Quality of elections: definitions and measures

Stefano Rombi1 https://orcid.org/0000-0003-2652-6097

Fulvio Venturino1 https://orcid.org/0000-0002-2850-6017

1
University of Cagliari

Corresponding author: Stefano Rombi, [email protected]

This article has been accepted for publication and undergone full peer review
but has not been through the copyediting, typesetting, pagination and proofreading process,
which may lead to differences between this version and the Version of Record.

Accepted: September 12, 2023 Published: September 20, 2023

Abstract:
The study of the quality of elections is spreading very rapidly. This article aims to provide a general
overview of the state of the art to introduce an empirical analysis of the quality of elections for the
Italian parliament. After reviewing the definitions of ‘quality of elections’, the article thoroughly
examines the two main measurement tools. One is based on the experts’ perceptions and promoted as
part of the Electoral Integrity Project. The other is based on objective data called the Elections
Performance Index. Finally, the article applies the measurement method based on the residual vote to
the Italian case.

Keywords:
Quality of Elections; Italian Politics; Electoral Integrity; Electoral Systems.

Please cite this article as:

Rombi, S., & Venturino, F. (2023). Quality of elections: definitions and measures. Italian Journal of
Electoral Studies QOE – IJES, Just Accepted. https://doi.org/10.36253/qoe- 14081

1
 Introduction

Most countries in the world provide for some form of election of their political rulers.
According to a recent count, 185 independent states out of 193 currently elect members of the lower
houses of parliament (Norris et al. 2016). Since the end of World War II, Brunei, China, Eritrea,
Qatar, and Saudi Arabia have been among the few countries where elections have not been held
(Golder 2005). This means that regardless of the democratic or autocratic nature of the regime,
elections currently are the main instrument of mobilisation or participation of the political
community. This simple observation helps us understand why the electoral process is so essential for
the social sciences and, above all, for political science.

Elections are widespread in both democratic and autocratic political systems; however, it is
apparent that they perform very different functions. In democratic regimes, elections are helpful both
in legitimising the government and allowing citizens to choose their representatives. Instead, in
autocratic regimes, elections have mainly the purpose of co-opting wider social groups, collecting
information on the oppositions, and allowing the expression of popular discontent (Gandhi and Lust-
Okar 2009). This state of affairs has involved the rise of hybrid regimes, such as the ‘electoral
authoritarianism’ proposed by Schedler (2006). Remarkably, using the adjective ‘electoral’ would be
utterly redundant in speaking of democracy: authoritarian regimes do not need elections; democracies
cannot exist without them.

For the reasons just mentioned, although it also covers authoritarian regimes, the study of the
quality of elections plays a critical role in democratic regimes. More precisely, it is closely linked
with an essential dimension of the quality of democracy: electoral accountability (Rombi 2015). As
studies on the quality of democracy have established (f.i., Diamond and Morlino 2004), quality in a
procedural sense is measured by looking at the functioning of electoral accountability mechanisms.
Moreover, the proper functioning of elections – their level of integrity – is a necessary, though not
sufficient, precondition for electoral accountability. When the elections are manipulated, the voter
register excludes parts of the population, the electoral campaign is not correctly regulated, and the
votes are not accurately tallied, then representatives have no incentive to act in the interests of the
voters (Pitkin 1967). When these – and other similarly negative – conditions occur, elections have
low integrity, political representation breaks, and accountability does not work suitably.

2
 Although the quality of elections is paramount in authoritarian and hybrid political systems,
it also concerns democratic representation. Within established democracies, interest in the quality of
elections emerged following the 2000 American presidential election due to the global political
scandal related to the malfunctioning of Florida’s punch card ballot system (Alvarez et al. 2008;
Alvarez et al. 2012; Bowler et al. 2015). Since then, many studies have shown that even established
democracies present widespread problems related to gerrymandering, vote fraud, voter registration,
vote buying, risks posed by voting technology, early voting procedures, challenges posed by social
media, and foreign interference in elections.1

This article aims to discuss some basic tenets of the literature on the quality of elections and
to adapt them to the analysis of the Italian political system. To do so, the present article proceeds
through two steps. First, we propose a review of the pertinent literature. Although relatively new, it
has already reached a notable amount that is impossible to manage in a single article. Hence, we
concentrate on two prominent approaches: the Electoral Integrity Project (EIP) proposed by Pippa
Norris and associates (Norris, Frank and Martinez i Coma 2013) and the Elections Performance Index
(EPI) developed by the Caltech/MIT research group (Gerken 2009; Stewart 2020). Second, we use
some measures provided by this literature and tentatively apply them to an exploration of Italian
politics. Of course, we do not deliver a systematic analysis; instead, we aim to show the potentialities
granted by applying consolidated theories to studying a polity such as Italy.

The article is organised as follows. The subsequent section provides a literature overview on
the quality of elections to discover recurrent definitions, methodological approaches and
measurement problems. Then a cursory empirical analysis of the Italian case is presented based on
the two examined theories. The final section contains some concluding remarks.

