after the second wave, 2 months after the start of the Draghi government (February 2021). This finding further contradicts H3, which anticipated a return to the typical differences between majority and opposition as the crisis progressed. In other words, our analysis shows that the rise in amendment submissions that began after the start of the Conte II government continued even during the early stages of the pandemic. Interestingly, our analysis suggests that the pandemic has even amplified this trend, at least until the second wave. Only once the pandemic situation improved did the number of submissions start to decline.

Although Figure 3 focuses on the differences in the slopes of the interaction between the time smoother and majority/opposition status, Figure 4 translates these variations into the metric of the response variable, allowing us to assess the substantive effect of our key independent variables. More precisely, the figure compares predicted amendment counts for a party between the two levels of our dummy (opposition vs. majority) across time. The graphs' initial bell shape tells us that most opposition activity is concentrated in the first 500 days of the legislative term, which coincide with the Conte I government. The remaining part of the term shows a significant decrease in amendments submitted by the opposition. Within this context of lower predicted amendment frequency, it is noteworthy that the only period showing a slight increase in opposition activity is the second wave of the pandemic crisis. However, in absolute terms, this translates to just a few more amendments compared to majority parties. This further reinforces the notion that the crisis impacted opposition amendatory activities, albeit in a direction contrary to the rally around the flag argument.

Figure 4.

Predicted differences in amendment counts between majority and opposition parties.

Note: Predicted differences are computed while other covariates in the model are held at their observed values.

Four line graphs labeled M1, M2, M3 and M4 showing comparison values over days, each with shaded variability bands and three marked wave periods. See long description.
Figure 4 Long description

The image contains four graphs arranged in a two by two grid, labeled M1 (top left), M2 (top right), M3 (bottom left) and M4 (bottom right). Each graph shares the same axis structure. The horizontal axis is labeled days and ranges from 0 to 1500. The vertical axis is labeled Comparison and ranges from 0 to 100. Each graph contains one solid trend line, a surrounding shaded variability band, a horizontal dashed reference line at 0, three vertical shaded bands labeled 1st wave, 2nd wave and 3rd wave positioned at approximately days 800, 1000 and 1100 respectively and rug marks along the horizontal axis indicating data density. The 1st wave marker appears at approximately day 800, the 2nd wave marker at approximately day 1000 and the 3rd wave marker at approximately day 1100. All four graphs share a common pattern: the solid line rises from a low value near day 0, reaches a peak between days 200 and 300, then declines steadily toward 0 by approximately day 700, remaining near 0 through the wave periods with minor variation. Key differences per panel are as follows. M1: The peak reaches approximately 50 at around day 250. The shaded variability band is wide at the peak and narrows toward day 1500. M2: The peak reaches approximately 30 at around day 250, lower than M1. The variability band is narrower than M1 at the peak. M3: The peak reaches approximately 40 at around day 250. The variability band is the widest among all four panels, extending from approximately 0 to 75 at the peak. M4: The peak reaches approximately 25 at around day 250, the lowest peak among the four panels. The variability band is moderate in width. In all panels, the solid line remains near 0 during the 1st wave, 2nd wave and 3rd wave periods, with only minor upward variation visible during the 2nd wave period. The dashed horizontal reference line at 0 is present across all four panels.

Moving to the control variables, Table 1 shows that most coefficients are statistically significant in the direction predicted by the existing literature. Larger parties are expected to present more amendments compared to smaller ones. On the contrary, parties with previous government experience are expected to show a lower number of amendments in comparison to those who have never shared government responsibility. Turning to government characteristics, ideological divisions between cabinet parties result in a higher count of the response variable. Considering the properties of the decree-law conversion bills, bills first introduced in the Chamber of Deputies, assigned to more than one committee for their examination and proposed by the Ministry of Health among others are all associated with a higher number of amendments. The only covariates that are not significant are those related to the ideological difference between the party and the government and the populist dummy. The dummy identifying majority opposition status probably cancels out the effect of these variables.Footnote 15

Discussion and conclusion

This article tested whether opposition parties changed their legislative behaviour during the pandemic. Given the severity of the pandemic in Italy, we expected the government, the opposition and the majority to unite under the same flag. However, the evidence provided by our models revealed that the COVID-19 crisis impacted legislative behaviour (RH1) without provoking a rally around the flag effect. Opposition parties did not refrain from obstructing the government's legislative initiatives through amendments and, more remarkably, contrary to what the theory suggests, in the first peak of the pandemic (March 2020), they increased their amendment activities (RH2). Then, during the third wave (RH3), they changed their behaviour, tabling less amendments. These results partially align with previous research indicating an end of the rally around the flag effect in parliamentary debates when public opinion becomes more aware of the economic implications of the crisis (i.e. during the second wave). Despite these findings, the decline in obstructionist activity during the third wave remains puzzling.

