Italian Political Science, VOLUME 17 ISSUE 1 Published in 2022 under Creative Commons BY-NC-ND license by: Italian Political Science. ISSN 2420-8434. All copyrights of article contents are retained by the authors. Volume 17, Issue 1, 34–54. Contact Author: Nicola Maggini, University of Milan E-mail address: nicola.maggini@unimi.it The Italian space of electoral competition in pandemic times Nicola Maggini UNIVERSITY OF MILAN Cristiano Vezzoni UNIVERSITY OF MILAN Abstract The polls on the voting intentions of Italians during the COVID-19 pandemic revealed substantial stability of electoral orientations in the first phase of the pandemic, while they detected a certain fluidity after the birth of the Draghi government, specifically with a decline of the League and M5s and the growth of Brothers of Italy (FdI). The results of the 2022 general election confirmed those trends with a clear-cut victory of the (centre-) right coalition, this time led by Meloni’s Brothers of Italy. In the meantime, the opponents experienced a poor electoral performance, and an even more deceiving result in terms of seats. All these upheavals have led many pundits to speculate about what would have been the results if the opponents to the right-wing coalition would have succeeded in building a coalition to achieve a higher level of competitiveness in the first-past-the-post electoral districts. But, beyond considerations based on vote intentions or electoral results, to what extent are these speculations consistent with the actual space of electoral competition among main Italian parties? In other words, on which areas of the electoral space does the competition unfold and how did those areas evolve? This article answers these questions using original survey data from the ResPOnsE COVID-19 project. In particular, through the scale analysis of a set of propensity-to-vote (PTV) measures, we investigate the configuration of the electoral competition space in the aftermath of 2022 general election and how (and if) this configuration changes over three distinct phases of the pandemic: during the first wave (spring-summer 2020), during the third wave (spring 2021) and during the fourth wave (autumn-winter 2021). Results show that regardless of the period analysed, party competition occurs mainly within the right, whose party electorates strongly overlap, whereas more barriers exist among party electorates of the opposite camp. 1. Introduction o any observer of the Italian political debate, the attention journalists, politicians and pundits pay to election polls is manifest, with a special focus on the estimates coming from the aggregation of the answers to the question on vote intention. De- spite the notable limitations of such predictive exercises, nowadays broadly recognised in literature (Sturgis et al. 2018), their popularity in the public debate is paramount, not only before an election but also after it and during the whole electoral cycle. If predicting the share of votes of the parties is thus problematic, there is something that is even more arduous if not altogether impossible to achieve by means of vote inten- tion questions. In fact, relying only on them does not allow us to derive direct information on party competition. T MAGGINI and VEZZONI 35 The scope of this article is firstly to illustrate the main limitations of vote intentions as an instrument for describing both proportion of votes and party competition. Sec- ondly, we will discuss an alternative to vote intention that supplies a broader reach, especially as far as the description of the political space of competition is concerned. We refer to propensity to vote (PTV) measures, illustrating their conceptual significance and how they work. Finally, we use PTV to reconstruct the space of competition in the last two years, from the beginning of the pandemic crisis to the general election of 2022. The sections of the article follow this scheme closely, with the addition, after the conceptual discussion of PTVs, of a brief presentation of the data that we use. 2. Vote intentions and their limitations in studying electoral competition The primary use of vote intentions within the framework of electoral polls is to generate a prediction of a forthcoming election. However, in recent years, their use became wide- spread during the whole course of the electoral cycle. It is now customary to have weekly polls and to follow their trend over time. Gathering multiple polls and offering an average for comparison over time also in periods not close to an election has become common practice.1 The attitudes towards this instrument are nonetheless ambivalent. Their limita- tions are, in principle, acknowledged, but in facts they are commented on ‘as if’ they would depict the state of the matter about the preferences of the voters, often having real consequences on the decisions of political actors. Nowadays, a great deal is known about the shortcoming of electoral forecasts based on polls, especially when far from elections (Sturgis et al. 2018). It is nonetheless useful to briefly review the main limitations that affect specifically the vote intention question. The first problem is that a large portion of respondents, generally exceeding a third, do not answer the question, either because they do not know or because they do not want to express their preference. This points to a clear problem of coverage, made more severe by the fact that refusing to express one’s vote intention is often not independent of the respondent’s political profile. In fact, respondents giving a valid answer are more politi- cally interested and engaged than those refusing to answer such a question. And this can significantly bias the estimates. Problems concerning both coverage and measurement error are not the only ones affecting the instrument. There is also a more profound problem connected with it. The main limitation is in fact conceptual. The question gives the interviewee the possibility to choose only one party from a list, while it supplies no information about the attitudes of the respondent towards all other non-selected parties. One could hold that the information we want to know is about the party to be voted for, and thus that is what we ask. If we constrain ourselves to this use, we can hold that the instrument supplies a valid measurement of what it purports to measure (Kelley 1927, quoted in Borsboom 2005, p. 150). It is, though, fair to recognize that such a ques- tion does not allow anything to be said about the preferences concerning parties other 1 The main example in Europe, collecting polls in multiple countries, is Poll of Polls produced by the online outlet politico.eu (https://www.politico.eu/europe-poll-of-polls/). The Italian space of electoral competition in pandemic times 36 than the one selected, and therefore on the competition that exists between parties and the contestable nature of portions of the electorate.2 This limitation is often forgotten and many arguments on the possible electoral gains and losses of parties are in fact derived from the comparison of mere marginal dis- tributions of the vote intention question. In the face of these shortcomings and holding that understanding party competi- tion is a relevant objective, the question is: does a viable alternative to vote intention exist? The answer is, in our opinion, positive. 