Defining the quality of elections

The concept of quality of elections has taken on different names. The literature has proposed
several definitions without finding a univocal proposal. In a somewhat confusing landscape, the most

1
Empirical analysis of the quality of elections in consolidated democracies regards countries such as the United States,
Great Britain, the Netherlands, Switzerland, and Ireland (Alvarez et al. 2008; Alvarez et al. 2012; Bowler et al. 2015;
James 2013; Clark 2017; Buckley and Reidy 2015; Norris et al. 2018).
3
effective method to put some order was presented by Carolien van Ham (2015; 2020), who deals with
electoral integrity, distinguishing between positive and negative definitions. Expressions such as ‘free
and fair elections’, ‘democratic elections’, and ‘elections quality’ all belong to the universe of positive
definitions (Elklit and Reynolds 2005; Elklit and Svensson 1997; Lindberg 2006; Kelley 2012).
Conversely, formulas such as ‘elections rigging’, ‘electoral malpractice’, and ‘elections fraud’ fall
within the negative definitions. In general, positive definitions are characterised by the ‘presence of
criteria (or fulfilment of norms) for democratic elections’; the negative ones are instead characterised
by the ‘absence of criteria (or norm-violations) that render elections less-than-democratic or plainly
un-democratic’ (van Ham 2015, 716).

The type of definition – positive or negative – is consequential for empirical research.

Negative definitions refer to intentional acts aimed at modifying election results. In contrast, positive
definitions cover a broader spectrum of behaviours, including intentional and unintentional ones. The
latter, which can be empirically measured as irregularities ‒ for example, the number of null ballots
‒ also depends on the electoral process quality; therefore, it may be helpful to examine them.

As effectively shown (Hartlyn and McCoy 2006; van Ham 2015), the available
conceptualisations of the quality of elections are also distinguished according to the nature of the
criteria used. On the one hand, some approaches refer to universalistic principles based on the theory
of democracy (Dahl 1971) or the standards established by international organisations (Norris 2015);
on the other hand, alternative approaches are based on the perception of the quality of elections by
citizens or parties (Elklit and Reynolds 2005). Of course, the first approach assumes that it is possible
to identify objective standards valid in all countries; the second type maintains that it is impossible to
disregard the context in which the elections are held. The first approach adapts more effectively to
the point of view of comparative politics.

According to Elklit and Svensson, ‘the phrase «free and fair» cannot denote compliance with
a fixed, universal standard of electoral competition: No such standard exists, and the complexity of
the electoral process makes the notion of any simple formula unrealistic. Election observation
requires the simultaneous use of multiple scales to achieve valid and reliable measurements of
complex phenomena’ (1997, 43). This perspective places in antithesis the approach based on
universal norms and that based on a processual conception of the quality of elections. In reality, most

4
conceptions of the quality of elections consider both normative aspects of a universalistic nature and
a complex analysis of the phenomenon characterised by cyclical phases.

Figure 1 The electoral cycle

Source: Institute for Democracy and Electoral Assistance (IDEA), www.idea.int/elections/

Figure 1 helps us to understand the components of the electoral process, which, although with
some differences from case to case, are taken into account by the primary empirical analyses
concerning the quality of elections (Kelley 2013; Simpser 2013; Norris 2014; 2015). The electoral
process is articulated in many phases, most of which fall under the pre-voting period, others concern
the voting period, and a relatively small number concern the post-voting period. The pre-voting period
5
includes several phases, each of which in turn consists of a variable number of rules and activities:
the legal framework (f.i., electoral system); planning and implementation (f.i., electoral calendar);
training and education (f.i., voters’ information); voter registration; electoral campaign (f.i., media
access). On the other hand, the voting period is composed of two dimensions: voting operations and
election day (f.i., vote counting); verification of results (f.i., tabulation of results). Finally, the post-
voting period consists of a single dimension: post-election (f.i., audits and evaluations).

Following this approach, Pippa Norris (2014, 21) affirms that ‘the overarching notion of
electoral integrity refers to agreed-upon international conventions and universal standards about
elections reflecting global norms applying to all countries worldwide throughout the electoral cycle,
including during the pre-electoral period, the campaign, on polling day, and its aftermath’.

From this point of view, the debate between the alleged universality of the norms related – in
this case – to the quality of the electoral process and the supposed Western bias can be resolved by
considering two aspects. First, by pointing out that almost all international organisations have
formally supported international standards on the quality of the electoral process, starting with the
United Nations (UN), the Organization for Security and Co-operation in Europe (OSCE), the
Organization of American States (OAS), and the African Union (AU); secondly, by focusing on
empirical analysis as the only tool capable of determining the gap between international standards –
supported by all regional organisations – and the actual unfolding of the electoral process.

