The electoral geography of the 2016 Presidential Election in Portugal

Giorgian-Ionuţ GUŢOIU

Ph.D. Candidate, Doctoral School of Political Science, University of Bucharest

Sabin PANDELEA

Ph.D. Candidate, Doctoral School of Political Science, University of Bucharest

Abstract: Portugal elected a new president in January, this year. While the campaign was rather atypical, with a majority of independent candidates and a low involvement of the parties, we employ here an analysis of the election’s electoral geography, in order to identify if the geographical partisan delimitations influenced the electoral outcome. At this election a clear political geographical divide existed between the urban North and the rural South. Our findings suggest that the geographical distribution of the votes follows the candidates’ ideological identity.

Keywords: electoral geography, election, semi-presidentialism, presidents, Portugal

1. 1. INTRODUCTION

The January 2016 Portuguese presidential election campaign, with its outcome, are a good illustration of Bernard Manin’s audience democracy.1 Not only the winner, Marcelo de Sousa, enjoyed significant media coverage, but he is in fact a media specialist who activated as a TV moderator for over 10 years. His atypical campaign, doubled by the decreased involvement of both Social Democratic Party (PSD) and Socialist Party (PS) in the campaign, emphasized the increased personalization of the Portuguese politics, as well as the important role that media plays in shaping the voters’ decision. Moreover, media’s importance in the former authoritarian state is even more relevant as it enjoys one of the greatest level of trust in Europe (68% in December 2015 compared to France, for instance, with 34%), being the most trusted in Southern Europe and having an increasing trend over the last two years.2 In the same time, social media is increasingly relevant in shaping the electoral participation of the Portuguese voters.3

The January presidential elections were marked by a high number of independents, and a clear win, from the first round with 52% of the votes for the PSD-proposed candidate Marcelo de Sousa, who managed to overcome Sampaio da Nóvoa, who secured 23% of the votes, (an independent backed by several PS important figures and two extra-legislative leftist parties) and Marisa Matias, (BE candidate) with 10%. The great looser was Maria de Bélem, another independent, endorsed by a part of the Socialist Party, who got only a few votes over 4%.

Nonetheless, although at a first glance it may seem that the parties (and partisanship) played a rather minor role, this is far from being accurate. We employ here the electoral geography to test how present partisan geographical delimitations have been transferred to the presidential level during this election. Employing a multi-disciplinary approach, putting together electoral geography and a qualitative analysis of the main candidates’ platforms and their actual campaigns, our enquiry aims at identifying if partisanship plays a role in the Portuguese presidential elections, known for their special neutral status, in the specific context experienced by the today’s democracies: parties’ decline, voters’ dealignment and increased personalization.

As it has already been emphasized by Marina Costa Lobo, the Portuguese politics witnessed a presidentialization process, especially in regard to the relation between the president and the Prime-minister: the latter has become more independent from both the party and the parliament (partly due to the Assembly’s decreased capacity to hold the government accountable after the EU integration), henceforth enjoying a favorable position in the executive compared to the President.4Moreover, the study observed that the parties’ role during elections has decreased, increasing that of the leaders’ and making the personal traits of the candidates essential for the electoral outcome. Illustrative of the personalization’s degree in the Portuguese politics, as well as for the Presidents’ importance, is the role that the last president, Cavaco Silva, has played in the last 30 years. He was prime-minister for 10 years (1985-1995), becoming president afterwards for another 10 (2006-2016). During his period as Prime-minister, he managed to govern backed by two consecutive absolute majorities, pushing a series of reforms, especially in infrastructure, making the political commenters from Lisbon to call the 1985-1995 period as Cavaquismo.5

In a society like the Portuguese one, characterized by a neo-patrimonial nature of the public administration6 exploited by the mainstream parties,7 and doubled by a toleration of corruption (including in the electoral process) by the Portuguese citizens,8 elections represent an interesting political moment.