3. The alternative to vote intentions: propensity to vote For four decades now, political scientists developed a tool for studying voters’ prefer- ences that is referred to as propensity to vote for a party, and denoted by the acronym PTV (Van der Eijk and Franklin 1996; Van der Eijk et al. 2006). PTVs are measured by asking the respondents how likely they are to vote for a party in the future, on a scale from ‘not at all likely’ to ‘very likely’. Crucially, all major parties are tested, supplying a measure of the dyadic relation, in terms of probability to vote, that exists between them and each respondent. Propensities to vote are regularly asked in the questionnaire of the European Elec- tion Study (EES) which supplies the standard formulation of the question: “We have a number of parties in [country] each of which would like to get your vote. How probable is it that you will ever vote for the following parties? Please answer on a scale where 0 means ‘not at all probable’ and 10 means ‘very proba- ble’ (Schmitt et al. 2022). The concept of propensity to vote is not particularly complex to imagine. Indeed, it corresponds to the evaluation that a voter gives of a party, in terms of willingness to vote for it in an undefined future. Economists would indicate it as the utility of the various voting options, which is made explicit and observable precisely through the answers to the question. If the concept is simple to grasp, its conceptual implications are quite profound in terms of representation of the electoral choice process. In fact, considering the utility of the various choice options in the voting function means making clear the fact that the voting choice takes place in two steps. In the first step, the available options are consid- ered, and a specific utility is attributed to each one (propensity to vote); in the second step, a translation rule is applied which leads from the utilities, considered together and compared, to a choice of a preferred option that is, in our case, vote intention. Figure 1 clearly represents this process. 2 To fully understand this limitation, it is enough to refer to an example in market research. Assume that we want to study the pasta market. By asking only which brand of pasta a potential buyer would like to buy (purchasing intention), we would only know their preference for the pasta they choose, and nothing about all the other brands not chosen. It is clear that for an operator in the pasta sector this information would be rather poor and insufficient for designing a market strategy aimed at acquiring new customers, be- cause no information on effective competition between brands would be available. MAGGINI and VEZZONI 37 Figure 1. A two-stage model of electoral choice Source: Van der Eijk et al. 2006, p. 428. These propensities to vote, therefore, represent the immediate antecedents of the choice to vote, and summarize both individual considerations and the images of a party which lead a voter to be more inclined to choose one rather than another. There is something more to the PTVs, a characteristic that clearly distinguishes them from voting intentions and on which their inventors insist. Voting propensities are not ipsative. The adjective ‘ipsative’ indicates a forced choice, where by choosing one op- tion you give up the others, such as when you have to choose a party on the ballot paper. PTVs allow a probability to vote greater than zero to be expressed for several parties. In this way, it is possible to detect if a voter is available to more than one party and, there- fore, to study the potential competition between these parties (for the linkage between electoral availability and political competition, see Bartolini 2000). What is the typical distribution of the propensity to vote for a specific party? First, the majority of respondents usually indicate values equal to zero, expressing a total closure to- wards the party at stake. The other respondents are then distributed over the scale with increasing values, signalling a higher openness of considering the party as a vote option. Thus, the typical distribution of a PTV is skewed to right, with the modal value at zero. When considering the joint distribution of PTV for more parties, we can obtain dif- ferent profiles for our respondents. One can express a high score only for one party, supposedly being certain of her choice and not exposed to competition, while another can indicate two or more high PTVs. Those respondents who assign high and equal or similar scores to two or more par- ties are voters who are potentially open to different options and, therefore, open to competition between parties (Bartolini 2000). On the one hand, PTVs are a straightfor- ward instrument to detect the potential electorate of a party, namely those who express high scores for that party. On the other hand, they are an effective tool to study the struc- ture of the electoral competition, and its evolution over time, precisely because they supply non-ipsative information for several parties. There is a final attractive characteristic of PTVs with respect to voting intention: being a less directive question formulated in a non-ipsative way, it receives far fewer re- fusals and ‘don’t know’ answers from the interviewees, allowing analyses to be The Italian space of electoral competition in pandemic times 38 elaborated on almost the entirety of the sample (for an Italian example, see Vezzoni 2014). Since their ideation in the 1980s, PTVs have proven to be very successful. First and foremost, they became a major tool to study electoral behaviour in comparative perspec- tive (Van der Eijk and Franklin 1996). In the American context, the combined use of PTVs for the Democratic and Republican parties provides a better measure of party iden- tification than the traditional questions developed more than 50 years ago by the Michigan school (Campbell et al. 1960, Paparo, De Sio and Brady 2020). Finally, in the Italian context, PTVs have been extensively used to reconstruct the party space of com- petition, the location of the voters in such a space, and to investigate the left-right dimension without the need to explicitly use the measure that refers to the spatial di- mension of the competition (Schadee, Segatti and Vezzoni 2019). 