Measures

As in many other fields of political science, investigating the quality of elections requires
empirical analyses aiming to provide evidence. In this case, the research questions related to the main
topic may be challenging, as many aspects of the electoral process are involved, and sometimes ‒ f.i.,
electoral fraud ‒ they involve hidden (mal)practices. Therefore, researchers of the quality of elections
make use of a wide-ranging array of approaches and methods, including old and new institutionalism,
behaviouralism, technological determinism, cultural anthropology, radical theories, and rational
choice (James 2012, 6-15), often criticised for proposing an alternative perspective (James 2020).
This eclecticism notwithstanding, two methodological approaches have been commonly practised so
far. Some researchers adopt an approach based on expert surveys to gather qualified information from
few but skilled individuals; other scholars prefer to analyse electoral results aggregated to various

6
territorial levels, often blending these sources of information with mass surveys targeted at voters and
election officials. According to this partition, here we present two emblematical streams of research
on the quality of elections, namely the Electoral Integrity Project and the Election Performance Index.
Of course, some other approaches concern the empirical study of the quality of elections. However,
EIP and EPI are undoubtedly the two most structured – because they have been subject to a systematic
empirical test – and best able attempts to illuminate the differences between an approach based on
expert judgment – among them, it is worth mentioning Elklit and Reynolds (2005) – and one based
on the observed behaviour of voters.

The Electoral Integrity Project

Launched in 2012, the Electoral Integrity Project (EIP) builds on the framework of the
electoral cycle as initially proposed by IDEA and sketched in the previous sections. As adapted by
the EIP researchers, the electoral cycle is framed in the eleven stages shown in Figure 2 2. Remarkably,
this version of the electoral cycle maintains a comprehensive approach, in the sense that it entails a
full consideration of pre-election activities (phases 1 to 5), campaign regulations (6 and 7), the
election day (8 and 9), and post-election audits and potential judicial disputes (10 and 11).

2
Although the framework remains unaltered, the labels used to identify each phase are changed after the initial proposal
by Norris (2013, 568). We prefer the electoral cycle in the form proposed here inasmuch recently published.
7
 Figure 2 The 11-stage electoral cycle used in the Electoral Integrity Project
Source: Van Ham 2020, 116.

EIP covers many cases including, in principle, all the nation-states worldwide. A limited
number of exclusions is due to practical reasons (Garnett, James and MacGregor 2022, 29): micro-
states with a population of less than 100,000; states without a popularly elected legislature; states
constitutionally endowed with a representative parliament where de facto elections have not been
held since an extended period; one-party systems; and small (and primarily exotic) states with limited
data availability. According to these criteria, from 2012 to 2021, EIP has researched 480
parliamentary and presidential elections held in 169 countries: 87 per cent of the currently existing
194 polities.

According to the advocated methodology, EIP coordinators recruit experts for each country
under examination to obtain the required information. These experts are picked up due to their
knowledge of the involved political system ‒ with special reference to the electoral process ‒ usually
ascertained through their university employments and scientific publications. The general target is
enlisting 40 experts to be interviewed, but significant differences exist around this quantity. This data

8
collection feature brings about a major shortcoming of each expert survey, i.e., the low number of
respondents for some countries3.

To grasp the eleven dimensions of the general concept of electoral integrity shown above in
Figure 2, the questionnaire ‒ beyond the items for identifying countries/elections and some features
of the respondents ‒ contains 49 ‘core questions’ (Norris and Grömping 2019, 29). Data are collected
through Likert scales, and answer modalities are oriented ‒ eventually after a recode ‒ so that a high
score corresponds to a state of high integrity. The general Perceptions of Electoral Integrity (PEI)
index and the score for every stage are calculated as a sum of the related questions and then
standardised to build a 100-point scale. This procedure warrants the availability of raw data usable
both for comparative analyses and in-depth case studies. Figure 3, for instance, shows a classification
of all the currently existing countries with the PEI scores split into five classes of electoral integrity,
from very low to very high.

Figure 3 Electoral integrity worldwide, 2012 to 2018

Source: Norris and Grömping 2019, 5.

3
In the report by Norris and Grömping (2019), the number of interviewed experts spans from two (Antigua and Barbuda)
to 125 (Czech Republic). The authors suggest dealing with caution in eight out of 166 countries due to the low number
of respondents. The following report by Garnett, James and MacGregor (2022), referred to years from 2019 to 2021
instead, covers 142 elections held in 115 countries and acknowledges 27 cases where the number of experts was too low
to administer the questionnaire.
9
 The first wave of PEI data, collected in 2012 and amounting to 20 cases, has been used to
implement test validity and reliability of indicators (Norris, Frank and Martinez i Coma 2013). First,
the PEI index has shown external validity to a reasonable extent. This has been calculated by
correlating PEI scores with independent measures of democracy, such as the political rights and civil
liberties scales proposed by Freedom House4. Second, internal validity means that PEI scores are not
influenced by the individual characteristics of experts expressing their judgments. Third, PEI scores
reach a good level of legitimacy, as they are congruent with the assessments of the public opinion
collected during the sixth round of the World Values Survey by using the same questions previously
used to address the countries’ experts.

The Elections Performance Index

Although also usable for analysing specific countries or elections, the Electoral Integrity
Project is oriented to broad comparative politics. Alternative approaches often focus on a single
country, particularly the United States. Here the electoral management had a long history of
intervention by politicians, academics, and practitioners, largely due to racial issues and their links
with the enfranchisement of black people started in the 1860s. A key feature of the American electoral
landscape is decentralisation, a consequence of federalism that empowers states and counties to
implement elections. This organisation has created very different state legislations about voter
registration and turnout. Therefore, researchers have exploited the possibility of analysing formal
variations and their effects on political participation. While in many countries electoral management
is considered a technical problem, in the United States it is intertwined with hot partisan issues.
Democrats are mainly engaged in adopting expansive registration procedures, while Republicans are
concerned about possible frauds made easier by some types of vote casting, mainly postal. Thus, both
parties think that their electoral fortunes are significantly affected by the makeup of the electoral
process.