1. 1.1 Presidents’ role in the Portuguese regime

Portugal opened what was later labeled by Samuel P. Huntignton as the Third Wave of Democratization,9 when the Armed Forces Movement openly opposed Marcelo Caetano, Salazar’s successor, in April 1974. Two years later, Portugal will have a new constitution and start the democratization processes. From the very beginning, Maurice Duverger, in his article on the nature of the French mixed constitutional regime, qualified Portugal as having a semi-presidential constitutional design, characterized by a dual executive.10 Similarly to other such regimes, Portugal is exposed to conflictual relations between the President, and the legislative majority, through cohabitation with the Prime-minister. Thus, in many instances, the president can use his mediator position to weaken the majority supporting the Prime-minister.11 One such example is that of General António Ramalho Eanes, the first democratically elected president, after the adoption of the 1976 constitution: not only had he cohabited with several prime-ministers, but he also engineered his own personal party, Partido Renovador Democrático (PRD), which he used in order to change the government in 1987. As a matter of fact, before the 1982 constitutional change, which excluded the president’s power to dismiss the PM, the presidents made use of their role in the cabinet formation process, in some instances crafting favorable governments (the so-called governos de incitativa presidencial). Even after this constitutional change, which eliminated the possibility of governments of presidential initiative, the presidents managed to influence the cabinet-formation process. One illustration of these practices is that of the second Pedro Passos Coelho cabinet, named by Cavaco Silva, after the October 2015 legislative election and in spite of the fact that a new coalition secured the majority in the Assembly around the Socialist Party.12Consequently, Passos Coelho resisted as PM for less than a month. Cavaco Silva had a similar attitude towards José Sócrates (with whom he cohabitated), by trying to adapt the government and parliament’s decisions to his own, mainly by vetoing pieces of legislation.13 Previously, in 2004, Jorge Sampaio by-passed the impossibility of dismissing the government by dissolving the Assembly, securing a new, favorable, majority at the following election, and consequently a compatible cabinet.14

The relation between the President and the Prime-minister is also affected by the fact that the latter is accountable solely to the Assembly, making the Portuguese regime a case of what Matthew Shugart and John Carey called premier-presidentialism.15Furthermore, as Pedro Magalhães underlined, this affects the presidential elections’ nature: on one hand, the elections do not produce consequences for the government, thus cannot be used to “punish” an inefficient cabinet, and on the other hand the presidents do not lead the Assembly’s majority, therefore are not held responsible (or rewarded) for the legislative parties’ performance.16 Nonetheless, as the Portuguese political scientist observers, both the government activity and especially the parties’ activity influence the presidential election outcome, and, albeit the increased personalization of the campaigns, ideology and party endorsement explain to a great extend the electoral behavior of the Portuguese voters.17

1. 1.2 The 2016 Presidential elections: many candidates, few parties

Although at a first glance the parties’ role in the Portuguese presidential elections may seem as unimportant, in fact, the parties are an unreplaceable political actor in the candidates’ campaigning effort. As the former president Jorge Sampaio emphasized during a conference on semi-presidentialism, the presidential candidates have no chance of winning the electoral competition in the parties’ absence.18Therefore, the high number of independents who run in these elections leaves an open question: what role do the parties (and partisanship) play in the campaign effort, and especially in the electoral outcome?

The 24th of January elections were marked by some novelties. Firstly, ten candidates entered the presidential competition, which is almost double compared to the 2011 election, when just six politicians run for the executive position. Secondly, the partisan support was rather weak, demonstrated by the seven independent candidatures, as well as by the parties’ chief preoccupation with the state budget, which had to be afterwards approved by the European Commission. Moreover, this affluence of independents is not a constant of the presidential elections (See Appendix 1). The parties’ attitudes can be understood as a crisis’ outcome: Portugal signed a Memorandum of Understanding with the Troika (IMF, European Commission, and ECB) in May 2011, thus constraining particularly the executive’s policy, and generally Portugal’s sovereignty – or, to use José Magone’s observation, transforming Portugal from an independent country to an interdependent one.19 The Socialist Party’s position towards the candidates is illustrative in this sense: the main governmental party has not officially endorsed any candidate. Nonetheless, the party’s officials were split between supporting Maria de Belém, ex-president of the political formation, and Sampaio da Nóvoa, University of Lisbon’s honorific rector. The latter had, though, the official endorsement of several extra-legislative parties, like LIVRE/Tempo de Avançar or the Portuguese Workers’ Communist Party, alongside that of former presidents (Ramalho Eanes, Mário Suares, and Jorge Sampaio) and other members of PS and several leftist organizations. Moreover, the current prime-minister and general secretary of the Socialist Party, António Costa, endorsed both de Belém and da Nóvoa,20 describing the first round of elections as a kind of primaries for the Portuguese left.21 Consequently, the partisan mobilization was rather weak, leading to a monotone campaign. Thirdly, the main themes of the campaign were atypical, lacking debates around the most important political issues and outlining a rather neutral, unpartisan, role for the presidential institution in the Portuguese constitutional design.