4. Research questions and data In this article, we provide an empirical analysis of PTVs in the Italian context, describing their distribution at the time of the 2022 general election, and propose an application to study the electoral space of competition based on a multidimensional analysis3 of their joint distributions spanning from 2020 to 2022, a time span dramatically marked by the Covid-19 pandemic. In this context, the electoral outcome of the 2022 general election, with the clear-cut victory of the (centre-)right coalition led by Giorgia Meloni’s Brothers of Italy, has led many pundits to speculate about what would have been the results if the opponents to the right-wing coalition had succeeded in building a coalition to achieve a higher level of competitiveness in the first-past-the-post electoral districts (the so-called majoritarian part of the electoral law). But to what extent are these speculations consistent with the actual space of electoral competition among main Italian parties? In other words, in which areas of the electoral space does the competition unfold and how did these areas evolve since the outbreak of the pandemic? The analysis of party competition over the two years preceding the general election is functional to understanding whether its con- figuration in 2022 is contingent on the electoral campaign dynamics or rather it has more long-lasting roots. To answer these questions, we use data coming from the Re- sPOnsE COVID-19 study, which covers the period 2020-2022 with four waves of data collection on the adult Italian population (Vezzoni et al. 2020, Biolcati et al. 2021, Vez- zoni et al. 2022). The study is intended to investigate the social, political and economic impact of the Coronavirus pandemic on Italian citizens. The study counts over 30,000 interviews over five waves, part of them collected with a panel design. The samples come from a commercial online community of an Italian research institute (SWG), and are collected respecting the quotas for gender, age and geographical area of residence. Specifically, we use here the data coming from waves 1 (spring 2020), 3 (spring 2021), 4 (autumn-winter 2021) and 5 (autumn 2022), counting respectively 15,673, 3 As we will clarify in the next section, the multidimensional analysis is constrained to two dimensions. We are aware that this choice can have some limitations and an alternative solution based on three di- mensions could be employed. Nonetheless, the addition of a further dimension to the classical two- dimensional solution would accommodate for residual segments of the electorate that are in any case marginal and of ambiguous meaning (e.g., voters giving high scores to all parties). MAGGINI and VEZZONI 39 9,222, 3,032 and 4,768 interviews.4 We applied post-stratification weights based on the cross-classification of gender, area of residence, age group, level of education and (for wave 5 only) voting behaviour. In our research, the PTVs were asked for all those parties that reached the 3 per cent electoral threshold:5 Brothers of Italy (Fratelli d’Italia, FdI), the League (Lega), Go Italy (Forza Italia, FI), Action-Italy Alive (Azione-Italia Viva, Azione-Iv), Democratic Party (Partito Democratico, Pd), Green Europe-Italian Left (Europa Verde-Sinistra Italiana, Verdi-SI), Five Star Movement (Movimento 5 Stelle, M5s). Table 1 shows the distribution of the PTVs for the parties, divided into four groups and further distinguishing between those available and not available to a given party, with the threshold set at the value of 6. Table 1. Distribution of the propensity to vote for the main Italian parties in 2022 (percentages, N = 4768) Propensity to vote General election 2022 Not available to the party Available to the party % outcome for the party Don’t know 0 1 to 5 6 to 10 Total On valid votes On total voters FdI 9 43 16 32 100 25.9 15.9 Lega 8 49 21 22 100 8.8 5.4 FI 8 50 23 19 100 8.1 5.0 Azione-Iv 9 51 27 13 100 7.8 4.8 Pd 9 44 25 22 100 19 11.6 Verdi-SI 10 50 26 14 100 3.6 2.2 M5s 8 47 24 21 100 15.4 9.4 Source: ResPOnsE COVID-19. Observing the table, it is possible to detect some of the characteristics of the PTVs discussed above. First of all, it clearly emerges that the majority of respondents give a valid answer to the question and less than 10% adopt the ‘I don’t know’ option. As ex- pected, zero (complete unwillingness to vote for the party) is the modal value, i.e., the category receiving the most responses, for all parties. With increasing values, readiness to vote for the party increases as well. Where to place the cut-point between what we con- sider low and high values of PTV is an arbitrary matter. In the Italian context, it makes sense to indicate 6 as the cut-point, in line with the well-known scale of school marks, 6 being the lowest pass mark. Thus, values between 1 and 5 can still be considered low. Val- ues from 6 upwards are compatible with a potential willingness to vote for the party. FdI is clearly the party that attracts the largest number of potential voters, almost one third 4 In wave 5 the panel component (84.9 per cent of interviews) was integrated with new interviews to main- tain the representativeness of the sample according to the aforementioned quotas. 5 As regards the allocation of PR seats, national thresholds of 3 per cent apply to single party list votes. According to this criterion, we excluded the respondents who voted for +Europe, which in the 2022 elec- tion reached 2.8 per cent. We acknowledge that +Europa, though small, is a relevant option for voters, especially when considering the area of potential support for the Pd. Unfortunately, it is not only a matter of an arbitrary decision to cut parties below 3 per cent but also of available data, as the Ptv for +Europa was asked for only in one wave. However, the conclusions drawn in the paper do not change, including one party below the threshold. The Italian space of electoral competition in pandemic times 40 of the sample (32%). Pd, Lega and M5s follow with slightly more than one fifth (or 20%) of the sample. Lower portions of the sample are available to FI, Verdi-SI and Azione-Iv (below 15%). It should be noted that the percentages are expressed on the total sample N. Thus, for each party, that portion largely exceeds the percentage of votes received in the 2022 election computed on the total number of potential voters, thus including absten- tion, white ballot and invalid votes, as shown in the last column of Table 1. Figure 2 highlights this aspect, comparing the percentage of respondents in the sample that are available to a party (PTV ≥ 6, solid bars) and