The 2000 presidential election and its aftershock powerfully drove the analysis of the electoral
process. Because the troubles of that election originated from the voting equipment ‒ ballot design,
lever machines, punch cards ‒ most efforts addressed the voting technology and the possible solutions
to technological problems. A relevant endeavour in this direction has been the Voting Technology
Project (VTP) jointly launched by Caltech and MIT (Alvarez et al. 2012). Subsequently, the focus

4
https://freedomhouse.org/reports/freedom-world/freedom-world-research-methodology
10
enlarged to target additional problems, such as voter registration, polling places, absentee voting,
election finance, and the overall administrative structure of elections (Alvarez, Atkeson and Hall
2013).

In 2010, building on these experiences, a research group based at MIT Election and Data
Science Lab (MEDSL) launched a new and inclusive approach grounded on the Electoral
Performance Index (EPI) (Stewart 2020). In the same mood as the abovementioned Electoral Integrity
Project, EPI adopts a comprehensive approach considering the election day but also includes
investigating activities preceding and following it. However, while EIP segmented the electoral
process into four phases ‒ pre-election activities, campaign regulations, election day, and post-
election activities ‒ EPI identifies three phases: registration, voting, and counting. Moreover,
according to Gerken (2009), EPI considers two relevant dimensions of the concept of performance:
convenience measures the voting experience of the individual voter and how comfortable she is with
the election procedures; integrity measures the probability that her vote will be correctly included in
the final tally. Figure 4 summarises the EPI approach showing the two dimensions of the state
performance and the three stages of the electoral process.

Figure 4 A rubric for election administration

Source: MIT Election Data Science Lab 2022, 7.

11
 To tap these concepts, the EPI project uses 19 indicators to provide a global view of the
electoral administration deployed by the 50 American states and the District of Columbia5. At present,
they cover federal elections from 2008 to 2020, but it aims to scrutinise the American elections in the
long run. This imposes some limitations on the data collection, as information gathered according to
the EPI methodology is supposed to meet six principles:

1. be from a reliable source;

2. be available and consistent over time;
3. be available and consistent for all states;
4. reflect a salient outcome or measure of good elections;
5. be easily understood by the public and have a relatively unambiguous interpretation;
6. be produced in the near future.

Table 1 details the current indicators used to assemble the Elections Performance Index. By
and large, they have been grouped into seven areas (MIT Election Data Science Lab 2022, 15-20).
Indicators 7 and 16 refer to the online capabilities of a given state; indicators 11, 12, 15, and 18
measure the state’s performance on the hot issue of registration; indicators 2, 13, 14, and 17 are related
to in-person voting on election day; indicators 5 and 6 point to the role played by the military and
other overseas voters; indicators 3 and 4 measure the states’ efficiency in the implementation of the
mail ballots; indicators 9 and 10 concern provisional ballots, namely the temporary vote cast by a
voter whose eligibility is to be ascertained at a later time; finally, indicators 1, 8 and 19 assess the
level of data transparency warranted by a state.

5
The current metrics with 19 indicators were adopted in 2020; the previous version of the metrics made use of 17
indicators.
12
Table 1 List of the indicators used to build the Elections Performance Index
1. Data completeness: The degree to which a state’s local jurisdictions report critical election
statistics to the EAVS
2. Disability access: difference in turnout rates between people who reported having one of
six disabilities and those who reported having none of these disabilities
3. Mail ballots rejected: Number of mail ballots rejected, as a percentage of turnout
4. Mail ballots unreturned: Number of mail ballots unreturned for counting, as a percentage
of turnout
5. Military and overseas ballots rejected: Number of Uniformed and Overseas Civilian
Absentee Voters Act ballots rejected, as a percentage of UOCAVA ballots returned
6. Military and overseas ballots unreturned: Number of UOCAVA ballots unreturned, as a
percentage of ballots distributed
7. Online registration available: Whether a state provides the opportunity for citizens to
register online
8. Post-election audit required: Whether a state requires a post-election audit of election
returns
9. Provisional ballots cast: Number of provisional ballots cast, as a percentage of turnout
10. Provisional ballots rejected: Number of provisional ballots rejected, as a percentage of
turnout
11. Registration or absentee ballot problems: Percentage of non-voters who cite ‘registration
problems’ as the reason for not voting
12. Registrations rejected: Number of registration forms rejected, as a percentage of new
registration forms submitted
13. Residual vote rate: Over- and under-votes as a percentage of turnout
14. Turnout: Number of voters as a percentage of the voting-eligible population
15. Voter registration rate: Percentage of respondents who voted or stated they were registered
16. Voting information lookup tools: The number of voter information lookup tools on a state’s
election website, out of a possible five that are tracked
17. Voting wait time: Average amount of time reported waiting to cast a ballot
18. Electronic Registration Information Center (ERIC) membershipb: a binary coding of
whether a state is a member of ERIC
19. Risk-limiting audit requiredb: states that mandate risk-limiting audits state-wide in statutes
are coded at the highest value, while states that do not conduct risk-limiting audits are coded as
missing
Source: adaptation from MIT Election Data Science Lab 2022, 4; Stewart 2020, 123-124.
a: since 2020, has substituted an old indicator named ‘Disability- or illness-related voting
problems’.
b: added in 2020.