Marcelo de Sousa’s campaign was particularly atypical: he did not use the specific campaigning instruments (posters, banners, and flyers), preferring a personalized approach, based on the direct contact with the voters.22 This approach was made possible to a great extent by de Sousa’s popularity. Before running for president on behalf of the Social Democrat Party, he hosted various political talk-shows for the previous 16 years, earning him reputation, as well as influence. In fact, the entire campaign was focused on obtaining as much media coverage as possible, making even the candidate selection significantly influenced by the media presence. This particularity led some political commentators to consider this election as a step towards the transformation of the Portuguese regime into a “media-cracy”.23

Regarding the proposed electoral platforms by the main competitors, de Sousa distinguished himself with a program focused on the president’s involvement in “assuring the government’s success”, and promoting a message of consensus, especially between the two components of the executive power.24 The other candidates focused more on leftist massages: Sampaio da Nóvoa, for instance, emphasized the necessity of further developing the welfare system, or the need to solidarize with the Syrian refugees, and overall promoting the principle of equality.25 Marisa Matias, the third runner, had a similar approach, promising to defend the public sector’s interests against the multinationals and foreign banks that are threating the country’s independence, whilst promoting feminism.26 Interestingly, Matias, a European MP for the leftist Bloco de Esquerda, was endorsed by several Syriza MEPs. Moreover, the political parties have had a rather low involvement in the campaign effort, whilst the debates outlined a neutral, low profile for the presidential institution.

1. 2. THE ELECTORAL GEOGRAPHY OF THE 2016 PRESIDENTIAL ELECTION

The electoral geography can be employed for better understanding the role that partisanship plays in the presidential elections. Geographical divisions follow not only electoral lines, but also social and economic ones. Thus, the Northern part is more urbanized than the South, which is predominately rural. Moreover, the North is also more conservative and religious.27 For instance, during the 2007 referendum on decriminalizing abortion, the most votes against were cast in this region. During the last decades, the Northern part distinguished as a base of the center-right parties (Norte and Centro regions, but also the extreme-South region of Algarve), whilst the South favoring the leftist parties (Alentejo).28

These geographical delimitations have their origin in the period following the Carnation Revolution, during which the political and social situation was highly unstable, inheriting the historical North-South differences.29 Thus, the period between 1974 and 1976, known also as the Ongoing Revolutionary Process (Processo Revolucionário Em Curso), witnessed conflicts between the conservative North, which opposed several Communist policies, and the extreme-left, represented by the Portuguese Communist Party (one of the main opposition forces during Salazar’s Estado Novo), which attempted, without success, to install a Soviet-inspired Communist republic through a coup d’état.30

In our analysis of the 2016 Portuguese Presidential Election results, we propose a framework that draws its conceptual structure from developments within electoral geography. The main idea underling the research in electoral geography is that political behavior is related to a relevant geographical context31. Elections are events that take place within particular geographies. We analyze electoral geography at the level of 278 municipal divisions of mainland Portugal. These municipalities are not incorporated in our argumentation only with their role of administrative boundaries, but rather as social, political and economic constructs as a result of time-space dynamics between agency and structure32. Such relations transcend the everyday locale of human life, being rather connected to wider processes at national or global scale33.

1. 2.1 Methodology

To account for the register of geography in electoral behavior we employ spatistics. Working within the electoral geography framework implies bringing into analysis the existence of two geographical phenomena that are mostly neglected in electoral studies, namely contagion and diffusion34. These two processes implies one’s political behavior is not independent from its context instead is influenced by developments at a neighboring location. This was formalized by Waldo Tobler as the first rule of geography: everything is related to everything else, but near things are more related than distant things35. Such structuration in space demands of a contextual approach because of the inter-dependency of voters in their context.