the actual percentage of votes received in the 2022 elections (computed on the whole electorate). This comparison suggests that the success of a party is not only a function of the size of its potential electorate, i.e. those open to voting for it, but also of the ability of the party to translate that availability into votes. It is clear that the distance between potential voters and actual votes varies largely from party to party. FdI, Pd and M5s seem to be more effective in getting the votes of their potential electorate, while the other parties express poorer performances. The central focus to understanding these differences is on the translation rules from utilities to actual votes (see Figure 1) and, at the end of the day, on competition be- tween parties that share (at least partially) the same potential electorate. Once more, PTVs show their usefulness for studying electoral competition in full strength. In the following paragraph, we will show an innovative method to consider jointly all the PTVs for the main parties and study in this way the whole electoral space of com- petition. Figure 2. Percentage of available voters (PTV≥6) in the sample (N=4867) and percentage of votes received in the 2022 election (Low Chamber) on all voters (N ca. 46*106) for each of the main Italian parties Source: ResPOnsE COVID-19. 0 5 10 15 20 25 30 35 FdI Pd Lega M5s FI Left-Green Azione % PTV>=6 (on all respondents) % votes (on the total voters) MAGGINI and VEZZONI 41 5. The configuration of the political space after the 2022 Italian general election In the previous section we discussed how PTVs provide an effective tool to study the po- litical competition space. In fact, it is precisely by analysing the PTVs for the main Italian parties that we can fully answer the previous research questions, going beyond argu- ments built on the marginal distributions of voting intentions which represent a mere snapshot of the balance of power between parties. The analysis of the PTVs allows us to understand the level of electoral uncertainty and to make counter-factual reasoning about how difficult (or easy) it would have been for voters to vote differently from how they actually did. At an aggregate level, this result supplies an indication of the potential for electoral volatility (van der Eijk and Elkink 2017). This is a particularly relevant topic in light of the great electoral volatility that has characterized Italy and Europe in recent years (Emanuele, Chiaramonte and Soare 2018), as seen in the last political and EU Par- liament elections. In this regard, the 2022 general election marks for Italy the third highly-volatile election in a row, signalling the unprecedented instability experienced by the Italian party system over the past 15 years (Chiaramonte et al. 2023). The space of the Italian electoral competition in 2022 can be described with the di- agram shown in Figure 3 (a so-called Venn diagram).6 The size of the circles is an estimate of the size of each party’s potential electorate (PTV equal to or greater than 6).7 The larger the share of respondents that gave a PTV of at least 6 for a party, the larger its circumference. The overlapping areas between the circles represent the share of poten- tial voters ‘in common’ between two or more parties (i.e., those voters who express a PTV equal to or greater than 6 for the parties in question). It is worth repeating that this is not an estimate of the vote, but an estimate of the propensity to vote and therefore an esti- mate of the electoral potential of each party, namely its attractiveness. If we look at the electoral potential of the main Italian parties after the 2022 Italian general election (data collected in November 2022), we immediately notice that FdI has clearly the biggest elec- toral potential, followed by Pd and M5s, while more limited is the electoral potential of Lega, of FI and, especially, of Azione-Iv. 6 These diagrams were drawn based on the results of a Constrained Multidimensional Scaling (MDS) analysis applying the algorithm developed by Ben Frederickson in Javascript (https://github.com/benfred/venn.js) and implemented in Stata. Constrained MDS improves the clas- sic MDS layout in being aware of subsets and disjoint circles/relationships and this is particularly suitable when there are more than 3 circles. There are several alternatives to the constrained MDS to handle joint distributions of PTV, among which multiple correspondence analysis. Nonetheless, once decided for the space constraints (e.g. two dimension solutions), the results are comparable. Moreover, the option se- lected here, integrating the results with Venn diagrams, conveys a clear idea of the relative size of the electoral potential of a party and its overlapping with the electorates of other parties, which was the scope of the article. 7 Of course, we could have chosen a different cut-point between what we consider low and high values of PTV. In Appendix A we reproduced the political space shown in Figure 3, but with different cut-points for high PTV: at 5 (Figure A1), at 7 (Figure A2) and at 8 (Figure A3). As expected, the diagrams show that when we lower the cut-point, the overlapping areas (voters available to more than one party) increase, whereas these areas decrease (thus voters available to only one party increase) if we raise the cut-point. In particular, it should be noticed that results for PTVs≥8 approximate those for voting choices. We be- lieve that the cut-point we have chosen (Ptv≥6) offers an adequate solution for our purposes, also for the reasons outlined in the previous section. https://github.com/benfred/venn.js The Italian space of electoral competition in pandemic times 42 Figure 3. Venn diagram of the potential electorates of the main Italian parties (PTV≥6), November 2022 Source: ResPOnsE COVID-19. This overall picture is consistent with the outcome of the election, which saw the unprecedented success of the radical-right FdI, led by Giorgia Meloni (who would be- come the first female Prime Minister in Italian history). FdI, indeed, moved from 4.4 per cent of the votes in 2018 to 26 per cent in 2022. This incredible rise came at the expense primarily of FdI’s main allies, Lega and FI, which lost millions of votes compared to the previous election (Chiaramonte et al. 2023). Overall, the centre-right coalition sharply won the other line-ups, i.e., the centre-left coalition led by the Pd, the M5s and the new centrist subject Azione-Iv. A great deal of speculation has been produced about what would have been the results if the opponents to the right-wing coalition had succeeded in building a coalition to achieve a higher level of competitiveness in the first-past-the- post electoral districts and, therefore, a better