13
 All the above-reported indicators are scaled according to the same procedure. In practice, the
lowest actual score is anchored to 0, while the highest measure is anchored to 1; thus, once
normalised, all the actual values assume a score higher than 0 and lesser than 1. An essential
consequence of this choice is that all indicators are weighted equally. Indicators may be used
according to two different approaches: first, it may be illuminating to examine each indicator
separately to understand how a state performs on a particular aspect; second, indicators may be
combined for developing a summary measure of the performance in an election. This is an advantage
shared with the alternative Electoral Integrity approach.

Adapting existing methodologies: examples from Italy

As reported in the above sections, the investigations of the quality of elections may adopt
alternative points of departure. Generally, they may be summed up according to a simple dichotomy
between approaches based on expert surveys or objective data (citizens’ and poll workers’ surveys
included). There is, however, a large agreement on the fact that, as much as unique, each approach
should be based on some evidence, a point of view firstly vocally advocated by Gerken (2009). This
stance should also be seriously considered in Italy, where the problem of the inquiry ‒ and eventual
reform ‒ of the quality of elections is relatively new. This entails a situation recurrent in many
democracies, the United States being the only exception, where involved people must face a mere
lack of data or, at best, a lack of transparency. For instance, in Italy, neither voters nor poll workers
have been surveyed to know their experience on election day. However, this is a rarity outside the
United States (see at least Clark and James 2017; Partheymüller et al. 2022). Moreover, while the
electoral reforms have been hotly debated for thirty years, the registration, voting and counting
processes still go largely unnoticed by academics, politicians, and public opinion, the officials in
charge of running the elections being the only real experts. Therefore, the evidence available to
interested people is, at best, anecdotal; what matters more, it mainly comes from journalistic reports
that are usually negatively framed. Thus, even a limited number of poll station lines, misconduct by
the poll workers, or delayed result reports are headlined to emphasise the failures of the electoral
administration.

To avoid such an unbalanced account of the electoral processes, it is necessary to dispose of

a battery of indicators for collecting, gathering, and analysing data; explicit rules for aggregating
those indicators in an index are also requested. Both approaches considered above share these
14
features. For instance, EIP data often used for broad comparative analyses can also be used at a more
disaggregated level by exploiting the scores of each of the 49 indicators to describe the quality of
elections in a given country. And it is worth remembering that the Italian case has already been
examined according to the technique provided by the EIP. Figure 5 reports the 0-100 scores conferred
at each stage in Italy’s parliamentary election held in 2018. Being a long-standing democracy, not
surprisingly, it gains a score on the general PEI index greater than the mean reached by all countries
– 69 against 56. Moreover, Italy performs better than the grand mean of the whole sample on all
stages, although in some cases – i.e., party registration and campaign media ‒ the difference is quite
small.

15
 Figure 5 Italy’s performance on the PEI subdimensions, 2018 parliamentary election
Source: Norris and Grömping 2019, 12.

Unlike the EIP, the Caltech/MIT approach has been explicitly elaborated for the analysis of
the United States; thus, no application to other countries has ever been attempted. Here we aim to
apply part of the methodology to a tentative investigation of the Italian case. Within the scope of the
activities of the Voting Technology Project (VTP) launched by the Caltech/MIT research group, a
relatively simple indicator, named the residual vote rate, is defined as follows (Alvarez, Atkeson and
Hall 2013, 24):
16
 A residual vote rate is computed by determining the total number of ballots
cast in a given election race and subtracting the total number of votes cast in the
race. The percentage of ballots that were cast without a vote for a given race is the
residual vote rate for that race.

In practice, the residual vote is calculated as the sum of blank and null ballots or as the
difference between cast and valid votes. On the one hand, an advantage warranted by this indicator
is the availability and comparability of the raw data, usually publicised by electoral officials. On the
other hand, it is affected by some shortcomings, the most important being the impossibility of
distinguishing intentional nonvoting from invalid votes due to miscounting or voters’ errors. In any
case, the residual vote rate is recurrently used to analyse several aspects of the elections, such as the
choices of social minorities and low-income voters (Herron and Sekhon 2005). In the American
landscape, it is also employed to compare in-person and absentee voting. However, it has been
initially created and successfully used to analyse the consequences of different voting technologies.
It is important to recall once more that electoral administration in the United States is overseen by
states, counties, and sometimes municipalities. Therefore, several different ballot casting methods are
used, including paper ballots, punch cards, optical scans, lever machines, and electronic tools, the
latter often referred to as DRE (Direct Recording Electronic voting machines) (Alvarez, Atkeson and
Hall 2013, 41). In the first systematic analysis of this indicator, Ansolabehere and Stewart (2005)
correlated the level of the residual vote with different voting technologies. They discovered that punch
cards and mechanical lever machines perform worse than alternative methods. This damage the rights
of the citizens voting in those precincts where the electoral officials choose them.