Contagion and diffusion are commonly known in the literature of spatial statistics as spatial dependence. Luc Anselin defines spatial dependence as follows: in general terms, spatial dependence can be considered to be the existence of a functional relationship between what happens at one point in space and what happens elsewhere36. We employ statistical tools that measure the existence of this spatial dependence at global level (at the national level of Portugal) with Global Indicators of Spatial Autocorrelation (GISA) and at the local level (the municipalities) with Local Indicators of Spatial Association (LISA)37. GISA measures the degree of overall clustering within the spatial dataset, whilst LISA identifies local clusters.

We perform these analyses with GeoDa, a free software package, developed by Luc Anselin and his team of researchers38. The software conducts spatial data analysis, geovisualization, spatial autocorrelation and spatial modeling. In order to perform our analyses at both local and global scales we employ the Moran’s I statistic and derivates from it.

To evaluate the global level of clustering we use the Moran’s I statistic. We compute it for each candidate with the univariate function in GeoDa39. This statistic takes values between -1 (indicating perfect dispersion of values in space) and +1 (indicating perfect clustering). A zero value indicates no spatial pattern. Values between 0 and 1 denote positive autocorrelation with the degree of clustering increasing as we approach the value of 1.

Moran’s I is expressed as the linear relation between vectors of observed values, i.e. y, and the weighted average of the values that neighbors y. The weighted average values of y in location i are commonly known as the spatial lag of y in location i. Mathematically the formula for Moran’s I is be expressed as follows:

For our case study N represents the sample of 278 municipalities located in mainland Portugal. Their map is shown in Figure 1. The spatial lag of each location is constructed based on a spatial weights matrix. The matrix comprises all the relations of vicinity existing between municipalities. We use a queen contiguity-based spatial weights with two levels of neighbors accounted for. A queen weights matrix defines a location’s neighbors as those with either a shared border or vertex (in contrast to a rook weights matrix, which only includes shared borders)40.

Figure 1. The 278 municipal divisions of mainland Portugal

The command for univariate global analysis in GeoDa also produces a scatter plot. This is commonly known as Anselin’s Moran scatter plot and depicts a graphic representation of the regimes of spatial association existing in the dataset. On the x -axis are displayed the standardized values for each location, whilst on the y- axis are the standardized values for the spatial lag of each location. The slope of the regression line corresponds with the Moran’s I. Further in our analysis we shall refer to the locations mapped in scatter plots as observations. In Figure 2 we propose a guide for interpreting the scatter plot. The area of the scatter plot can be split in four parts, with each part depicting a type of spatial association:

• - Quadrant I: high values of y surrounded by high values;
• - Quadrant II: low values of y surrounded by high values;
• - Quadrant II : low values of y surrounded by low values;
• - Quadrant IV: high values of y surrounded by low values.

Figure 2. Anselin’s Moran scatter plot interpretation guide

The clusters of high values located in quadrant I point towards areas where the candidate is stronger in comparison with the rest of the country, whilst the clusters located in the third quadrant reveal palaces where the candidate achieved lower performances than in other parts. Further in our spatial analysis we also deconstruct the scatter plots and count the observations located in the first and third quadrant and are shared by candidates together. This will help us to see to what extent different candidates have attracted votes from the same territories.

In order to analyze the spatial dataset with LISA GeoDa offers a tool derivate from the Moran’s I. The univariate Local Moran’s I exploits the fact that global Moran’s I represents a summation of individual cross products41. LISA were first developed by Luc Anselin in a paper in 199542. They serve as a useful tool in exploring spatial data by indicating local spatial clusters (hot spots) and forming a basis for a sensivity analysis (outliers)43. We employ LISA by depicting local clusters on maps of the 278 municipal divisions. These maps serve for comparing the spatial association regimes between candidates thus allowing a more comprehensive image on the general political geographical space at this election.