We believe these results should be interpreted in light of the unique characteristics of the case study under investigation. Specifically, two contingent factors may be relevant: pandemic management and party polarization. How the government handled the pandemic can explain the behaviours in the first two waves, as Italy was the first country hit by COVID-19, and, from a comparative perspective, it was atypical in the use of secondary legislation. As a result, when the crisis mainly hit, the parliament had few occasions to change what the government already established. In these circumstances – that spanned during the first two waves – the government would have no incentives to curb parliamentary rights further. The use of the confidence procedure or the reduction of time devoted to the discussion of amendments would have exposed the government to public critiques, as it would have meant a further reduction of the scrutiny function of the parliament. Consequently, the government strategically allowed more amendments to be voted on in the first two waves, perhaps in an attempt to pre-empt criticism from opposition parties. At the same time, opposition parties exploited this opportunity, tabling more amendments than the majority, just as they do in ordinary times.

The peak of obstructionist activity during the second wave of the pandemic can be attributed to party polarization. Opposition parties fiercely criticized the government's lockdown and reopening decisions, leading to increased conflict and a rise in amendments. However, the establishment of the Draghi government in February 2021 marked a turning point. The new government's broader base of support, including some centre-right parties, may have reduced the differences between majority and opposition amendments, leading to a decrease in obstructionist activity during the third wave.

Our main findings add new perspectives on the rally around the flag hypothesis on opposition behaviour. Firstly, from a methodological standpoint, our article suggests that establishing a criterion to measure the agreement between majority and opposition could be critical. On the one hand, opposition parties can express a national unit sentiment in their speech (Louwerse et al., Reference Louwerse, Sieberer, Tuttnauer and Andeweg2021), but, at the same time, table many amendments to undermine government lawmaking. In other words, as highlighted by previous research (Laflamme et al., Reference Laflamme, Milot-Poulin, Desrosiers, Verreault, Fillion, Patenaude and Bodet2023), the opposition's attitude towards the government may not be consistent, as they can appear to support the government in speeches while also engaging in obstructionist tactics. The following research could tackle this distinction further.

Secondly, existing literature on the rally around the flag effect primarily focuses on public opinion, observing rally effects after major crises more prominently in the public sphere than within political institutions. Political institutions like the Italian parliament appear to follow their internal logic and are somewhat insulated from shifts in public opinion. Consequently, the behaviours of politicians in legislative and governmental arenas are less susceptible to volatility than changes in voter attitudes. Finally, we did not analyse committee-level amendments due to lack of data. Future research could examine these to see if majority and opposition parties behave differently in committees, which remain the primary venue for legislative revision despite their generally consensual nature.

Funding

This work was supported by the Italian Ministry of Education, University and Research (grant 2020NK2YHL_002, ‘DEMOPE: DEMOcracy Under Pressure’).

Data

The replication dataset is available at http://thedata.harvard.edu/dvn/dv/ipsr-risp.

Supplementary material

The supplementary material for this article can be found at https://doi.org/10.1017/ipo.2024.29.

Competing Interests

The authors declare none.

Footnotes

1 https://www.repubblica.it/politica/2020/04/16/news/salvini_e_i_cambi_di_rotta_sulle_aperture_attivita_-254162048/ (accessed 8-5-2024).

2 https://www.ilgiornale.it/news/politica/coronavirus-forza-italia-boccia-conte-delusi-governo-non-1835016.html (accessed 8-10-2024).

3 https://video.sky.it/news/politica/video/coronavirus-conte-a-salvini-e-meloni-no-governo-di-tenebre-586199 (accessed 8-5-2024).

4 COVID-19 was handled first by the second Conte government (2019–2021), then by the government of Mario Draghi (2021–2022). The first government was supported by an alliance between the Five Star Movement and the centre-left, while the second was a government of national unity, supported by all Italian parties except Brother of Italy and other small parties.