While the relative merits of different types of voting equipment is a key issue in the American
elections, the Italian electoral administration cannot be assessed from the point of view of different
voting technologies because since the first elections were held in the 1940s, a paper-and-pencil
method has been used with no exception or innovation6. However, the scores of the indicator can also

6
Recently, some experiments have been done by introducing innovative forms for expressing and counting votes.
However, these experiments should be considered pilot tests, only including a few poll stations rather than involving the
electorate as a whole.
17
be calculated for the Italian elections by using the existing procedures, and they are reported in Figure
6 for the Lower House (Camera dei Deputati) at the national level7.

Figure 6 Residual vote rate in Italy, 1948-2022

Note: data are referred to the Lower House elections.

Values of the residual vote rates maintained below 4 per cent in the first seven elections (1948-
1976), with a peak in 1953, then scores fluctuated about 4-5 per cent in the following four elections
(1979-1992). Values skyrocketed by around 7 per cent in the three elections held from 1994 to 2001,
to plummet again below 4 per cent in the four elections from 2006 to 2018. At the last election in
2022, the residual vote equals 4.5 per cent. How can we account for these trends? As just said, this
indicator has been shaped initially to gauge the performance of different types of voting equipment,
a groundless problem in Italy. Instead, Italian voters seem to react to the different types of electoral
systems used in the seventy years since World War II. All elections from 1948 to 1992 were held
under a PR system, allowing voters to pick up their preferred party. The 1953 election was an

7
Data are taken from the official website of the Italian Ministry of the Interior.
18
exception because a new mixed system provided a seat majority bonus for the most-voted coalition
while maintaining a proportional structure8. Thus, voters were involved in strategic evaluations of
parties and coalitions, which made their political reasoning more complex.

The medium level of the scores of the 1979-1992 period is difficult to explain with reference
to the electoral system. A reform was approved through a referendum for the 1992 parliamentary
election, curbing the number of preference votes admitted for each voter from four to one. This reform
should be crucial for a political and party system where intra-party competition is a key characteristic
of parliamentary elections. However, it is difficult to relate it to the values of the residual vote rate.
Rather, the growing scores of the period could be better accounted for by the mounting party
fragmentation affecting Italy after the 1976 election, when the two most voted parties ‒ Christian
Democrats and Communists ‒ summed up about three-quarters of the total vote.

The three elections of 1994-2001 correspond to the mixed electoral system mocked by
Giovanni Sartori with the label ‘Mattarellum’. The Lower House was partially elected with a PR
system in this case. At the same time, 75 per cent of the parliamentary seats were disputed through a
first-past-the-post system based on single-member districts, a novelty for the Italian voters. Moreover,
implementing this electoral system needed the simultaneous use of two ballots. Overall, this mixed
system was doubtless more intricate than the previous PR and may easily account for the unusual
level reached by the residual vote rate 9.

The new mixed system operated on a de-structured party system due to the consumption of
the historical parties following the fall of the Berlin Wall and the Clean Hands judicial investigation.
The voters, therefore, on the one hand, voted with new – and certainly more complex – rules; on the
other, their vote concerned an almost completely transformed political landscape. This also helps
explain the residual vote trend in the 1994-2001 phase.

The large decrease in the residual vote rate that occurred in the passage from 2001 to the 2006
parliamentary election is unquestionably a remarkable change. It elicited a hot debate due to a

8
The bonus would be provided so that the most-voted coalition reaches 50 per cent of the total votes. As this did not
happen, the bonus was not allocated, the parliamentary seats were distributed on a purely proportional base, and the
provision was immediately deleted.
9
Probably, the residual vote growth during the 1994-2001 phase is also due to the lack of coherence among the electoral
systems at the various levels of government (state, regional, provincial, and municipal). That did not allow voters to learn
the proper functioning of each system, favouring voting errors.
19
journalistic inquiry that pretended the 2006 election had been rigged by transforming blank ballots
into votes for Forza Italia, then the party of the incumbent Prime Minister Silvio Berlusconi 10.
However, not only these accusations did not produce any lawsuits11. Once more, they may be instead
explained with the electoral reform approved by the Italian parliament in December 2005. It
established a mixed system radically different from the pre-existing one. The new rules envisaged a
PR system with a seat majority bonus for the most voted coalition, where voters could support their
preferred party ‒ with no strategic reasoning referred to coalitions ‒ using only one ballot. In a
nutshell, the electoral system used in the 2006-2013 years was extremely simpler than ‘Mattarellum’,
and this should be considered a sufficient condition to vindicate officials from the charge of fraud.
Almost the same holds for the new mixed system used for the 2018 and 2022 parliamentary elections
when residual votes maintained a low level.

Conclusion

The study of the quality of elections is characterised by a plethora of approaches and methods,
which can be traced back to two macro-sets: those based on the perception of qualified observers and
those based on hard and objective data. In this article, we have illustrated the peculiarities of two
prominent approaches – the Electoral Integrity Project and the Elections Performance Index – to
introduce the study of the quality of elections in the Italian political science community. Apart from
the comparative analyses promoted by the EIP, Italian political science has no established tradition
in studying the quality of elections12. Therefore, launching a research program to fill this gap is
worthwhile.