1. 2.2 Electoral geography analysis

The political geographer John Agnew has described how parties are not only receptacles for electoral purposes, but also mediators between state and society44. Electoral geography becomes in many stances also a geography of loyalties or sympathies towards parties, and understanding electoral geography also serves understanding the political system, the party system or the political culture we are interested in. Henceforth, our spatial analysis of electoral results is performed together with discussions regarding the political system, and the rhetoric during the electoral campaign. Our aim is to include both local and national scale in the analysis by using the geographical place in a wider scope, one called by John Agnew as a place-as-context approach.45

In this section we explore the electoral geographies of interest at this election and account for the particular relationships that exist between these. We search these relationships as a function of the existing political context at that time. Such a discussion also regards the ideological profile of candidates and their position within the political space in relation with the other candidates.

Before we explore the electoral geographies in mainland Portugal, we present the global results at this election. The results aggregated from mainland Portugal, from the autonomous regions of Azores and Madeira and from the Portuguese diaspora. These are presented in Table 1 and can be used as benchmark for comparison with the spatial analysis. Depicted here are the five candidates of interest and also the percentage of the turnout. Marcelo de Sousa, the candidate of the center right PSD, achieved a score slightly above 50%, allowing him to won the race in the first round.

Table 1. Descriptive statistics of the global results

Source: http://www.eleicoes.mai.gov.pt/presidenciais2016/resultados-globais.html

We begin our exploratory analysis of electoral geographies by depicting the degree of spatial autocorrelation existing within the dataset from the 278 municipalities. For this analysis we use the Moran’s I statistic. The scatter plots in Figure 3 and the statistics in Table 2 are deployed in this sense The statistics and the scatter plots can be read in parallel, since the slope of the regression line is also the Moran’s I. Accounting for the degree of spatial autocorrelation entail us to see the extent of geographical polarization and spatial patterns that exist for each candidate.

Table 2. Global Moran’s I statistics

The general idea that we can extract from the statistics in Table 2 indicates the existence of an important geographical polarization. Maria de Bélem is the sole candidate that doesn’t display the existence of a strong geographical polarization. The Moran’s I for de Bélem is 0,14. Her electoral performance is rather scattered thorough the country. The former president of the Socialist Party presents an electoral geography that can be mostly attributed to her independent candidacy. More important is important to note that the degree of clustering in her electoral geography does not resemble those of other left or center left candidates, like Marisa Matias (I=0,47), António da Nóvoa (I=0,53) and Edgar Silva (0,63). Marisa ran with the support of the Left Bloc while Edgar Silva ran with the support of the Communist Party. António da Nóvoa, although also independent as de Bélem, was backed by some socialist structures. Among all the left candidates, Edgar Silva has the highest degree of polarization. It is a similar degree with that of de Sousa, namely 0,63. These statistics show us the degree of geographical polarization for each candidate, yet little can be told about the polarization existing between the candidates. Our hypothesis is more interested in the geographical divide that exists between the candidate of the center right, namely Marcelo de Sousa and the other candidates of the center-left and left.

Figure 3. Anselin’s Moran scatter plots

To further explore the polarization between the left and the right and its correspondence at the geographical level we use Local Indicators of Spatial Association. This enables us to evaluate the clustering at a local level by identifying the actual clusters and visualize them on 278 municipalities map. We computed these statistics with the help of univariate option in GeoDa. The maps shown in Figure 4 present the clusters for each candidate. The identified clusters are those with a level of significance at 0.05.

The maps displaying the LISA are shown for each candidate in Figure 4. They portray the existence of a divide between the north and the south. This split is greater in the case of the candidates with the highest spatial autocorrelation as Edgar Silva and Marcelo de Sousa. Edgar Silva recorded his best scores in the southern half of the country, whilst de Sousa gained his highest scores in the Northern part of the country. The electoral map of these two candidates are in a complete opposition one with the other. Other candidates of the left resemble this pattern, yet to a lesser extent than Silva. Da Nóvoa also performs best in the southern part. However, for da Nóvoa, unlike Silva the Northern part does not reveal itself entirely as a forbidden region. For Marisa Matias the geographical map is polarized only at it two Northern and Southern ends. Matias, similar to Silva and da Nóvoa presents a map with weaker performances in the north and better scores in the south. The map for de Bélem indicates some clusters, yet their geographical position is scattered through the country since small areas with high performances are present both in the south and in the north. In spite of de Bélem’s commitments to the left, her electoral geography does not appear to fit the pattern of the other three left or center left candidates.