5 https://rollcall.com/2021/08/31/rip-rally-round-the-flag-afghanistan-trump-biden-republican-democrat-9-11-bush/ (accessed 10-12-2024).

6 Amendments, like any other legislative activity, can be pursued by representatives for electoral, office-seeking or policy reasons (Fenno, Reference Fenno1973). In this article, we are not interested in investigating the specific reasons for tabling amendments, as our primary objective is to verify whether opposition representatives, compared to their majority counterparts, interfered with government legislation during the pandemic.

7 Specifically, the government during the pandemic utilized both primary and secondary legislation to impose COVID-19 restrictions. However, representatives could effectively modify government secondary COVID-19 legislation by introducing new principles through amendments to government decree-law conversion bills (e.g. Lippolis, Reference Lippolis2021: 270).

8 The literature has demonstrated that, particularly since the 1990s, Italian executives have employed the confidence procedure, along with other instruments such as decree-laws, delegating laws, and maxi-amendments, to bolster their prerogatives in the legislative arena and ensure the enactment of their policy priorities (Capano and Giuliani, Reference Capano and Giuliani2001; Zucchini, Reference Zucchini2013; De Micheli, Reference De Micheli2014).

9 The total number of observations is 125 multiplied by 6, which equals the number of parliamentary parties existing throughout the entire legislature. Additionally, there are 100 observations related to Italy Alive (IV), a splinter of the Democratic Party. These observations are only included starting from the formation of IV in September 2019 (see Table A1 in the online Appendix).

10 We additionally calculated the difference between party ideology and the government's median position. The results using the mean and median are practically identical.

11 Given that FI is categorized as a borderline case in the PopuList database, we classify it as a non-populist party.

12 While a variable indicating a vote of confidence being linked to the bill is relevant, it is excluded from our models. This is because all bills associated with a vote of confidence consistently receive no amendments, making this variable a perfect predictor.

13 We rely on the quasi-Poisson distribution which extends the standard Poisson model by relaxing the assumption that the variance equals the mean. Quasi-Poisson distribution works well with count data that exhibits overdispersion and a significant number of observations with zero counts.

14 There are two main considerations when building a GAM (Wood, 2017). Choosing the basis for the smooths and determining the dimension of the smoothing basis. The first point is related to the underlying mathematical structure used to capture the non-linear relationships in the data. The second one defines the maximum level of complexity for the smooths. For basis selection, we opt for a thin plate regression spline. As for dimensionality, we set the dimension of the smoothing basis to 10. This choice aligns with the default setting in the R package mgcv (Wood, 2017), which is the tool fitting GAMs used in this analysis. This establishes the upper limit on the flexibility of the smooths. The final smoothness of each curve within the basis (the effective degrees of freedom) is determined automatically by the model fitting algorithm. This is achieved by introducing a wiggliness penalty. The penalty discourages overly complex curves, helping to find the right balance between capturing the trends and avoiding overfitting the data.

15 Alternative models excluding the majority/opposition dummy reveal a correlation between the number of amendments proposed and both the party-government ideological difference and the populist dummy. This indicates that the majority/opposition status neutralizes the effects of these variables (see the online Appendix).