result in terms of seats. Nevertheless, this argument relies on the assumption that voters in this area were ready to vote for a coali- tion that included parties like Pd, M5s and Azione-Iv, which had clearly distinct platforms and strategies. In this regard, Figure 3 is particularly useful for a twofold rea- son. On the one hand, it shows that there is a strong overlap of the potential electorates of the centre-right parties, as can be seen from the large overlapping areas between the circles of FdI, Lega and FI (i.e., respondents who show a PTV ≥6 for two or three parties of the centre-right coalition). This means that an overwhelming majority of right-wing voters are open to moving from one party to another of the centre-right coalition, reward- ing especially the biggest party (FdI), which has also a significant share of ‘uncontestable’ voters who do not consider the possibility of voting for another party (contrary to FI and Lega). On the other hand, Pd, Azione-Iv and M5s do not share the same potential electorate. Although there is a significant share of voters who are open to voting for both the Pd and M5s, even greater is the share of voters who consider the pos- sibility of voting only for the Pd or only for the M5s. The same occurs as regards the relationship between Pd and Azione-IV (although the loyal voters of the latter are far MAGGINI and VEZZONI 43 fewer, considering also that this party is much smaller than Pd). Furthermore, Azione- Iv is located on the opposite side of the political space compared to the M5s, sharing few potential voters with the party led by Giuseppe Conte, while sharing a significant amount of potential voters with centre-right parties. In this regard, Azione-Iv can be con- sidered a sort of ‘bridge’ between Pd and centre-right parties, consistently with its centrist nature. A similar role is played by Verdi-SI, but in this case the ‘bridge’ is be- tween M5s and Pd. This is consistent with the fact that Verdi-SI ran allied with Pd at the 2022 election, while advocating the need to include the M5s in the alliance. However, Verdi-SI seems to have no reserve of loyal voters: practically all its potential voters con- sider the possibility of voting also for Pd or for M5s (or for both). To sum up, the configuration of the political space based on the PTVs confirms for the electorate what we witnessed at the level of the party elites during the 2022 electoral campaign and afterwards: there is substantial unity on the right, whereas there are many more distinctions in the opposite camp. Where does this configuration of the electoral competition space come from? Is it only the result of the recent events that marked the 2022 general election and the birth of Meloni’s government or does it have deeper roots? To answer this question, in the next section we will analyse the evolution of the party competition space between 2020 and 2021, a time span marked by the pandemic of Covid-19. 6. The configuration of the political space during the pandemic Since the beginning of 2020, Italy has been hit by an intertwining of social, economic and political changes with few precedents in republican history. In March 2020 the Covid-19 pandemic exploded in Italy. In addition to the health emergency and related economic difficulties, the country faced a significant reconfiguration of the political framework. A year after the outbreak of the pandemic, a political crisis concerning the management of the funds of the European economic recovery programme – the Next Generation EU – decreed the end of the Conte II government, which had led the country since the spread of Covid-19. In February 2021 the so-called ‘yellow-red’ government, built on the alliance between the M5s, the Pd, Iv and Leu (Liberi e Uguali, Free and Equal), was replaced by a national unity executive headed by former president of the European Central Bank Mario Draghi and supported by all the main political forces present in parliament, with the exception of the FdI. 6.1. National patterns between spring 2020 and spring 2021 If we look at the electoral potential of the main Italian parties during the first wave of the Covid-19 pandemic (spring-summer 2020, see Figure 4), we immediately notice that Pd, Lega and FdI, followed closely by M5s, have the biggest electoral potential, while that of FI is much more limited. We have to recall that this period was initially marked by a de- cision of the Conte II cabinet to impose a strict national lockdown to contain the spread of Covid-19. After May 2021, the government gradually reopened economic and social activities, thus ending lockdown, while maintaining mandatory face masks, social dis- tancing and isolation measures. This stance was criticized by the right-wing opposition parties, especially FdI and Lega, who suggested more emphasis on the economy. The Italian space of electoral competition in pandemic times 44 Looking at Figure 4, the other fact that immediately catches the eye is the strong overlap of the potential electorates of the centre-right parties, as can be seen from the large overlapping area between the circle of Lega and that of FdI. In particular, the share of potential voters in common with Lega and FdI (i.e. those who show a PTV ≥6 for both parties) is greater than the share of Salvini’s and Meloni’s parties voters who do not se- riously consider the possibility of voting for both parties. The same applies to Forza Italia, with the difference that the latter appears at the centre of the intersections of all potential electorates, showing the vulnerability of the party, competing not only with its centre-right allies, but also (partially) with the Pd and the M5s. Lega and FdI instead show a smaller electoral overlap area with the Pd, while the overlap area with the M5s is greater (similar to that between FI and the M5s). In the latter case, it should be noted that unlike in the past (Emanuele and Paparo 2018), among the centre-right parties it is no longer Lega that shows the greatest overlap with M5s, but FdI. Figure 4. Venn diagram of the potential electorates of the main Italian parties (PTV≥6), spring-summer 2020 Source: ResPOnsE COVID-19. After a year of pandemic, the picture we have just described is substantially con- firmed if we look at the same data for the spring of 2021 (Figure 5) when the national- alliance cabinet led by Draghi had recently replaced the Conte II cabinet, the country was facing a further pandemic wave and the anti-Covid-19 vaccination campaign had started. The analysis of the Italian political space confirms that on the right there is a compact bloc: conservative voters are open to shifting from one party to the other with- out major problems, with the major potential electorates shown by Lega and