In achieving this objective, it is also worth drawing lessons for Italy from one of the most
analysed case studies: the United States. In comparison with the United States, some profound
differences emerge. First of all, in approaching the analysis of the quality of elections, American
researchers must consider at least two peculiar elements. First, in the US, the rules governing the
administration of elections are set at the state level. Second, electoral administration is a highly

10
https://www.repubblica.it/2006/11/sezioni/politica/polemica-film-deaglio/uccidete-la-democrazia/uccidete-la-
democrazia.html
11
https://www.lastampa.it/cronaca/2007/01/22/news/chiusa-l-indagine-sui-brogli-deaglio-1.37135613/
12
Several studies in the field of law have been conducted on the shortcomings of Italian electoral legislation (i.e., Pavani
et al. 2011; Tarli Barbieri 2018; 2021). The volume edited by Roberto D’Alimonte and Carlo Fusaro (2008) presents,
instead, an intermediate approach between political science and law.
20
politicised issue on which Democrats and Republicans provide opposing and conflicting
interpretations. In contrast, regarding this issue, decentralisation and partisanship are absent in Italy.
This difference must be considered seriously when approaching this issue, starting from the
pioneering research that began in the United States.

The caution applies all the more when using residual voting. In this article, we have followed
one of the paths some US researchers took, starting from the critical 2000’s presidential elections.
Examining the residual vote is the first step in analysing the quality of elections in Italy. It is an
indicator whose increase could indicate some problems in electoral integrity. It cannot be excluded
that in the future, voting technology may also become an explanatory variable in the Italian case. To
date, however, the most interesting independent variables are the type of electoral system and the
format of the party system. Concerning the first variable, it can be assumed that the greater the
complexity of the electoral system, the greater the possibility that voters will cast an invalid vote, thus
increasing the residual vote share. Regarding the second variable, we can speculate that the higher
the number of parties in competition, the higher the ballot confusion and the higher the level of
residual voting.

The start of a research program on the quality of elections in Italy must have at least two
objectives. The first and most immediate is the transition from a national scale of analysis to one at
the sub-national level down to the municipal level. In this way, we will have the opportunity to
evaluate the impact of local contexts on the trend of residual voting. We will thus be able to
understand better when and under what conditions can be attributed to defects in the electoral process.
These include, for example, the complexity of the electoral system and that of the voting paper. The
second, instead, requires the involvement of practitioners and the use of qualitative research
techniques, such as focus groups and in-depth interviews with officials and poll workers. This allows
us to understand the strengths and weaknesses of the Italian electoral process and to find out in which
phases of the electoral cycle harmful elements for the quality of the elections may emerge.

21
 References

Alvarez, R. M., Hall, T. E., & Hyde, S. D. (2008), Election Fraud. Detecting and Deterring
Electoral Manipulation. Washington, Brooking Institution Press.

Alvarez, M. R., Ansolabehere, S., Hall, T. E., Katz, J. N., Rivest, R. L., & Stewart, C. I. (2012).
Voting. What Has Changed, What Hasn’t, and What Needs Improvement. Caltech/MIT, Voting
Technology Project.

Alvarez, R. M., Atkeson, L. R., & Hall, T. E. (2013). Evaluating Elections. A Handbook of
Methods and Standards. Cambridge, Cambridge University Press.

Ansolabehere, S., & Stewart, C. (2005). Residual Votes Attributable to Technology. The
Journal of Politics, 67(2), 365–389. https://doi.org/10.1111/j.1468-2508.2005.00321.x

Bowler, S., Brunell, T., Donovan, T., & Gronke, P. (2015). Election Administration and
Perceptions of Fair Elections. Electoral Studies, 38, 1-9.
https://doi.org/10.1016/j.electstud.2015.01.004

Buckley, F., & Reidy, T. (2015). Managing the Electoral Process: Insights from, and for,
Ireland. Irish Political Studies, 30(4), 445-453. http://dx.doi.org/10.1080/07907184.2015.1100815

Clark, A. (2017). Identifying the Determinants of Electoral Integrity and Administration in

Advanced Democracies: The Case of Britain. European Political Science Review, 9(3), 471–492.
https://doi.org/10.1017/S1755773916000060

Clark, A., & James, T. S. (2017). Poll Workers. In P. Norris & A. Nai (Eds.), Election
Watchdogs. Transparency, Accountability and Integrity (pp. 144–164). Oxford, Oxford University
Press.

D’Alimonte, R., & Fusaro, C. (2008). La legislazione elettorale italiana. Come migliorarla e
perché. Bologna, il Mulino.

Dahl, R. (1971). Poliarchy. Participation and Opposition. New Haven, Yale University Press.

22
 Diamond, L., & Morlino, L. (2004). The Quality of Democracy: An Overview. The Journal
of Democracy, 15(4), 20-31.

Elklit, J., & Reynolds, A. (2005). A Framework for the Systematic Study of Election Quality.
Democratization, 12(2), 147-162, https://doi.org/10.1080/13510340500069204

Elklit, J. & Svensson, P. (1997). The Rise of Election Monitoring: What Makes Elections Free
and Fair? The Journal of Democracy, 8(3), 32-46.