References

Beck, N and Jackman, S (1998) Beyond linearity by default: generalized additive models. American Journal of Political Science 42, 596–627.CrossRefGoogle Scholar
Bol, D, Giani, M, Blais, A and Loewen, PJ (2021) The effect of COVID-19 lockdowns on political support: some good news for democracy? European Journal of Political Research 60, 497–505.CrossRefGoogle Scholar
Brody, RA (1991) Assessing the President: The Media, Elite Opinion, and Public Support. Stanford: Stanford University Press.CrossRefGoogle Scholar
Brody, RA and Shapiro, CR (1989) Policy failure and public support: the Iran-Contra affair and public assessment of President Reagan. Political Behavior 11, 353–369.CrossRefGoogle Scholar
Bromo, F, Gambacciani, P and Improta, M (2023) Executive power and accountability in Italy and the government's response to COVID-19. Interdisciplinary Political Studies 9, 43–64.Google Scholar
Burk, J (1999) Public support for peacekeeping in Lebanon and Somalia: assessing the casualties hypothesis. Political Science Quarterly 114, 53–78.CrossRefGoogle Scholar
Capano, G and Giuliani, M (2001) Governing without surviving? An Italian paradox: law-making in Italy, 1987–2001. Journal of Legislative Studies 7, 13–36.CrossRefGoogle Scholar
Chiru, M (2024) The resilience of parliamentary oversight during the COVID-19 pandemic. West European Politics 47, 408–425.CrossRefGoogle Scholar
Chowanietz, C (2011) Rallying around the flag or railing against the government? Political parties’ reactions to terrorist acts. Party Politics 17, 673–698.CrossRefGoogle Scholar
Christensen, T, Jensen, MD, Kluth, M, Kristinsson, GH, Lynggaard, K, Lægreid, P, Niemikari, R, Pierre, J, Raunio, T and Skúlason, GA (2023) The Nordic governments’ responses to the COVID-19 pandemic: a comparative study of variation in governance arrangements and regulatory instruments. Regulation & Governance 17, 658–676. https://doi.org/10.1111/rego.12497.CrossRefGoogle Scholar
Colloca, P, Roccato, M and Russo, S (2024) Rally ‘round the flag effects are not for all: trajectories of institutional trust among populist and non-populist voters. Social Science Research 119, 102986.CrossRefGoogle Scholar
Costello, R (2021) Rally around the EU flag: Irish party positions on the EU in the wake of Brexit. Journal of Contemporary European Studies 29, 502–518.CrossRefGoogle Scholar
De Giorgi, E and Moury, C (2015) Conclusions: great recession, great cooperation? The Journal of Legislative Studies 21, 115–120.CrossRefGoogle Scholar
De Micheli, C (2014) Parlamento e governo in Italia. Partiti, procedure e capacità decisionale (1948–2013). Milano: Angeli.Google Scholar
De Vries, CE and Hobolt, SB (2020) Political Entrepreneurs: The Rise of Challenger Parties in Europe. Princeton: Princeton University Press.Google Scholar
Fenno, R (1973) Congressmen in Committees. Boston: Little Brown.Google Scholar
Giannetti, D and Pedrazzani, A (2016) Rules and speeches: how parliamentary rules affect legislators' speech-making behavior. Legislative Studies Quarterly 41, 771–800.CrossRefGoogle Scholar
Giannetti, D, Pedrazzani, A and Pinto, L (2022) Faraway, so close: a spatial account of the Conte I government formation in Italy, 2018. Italian Political Science Review/Rivista Italiana di Scienza Politica 52, 83–100.CrossRefGoogle Scholar
Hájek, L (2023) Legislative behaviour of MPs in the Czech Republic in times of COVID-19 pandemic. Parliamentary Affairs 76, 401–420.CrossRefGoogle Scholar
Hastie, TJ and Tibshirani, RJ (1990) Generalized Additive Models. New York: Chapman & Hall.Google Scholar
Hohendorf, L, Saalfeld, T and Sieberer, U (2021) Veto power fosters cooperative behaviour: institutional incentives and government-opposition voting in the German Bundestag. West European Politics 44, 921–945.CrossRefGoogle Scholar
Jensen, M, Lyngg, K and Kluth, M (2022) Paths, punctuations and policy learning – comparing patterns of European use of scientific expertise during the COVID-19 crisis. Public Organization Review 22, 223–247.CrossRefGoogle Scholar
Kritzinger, S, Foucault, M, Lachat, R, Partheymüller, J, Plescia, C and Brouard, S (2021) ‘Rally round the flag’: the COVID-19 crisis and trust in the national government. West European Politics 44, 1205–1231.CrossRefGoogle Scholar