FdI. Again, a significant share of voters shows high PTVs for both the M5s and the FdI, whose anti- establishment rhetoric resembles that originally characterizing M5s. Berlusconi’s party is confirmed as the most central in the political space, but also as the most vulnerable MAGGINI and VEZZONI 45 because it is subject to multiple competitive pressures: almost all of its potential voters seriously consider the possibility of voting for another party. On the other side, the Pd led by Enrico Letta, like that previously led by Nicola Zin- garetti, is confirmed as the most peripheral within this political space, even if once again it shares a consistent share of voters with the M5s and, to a lesser extent, with the centre- right parties (in particular FI). The isolation of the Pd had already been captured by other analyses that previously relied on Venn diagrams (Emanuele and Paparo 2018), alt- hough the overlap between potential voters of the Pd and M5s has increased over time, probably because of their common experience in both the Conte II and Draghi cabinets. Finally, the M5s is confirmed to be more vulnerable than the Pd, but more crosscutting and attractive also to non-left voters, as we said earlier. Figure 5. Venn diagram of the potential electorates of the main Italian parties (PTV≥6), spring 2021 Source: ResPOnsE COVID-19. 6.2. National patterns in autumn-winter 2021 During the fourth pandemic wave and after the municipal elections held in October 2021, we carried out a further wave of our survey from November 2021 to December 2021 (see Figure 6). We can see some elements of novelty compared to previous survey waves. The electoral potential of the M5s has reduced and, at the same time, it is less cross-cut- ting: indeed, the area of competition between M5s and centre-right parties is smaller. Conversely, the Pd is less isolated in this political competition space. During the years of the pandemic the readiness of right-wing voters to change electoral preferences has ben- efited Meloni’s party, whose position of ‘coherent’ opposition has increased its attractiveness. The Italian space of electoral competition in pandemic times 46 Figure 6. Venn diagram of the potential electorates of the main Italian parties (PTV≥6), autumn-winter 2021 Source: ResPOnsE COVID-19. Another novelty compared to the previous waves is that in the questionnaire we asked for the PTVs also for the smaller parties. In particular, we asked for the PTV for the Italian Left-Progressive Democratic Movement (Sinistra Italiana-Movimento Dem- ocratico Progressista, SI-Mdp),8 for Italy Alive (Italia Viva, Iv) and for Action (Azione). The question was put randomly to one third of respondents for each party. Figure 7 shows the potential electorates of Italian parties, including the minor ones. In general, the po- tential electorates of SI-Mdp, Azione and Iv are much smaller than those of the major parties. Furthermore, the three parties are located in different areas of the political space. The SI-Mdp electoral constituency is a sort of trait d’union between that of the Pd and that of the M5s. This is in line with the favourable attitude shown by the leadership of SI-Mdp as regards the electoral alliance between centre-left parties and M5s, which then collapsed after the crisis of Draghi’s government triggered (among others) by the leader of M5s, Giuseppe Conte. Conversely, both Azione and Iv are more central in the political space: they compete with all parties, in particular with the Pd and FI. That said, it should be noticed that the circles representing the potential electorates of Iv and Azione are both included in the circle of the Pd. This means that their main area of electoral competition, indeed, was with the centre-left party led by Letta. This is not totally surprising given that the leader of Iv, Matteo Renzi, was a former prime minister and leader of the Pd, whereas the leader of Azione, Carlo Calenda, was a former Pd member of the European Parliament. 8 At the time Mdp and SI were together in the wake of the joint electoral list created at the 2018 general election (LeU). For the 2022 general election, Mdp-Article 1 decided to present its candidates within the list of the Pd, while SI created a joint list with the Greens. Similarly, Azione and IV created a joint list. MAGGINI and VEZZONI 47 Figure 7. Venn diagram of the potential electorates of the main Italian parties (PTV≥6) including some minor parties, autumn-winter 2021 Source: ResPOnsE COVID-19. So far, we have analysed the configuration of the space of electoral competition at the national level. However, we know that in Italy territorial differences have always mattered in explaining the electoral results. Therefore, for sake of completeness, we re- peated the analysis using both the 2021 spring data and the 2021 autumn-winter data, disaggregated into three geopolitical zones: North, former Red Belt, South and Islands (see Figure B1 and Figure B2, respectively, in Appendix B). In a nutshell, the configura- tion of the electoral competition space at the macro-area level does not differ much from that previously observed at the national level (Figure 6).9 7. Concluding remarks Focusing on public opinion, this article has examined Italians’ actual electoral prefer- ences, which are the most immediate antecedents of the vote choice. In this regard, we have clarified – on a conceptual and empirical level – the distinction between two instru- ments to detect political attitudes: propensities to vote and the traditional voting intentions. On the basis of the configuration of the propensities to vote, it has been pos- sible to reconstruct the configuration of the overlapping areas between the potential electorates of the main Italian parties, which in turn represent the space of electoral competition. The latter was analysed after the 2022 general election and in the two pre- vious years to understand whether Italians’ political attitudes towards the main parties have changed during the pandemic. In conclusion, this analysis has pointed out some results that can be useful for inter- preting the dynamics of party competition at the time of Covid-19. The first is that in the 9 The actual difference across areas is that, as expected, the size of the potential electorate of the main Italian parties varies according to the zone considered, but to a different degree. Especially, the M5s shows a much smaller potential electorate in the north than in the south (with the electorate in the former Red Belt in an intermediate position), confirming the ‘southernization’ of the party that has taken place since the general election of 2018. Furthermore, the south is the area where there is more competition (consistently with the area’s high electoral