Gandhi, J. & Lust-Okar, E. (2009). Elections Under Authoritarianism. Annual Review of

Political Science, 12, 403-422, https://doi.org/10.1146/annurev.polisci.11.060106.095434

Garnett, H. A., James, T. S., & MacGregor, M. (2022). Codebook – The Expert Survey of
Perceptions of Electoral Integrity, Release 8.0 (PEI_8.0).

Gerken, H. K. (2009). The Democracy Index. Why Our Election System Is Failing and How
to Fix It. Princeton University Press.

Golder, M. (2005). Democratic Electoral Systems Around the World, 1946-2000. Electoral
Studies, 24, 103-121.

Hartlyn, J., & McCoy, J. (2006). Observer Paradoxes: How to Assess Electoral Manipulation.
In Schedler, A. (ed.), Electoral Authoritarianism: The Dynamics of Unfree Competition, 41–56.
Boulder, CO: Lynne Rienner Publishers.

Herron, M. C., & Sekhon, J. S. (2005). Black Candidates and Black Voters: Assessing the
Impact of Candidate Race on Uncounted Vote Rates. Journal of Politics, 67(1), 154–177.
https://doi.org/10.1111/j.1468-2508.2005.00312.x

James, T. S. (2012). Elite Statecraft and Election Administration. Bending the Rules of the
Game?, Basingstoke, Palgrave Macmillan.

James, T. S. (2013). Fixing failures of UK electoral management. Electoral Studies, 32(4),

597-608.

23
 Kelley, J. (2012). Monitoring Elections. When International Election Observation Works, and
Why It Often Fails, Princeton, Princeton University Press.

Lindberg, S. (2006). Democracy and Elections in Africa. Baltimore, Johns Hopkins University
Press.

MIT Election Data Science Lab. (2022). Elections Performance Index. Methodology Report.

Norris, P. (2013). The New Research Agenda Studying Electoral Integrity. Electoral Studies,
32(4), 563–575. https://doi.org/10.1016/j.electstud.2013.07.015

Norris, P. (2014). Why Electoral Integrity Matters. New York, Cambridge University Press

Norris, P. (2015). Why Elections Fail. New York, Cambridge University Press.

Norris, P., Martinez i Coma, F., Nai, A., & Grömping, M. (2016). Perceptions of Electoral
Integrity, Harvard Dataverse.

Norris, P., Frank, R. W., & Martinez i Coma, F. (2013). Assessing the Quality of Elections.
Journal of Democracy, 24(4), 124–135. https://doi.org/10.1353/jod.2013.0063

Norris, P., Cameron, S., & Wynter, T. (Eds.) (2018). Electoral Integrity in America. Securing
Democracy. Oxford, Oxford University Press.

Norris, P., & Grömping, M. (2019). Electoral Integrity Worldwide.

https://www.electoralintegrityproject.com/

Partheymüller, J., Müller, W. C., Rabitsch, A., Lidauer, M., & Grohma, P. (2022).
Participation in the Administration of Elections and Perceptions of Electoral Integrity. Electoral
Studies, 77, 102474. https://doi.org/10.1016/j.electstud.2022.102474

Pavani, G., Pegoraro, L., & Pennicino, S. (2011). Chi controlla le elezioni? Verifica
parlamentare dei poteri, tribunali, commissioni. Bologna, Bononia University Press.

Pitkin, H.F. (1967). The Concept of Representation. Berkeley, University of California Press.

24
 Rombi, S. (2015). Cosa non è e cosa è l’accountability elettorale. Quaderni di Scienza
Politica, 22(1), 89-112.

Schedler, A. (Ed.) (2006). Electoral Authoritarianism: The Dynamics of Unfree Competition.

Boulder, Lynne Rienner Publishers.

Simpser, A. (2013). Why Governments and Parties Manipulate Elections. Theory, Practice,
and Implications. Cambridge, Cambridge University Press.

Stewart, C. (2020). The Elections Performance Index: Past, Present, and Future. In M. Brown,
K. Hale, & B. A. King (Eds.), The Future of Election Administration (pp. 119–153). Springer
International Publishing. https://doi.org/10.1007/978-3-030-14947-5_8

Tarli Barbieri, G. (2018). La legislazione elettorale nell'ordinamento italiano. Milano, Giuffrè

Francis Lefebvre.

Tarli Barbieri, G. (2021). La necessaria ed auspicabile riforma della disciplina del contenzioso
elettorale preparatorio riferito alle elezioni politiche a seguito della sent. 48/2021 della Corte
costituzionale., 2, 658-672.

van Ham, C. (2015). Getting Elections Right? Measuring Electoral Integrity.

Democratization, 22(4), 714-737, http://dx.doi.org/10.1080/13510347.2013.877447.

van Ham, C. (2020). Electoral Integrity. In R. Rohrschneider & J. Thomassen (Eds.), The
Oxford Handbook of Political Representation in Liberal Democracies (pp. 113–133). Oxford
University Press.

25