Laflamme, L, Milot-Poulin, J, Desrosiers, J, Verreault, C, Fillion, C, Patenaude, N and Bodet, MA (2023) Opposition parties in times of pandemics. The Journal of Legislative Studies, 1–26. https://doi.org/10.1080/13572334.2023.2171542.Google Scholar
Lehmann, P and Zehnter, L (2022) The self-proclaimed defender of freedom: the AfD and the pandemic. Government and Opposition 59(4), 1109–27. doi: 10.1017/gov.2022.5.CrossRefGoogle Scholar
Lippolis, V (2021) Il Rapporto Parlamento-governo nel Tempo della Pandemia. Rivista AIC 1/2021, 268–277.Google Scholar
Louwerse, T, Sieberer, U, Tuttnauer, O and Andeweg, RB (2021) Opposition in times of crisis: COVID-19 in parliamentary debates. West European Politics 44, 1025–1051.CrossRefGoogle Scholar
Lynggaard, K, Exadaktylos, T, Jensen, MD and Kluth, M (2024) Mapping the changing role of expertise in COVID-19 politics in Europe. Policy & Politics 52, 44–66.CrossRefGoogle Scholar
Maggini, N and Pedrazzani, A (eds) (2021) Come siamo cambiati? Opinioni, orientamenti politici, preferenze di voto alla prova della pandemia. Milano: Fondazione Giangiacomo Feltrinelli.Google Scholar
Martin, LW and Vanberg, G (2011) Parliaments and Coalitions: The Role of Legislative Institutions in Multiparty Governance. Oxford: Oxford University Press.CrossRefGoogle Scholar
Massa Pinto, I (2022) Il ‘monocameralismo di fatto’ e la questione della perdurante validità della Costituzione. Costituzionalismo.it 3, 87–112.Google Scholar
Mattson, I (1995) Private members’ initiatives and amendments. In Döring, H (ed.), Parliaments and Majority Rule in Western Europe. New York: St Martin's Press, pp. 448–487.Google Scholar
Mazza, J and Scipioni, M (2022) The brief rally around the flag effect of COVID-19 in Europe. Publications Office of the European Union Luxembourg, 2022, doi:10.2760/657577, JRC126494.CrossRefGoogle Scholar
Mueller, JE (1970) Presidential popularity from Truman to Johnson. American Political Science Review 64, 18–34.CrossRefGoogle Scholar
Oneal, JR and Bryan, AL (1995) The rally ‘round the flag effect in US foreign policy crises, 1950–1985. Political Behavior 17, 379–401.CrossRefGoogle Scholar
Pedrazzani, A and Zucchini, F (2013) Horses and hippos: why Italian government bills change in the legislative arena, 1987–2006. European Journal of Political Research 52, 687–714.CrossRefGoogle Scholar
Rooduijn, M, Pirro, ALP, Halikiopoulou, D, Froio, C, Van Kessel, S, Lange, SL, de Lange, S, Mudde, C and Taggart, P (2024) The PopuList: a database of populist, far-left, and far-right parties using expert-informed qualitative comparative classification (EiQCC). British Journal of Political Science 54, 969–978.CrossRefGoogle Scholar
Tuttnauer, O (2018) If you can beat them, confront them: party-level analysis of opposition behavior in European national parliaments. European Union Politics 19, 278–298.CrossRefGoogle Scholar
Vande Walle, B, Wolfs, W and Van Hecke, S (2021) Opposition in times of COVID-19 – to support or not to support? Politics of the Low Countries 3, 138–159.CrossRefGoogle Scholar
Wood, SN (2017) Generalized Additive Models: An Introduction with R (2nd ed.). Chapman and Hall/CRC. https://doi.org/10.1201/9781315370279.CrossRefGoogle Scholar
Zucchini, F (2013) La repubblica dei veti. Un'analisi spaziale del mutamento legislativo in Italia. Milano: Egea.Google Scholar
Figure 0

Figure 1. Government Conte II's approval ratings.Source: Demos & PI, Atlante Politico 82–91.Figure 1 long description.

Figure 1

Figure 2. Monthly distribution of decree-law conversion bills in the Italian Chamber of Deputies (2018–2022).Source: Authors' elaboration on Chamber of Deputies linked data.Figure 2 long description.

Figure 2

Table 1. GAMs of the number of amendments in the Italian Chamber of Deputies (2018–2022)Table 1 long description.

Figure 3

Figure 3. Marginal impact of opposition status on the number of amendments.Note: Marginal effects are computed while other covariates in the model are held at their observed values.Figure 3 long description.

Figure 4

Figure 4. Predicted differences in amendment counts between majority and opposition parties.Note: Predicted differences are computed while other covariates in the model are held at their observed values.Figure 4 long description.

Supplementary material: File

Gambacciani et al. supplementary material

Gambacciani et al. supplementary material
Download Gambacciani et al. supplementary material(File)
File 229.5 KB