volatility during the so-called Second Republic). Compared to the past (with the remarkable exception of the 2018 general election, see Cataldi and Emanuele 2019), the novelty is that the former Red Belt, once an absolutely non-competitive area from an electoral stand- point, now is more similar to the south than to the north. In the latter, the predominance of the centre- right is clear and the possible voting shifts across the left-right divide, in particular between the Pd and the centre-right parties, are minimal. The Italian space of electoral competition in pandemic times 48 electorate there was a significant willingness to change vote choice. This depends not so much on the fact that individual voting propensities radically changed over time, as on the fact that there were multiple availabilities: significant segments of the electorate were effectively contestable between multiple parties. In particular, within the centre- right area, there was a large share of voters open to moving from one party to another. Currently, it is the party led by Giorgia Meloni that is exploiting this greater openness to vote switching within the centre-right area, as shown also by the results of the general election held on 25 September 2022. Undoubtedly, the decline recorded by the Lega and FI at the ballot box is explained by the strong competition exercised on the right by FdI, which exploited its role as the (almost) only parliamentary opposition during the past legislature (2018-2022). This role probably allowed FdI to leverage also the malaise caused by a year of pandemic among specific sections of the population, in particular those most affected by the restrictive measures imposed by the two governments (Conte II and Draghi) to face the pandemic waves. And among the most affected categories are undoubtedly the self-employed, the traditional constituency of the centre-right parties. As we saw in a previous analysis (Ladini and Maggini 2021), Lega voters were more luke- warm than those of the Pd towards the Draghi government, probably being more in line with the anti-establishment position taken by FdI. Similarly, Forza Italia has been sub- ject to much competition, not only on the right (FdI), but also in the centre (Azione-Iv). On the other side of the political space, the common experience of being in govern- ment during the pandemic seems to have made the potential electorates of the Pd and M5s more compatible than in the past. However, the overlap is much less than that found in the centre-right. Furthermore, the Pd on the one hand appears to be the party with the greatest share of ‘loyal’ voters, who do not consider voting for other parties; on the other hand, it appears also as the most peripheral party in the space of political competition, although this isolation decreased after the municipal elections of October 2021. Hence, the Pd on the one hand has a reserve of loyal voters that allows it to be resilient even in the event of adversity; on the other hand, it has to compete for significant shares of votes with both the M5s and Azione-Iv. Furthermore, the Pd shows low attractiveness to cen- tre-right electorates. In general, the left-right dimension in its symbolic meaning (Schadee, Segatti and Vezzoni 2019) still seems to strongly limit the voting movements between areas of the opposite political colour. Compared to the Pd, the M5s and espe- cially Azione-IV appear to be more attractive to right-wing voters. For the M5s, however, the electoral attractiveness cross-cutting left and right boundaries has significantly di- minished compared to the past. Furthermore, this competitive profile means that the M5s is also more sensitive to electoral volatility, with the risk of losing voters in different directions. This occurred mainly towards abstention and, partially, towards FdI in the last general election (Chiaramonte et al. 2023). Electoral volatility could affect even more Azione-Iv, which has a much smaller potential electorate (and fewer ‘loyal’ voters) compared to both the Pd and the M5s. Finally, the M5s potential electorate has signifi- cantly reduced compared to the great success of the 2018 general election, especially in the north, which accentuates its ‘southern’ electoral profile in line with the territorial pattern actually observed in the last general election. Before the electoral campaign for the 2022 general election there was a certain debate on a possible alliance between Pd and M5s, which then failed when Letta refused to MAGGINI and VEZZONI 49 forgive Conte for having triggered the crisis in the Draghi government. In the light of our data, an alliance between the two parties would not have been unreasonable, even if it would have been less organic and feasible than the alliance in the centre-right, given that there are significant shares of voters of both M5s and Pd who rank the (possible) ally poorly. And the latter point is not something contingent, related to the events that marked the fall of the Draghi cabinet and the latest electoral campaign. As shown by our analysis, this aspect characterised both the electorates throughout the entire period marked by the Covid-19 pandemic, when both parties were together in government. An eventual alli- ance including Azione-Iv appears even more difficult, given the low compatibility between the M5s electorate and the Azione-Iv electorate. In other words, as of today, strong reasons for division between the main opposition parties persist, which had pre- vented these parties from running in the same coalition at the 2022 general election. Conversely, the governing coalition can rely on a rather homogenous electoral bloc: right-wing voters are open to switching their vote from one party to another within the coalition. This time they awarded an astonishing result to the previous junior partner FdI, while assuring victory to the overall coalition. This sort of interchangeability across cen- tre-right electorates shows both the existence of a (almost) single right-wing electorate and a high level of competitiveness within the centre-right coalition. As for the first as- pect, it is significant that a single right-wing electorate existed throughout the 2020-2021 time span we analysed, regardless of the fact that the reference parties were together in opposition (Conte II government) or divided (Draghi government). The unity of the right- wing electoral bloc seems to have deep roots: after all, the format of the coalition has more or less been the same since 1994, although party labels have changed over time as well as the balance of power within the coalition (with a clear right-wing turn) and the coalition’s leaders (first Berlusconi, then Salvini, now Meloni). However, the high level of competi- tiveness within the centre-right coalition can be a powerful factor of instability in the relations between the coalition parties and therefore in the long run it can cause troubles for the current government led by Meloni. It all remains to be seen. Of course, future political development could change in the long run the configura- tion of the party competition space we have outlined so far. Furthermore, the structure of the political space and (in)compatibility across electorates is something that goes be- yond the analysis of the propensities to vote. In this regard, more clues could be provided in further research investigating how the different electoral segments as measured through the PTVs are ideologically characterised, both in terms of self-placement on the left-right continuum and in terms of positions on policy issues. Future research, building up on our contribution, might also explore even more innovative solutions to map and visualize multidimensional spaces of electoral competition. References Bartolini, S. (2000), ‘Collusion, Competition and Democracy’, Journal of Theoretical Politics, 12(1): 33–65. Biolcati, F., Vezzoni, C., Ladini, R., Chiesi, A. M., Dotti Sani, G. M., Guglielmi, S., Maggini, N., Maraffi, M., Molteni, F., Pedrazzani, A. and Segatti, P. (2021), ‘Come monitorare la The Italian space of electoral competition in pandemic times 50 risposta dell’opinione pubblica a eventi imprevisti? Il progetto ResPOnsE COVID-19 (How to monitor the public opinion response to unforeseen events? The ResPOnsE COVID-19 project)’, Polis, Ricerche e studi su società e politica, 1: 165-177, https://www.rivisteweb.it/doi/10.1424/100292 Borsboom, D. (2005), Measuring the Mind. Cambridge: Cambridge University Press Campbell, A., Converse, P.E., Miller, W.E. e Stokes, D.E. (1960), The American Voter, Chicago: University of Chicago Press. Cataldi, M. and Emanuele, V. (2019), “Voto sul territorio e competizione nei collegi: una geografia elettorale rivoluzionata”, in A. Chiaramonte and L. De Sio (eds.), Il voto del cambiamento. Le elezioni politiche del 2018, Bologna: il Mulino, pp. 151-175. Chiaramonte, C., Emanuele, V., Maggini, N. and Paparo, A (2023), ‘Radical-Right Surge in a Deinstitutionalised Party System: The 2022 Italian General Election’. South European Society and Politics, https://doi.org/10.1080/13608746.2022.2160088 Emanuele, V., Chiaramonte, A. and Soare, S. (2018), ‘Does the Iron Curtain Still Exist? The Convergence in Electoral Volatility between Eastern and Western Europe’, Government and Opposition, https://doi.org/10.1017/gov.2018.25 Emanuele, V. and Paparo, A. (2018), “Centralità della Lega e isolamento del PD: il nuovo spa- zio politico italiano”, published online on the website of the Italian Centre of Electoral Studies, https://cise.luiss.it/cise/2018/12/21/centralita-della-lega-e-isolamento-del-pd- il-nuovo-spazio-politico-italiano-2/ Kelley, T.A. (1927), Interpretation of educational measurements, Yonkers, NY: World. Paparo, A., De Sio, L. and Brady, David W. (2020), ‘PTV gap: A new measure of party identifi- cation yielding monotonic partisan attitudes and supporting comparative analysis’. Electoral Studies, 63, https://doi.org/10.1016/j.electstud.2019.102092 Schadee, H., Segatti, P. and Vezzoni, C. (2019), L’apocalisse della democrazia italiana. Alle origini dei due terremoti elettorali, Bologna: il Mulino. Schmitt, H., Hobolt, S. B., van der Brug, W. and Popa, S. A. (2022), European Parliament Election Study 2019, Voter Study. GESIS, Cologne. ZA7581 Data file Version 2.0.1, https://doi.org/10.4232/1.13846. Sturgis, P., Kuha, J., Baker, N., Callegaro, M., Fisher, S., Green, J., Jennings, W., Lauderdale, B. E. and Smith, P. (2018), ‘An assessment of the causes of the errors in the 2015 UK Gen- eral Election opinion polls’, Journal of the Royal Statistical Society, Series A: Statistics in Society, 181(3): 757-781, https://doi.org/10.1111/rssa.12329 van der Eijk, C. and Elkink, Johan A. (2017), “How Generational Replacement Undermined the Electoral Resilience of Fianna Fáil”, in M. Marsh, D. Farrell and G. McElroy (eds.), A Conservative Revolution? Electoral Change in 21st Century Ireland, Oxford: Oxford Uni- versity Press, pp. 102-122. Van der Eijk, C. and Franklin, M. N. (1996), Choosing Europe? The European electorate and national politics in the face of union, Ann Arbor (Mich.): The University of Michigan Press Van der Eijk, C. van der Brug, W., Kroh, M. and Franklin, M. (2006). ‘Rethinking the Depend- ent Variable in Voting Behaviour: on the Measurement and Analysis of Electoral Utilities’, Electoral Studies, 25: 424–47. Vezzoni C. (2014), ‘Italian National Election Survey 2013: a further step in a consolidating tra- dition’, Rivista Italiana di Scienza Politica, 1: 81-108, https://www.rivisteweb.it/doi/ 10.1426/76399. https://www.rivisteweb.it/doi/10.1424/100292 https://doi.org/10.1080/13608746.2022.2160088 https://doi.org/10.1017/gov.2018.25 https://cise.luiss.it/cise/2018/12/21/centralita-della-lega-e-isolamento-del-pd-il-nuovo-spazio-politico-italiano-2/ https://cise.luiss.it/cise/2018/12/21/centralita-della-lega-e-isolamento-del-pd-il-nuovo-spazio-politico-italiano-2/ https://doi.org/10.1016/j.electstud.2019.102092 https://doi.org/10.4232/1.13846 https://doi.org/10.1111/rssa.12329 https://www.rivisteweb.it/doi/10.1426/76399 https://www.rivisteweb.it/doi/10.1426/76399 MAGGINI and VEZZONI 51 Vezzoni, C., Ladini, R., Molteni, F., Dotti Sani, G. M., Biolcati, F., Chiesi, A., Maraffi, M., Gug- lielmi, S., Pedrazzani, A. and Segatti, P. (2020), ‘Investigating the social, economic and political consequences of Covid-19: A rolling cross-section approach. Survey Research Methods’, 14(2): 187-194. https://doi.org/10.18148/srm/2020.v14i2.7745. https://doi.org/10.18148/srm/2020.v14i2.7745 The Italian space of electoral competition in pandemic times 52 Appendix A. Figure A1. Venn diagram of the potential electorates of the main Italian parties (PTV≥5), November 2022 Source: ResPOnsE COVID-19. Figure A2. Venn diagram of the potential electorates of the main Italian parties (PTV≥7), November 2022 Source: ResPOnsE COVID-19. FdI FI M5s Lega AzioneIv VerdiSI Pd M5s Pd AzioneIv FdI FI Lega VerdiSI MAGGINI and VEZZONI 53 Figure A3. Venn diagram of the potential electorates of the main Italian parties (PTV≥8), November 2022 Source: ResPOnsE COVID-19. FdI M5s FI VerdiSI Lega AzioneIv Pd The Italian space of electoral competition in pandemic times 54 Appendix B. Figure B1. Venn diagrams of the potential electorates of the main Italian parties (PTV≥6) by geopoliti- cal zone, spring 2021 Source: ResPOnsE COVID-19. Figure B2. Venn diagrams of the potential electorates of the main Italian parties (PTV≥6) by geopoliti- cal zone, autumn-winter 2021 Source: ResPOnsE COVID-19.