Figure 4. LISA maps for the candidates of interest

The analysis of the LISA maps reveal a major divide existing between the Northern and the Southern part of the country. The candidate of the center right, namely Marcelo de Sousa recorded his better performances in the Northern part, while the candidates of the center left and the left gained more votes in the Southern half. We identified with the help of Moran’s I the existence of a polarization within our spatial dataset. Since the Moran’s I is only a global indicator of spatial clustering we employed Local Indicators of Spatial Association to explore the clusters at a more local scale.

Our next section in the exploration of the electoral geography implies a cross analysis between the clusters of the candidates. The LISA maps helped us identify the spatial distribution of hot-spots for each candidate, yet a comparison between the candidates by qualitatively comparing the maps in Figure 4 is preferable to be avoided. In order to search for the dynamics existing between the electoral geographies of candidates we make cross-comparisons between the clusters for each candidates. These clusters are those placed in the first and third quadrants of the scatter plots. The observations in the first quadrant are those indicating high values neighbored by high values, whilst those in the third quadrant indicate low values surrounded by low values. This analysis can depict the extent at which different candidates attracted votes from the same milieu and enable us a better comprehension of how candidates occupied the geographical space at this election. Their presence within the geographical space is an indication of how the candidates were positioned within political space and also how voters perceived them. The relationships between electoral geographies of candidates can reveal tensions, divides or other potential political conflicts existing between political actors.

In this regard, in Table 3 and Table 4 we count the number of clusters shared in common by our candidates of interest. Table 3 depict the clusters of high values, whilst Table 4 the clusters of low values. It is important to mention that in this counting we used all the clusters, including those that are not statistically significant. Clusters that are accounted in these two tables may be missing from representation in the previous LISA maps. The percentages are computed from the total numbers of clusters located in the quadrant of interest for each candidate displayed on the columns.

Table 3. Cross comparison between scatter plots for the 1st Quadrants

Table 4. Cross comparison between scatter plots for the 3st Quadrants

From the data in Table 3 and 4 we see that among all the candidates, de Sousa is the only one who presents an electoral geography constructed in complete opposition with the others. He shares only some scattered patterns with de Belém (16% in the first quadrant and 11% in the third quadrant) who as we have already discussed depicted a rather dispersed electoral geography. As a candidate of the center-right de Sousa shares no clusters of high values with communist Edgar Silva and only one cluster with da Nóvoa. Of the 140 total of clusters such results indicate towards an important polarization between him and the other candidate. These indications of polarization for de Sousa are sustained also by the results drawn from Table 4 depicting the shared clusters of low values, which indicates no major shared clusters with the left candidates.

Moreover there is an important consistency between the electoral geographies of candidates of the left. For example as we see in Table 3 all of the 67 Edgar Silva’s positive clusters are also present in the first quadrant of da Nóvoa. In fact, Silva and da Nóvoa share together the vast majority of clusters, both of positive and negative electoral performances. From all candidates Edgar Silva is the one with the lowest number of positive clusters and with the highest number of negative clusters (177 clusters in the third quadrant whilst only 67 present in the first quadrant). This indicates that the electoral geography of Edgar Silva is constructed as a clear divide between two regions, north versus south. Whilst the former is completely unsympathetic towards Silva, the latter is developed rather as favorable towards him. However, the unsympathetic geography is larger. This is also the case for da Nóvoa, who presents 96 clusters in the first quadrant and 149 observations in the third quadrant. At the other pole, the winner of this election, de Sousa presents the highest number of positive clusters, namely 140. Almost with 50% higher than the ones of da Nóvoa. This is again an indication of the polarization existing between the center right and the left/center-left at this election. The strongest areas for de Sousa were also the weakest areas for da Nóvoa, and vice versa.

Matias and de Bélem are two candidates with a consistency between the positive and negative clusters. For these two candidates there is almost a similar number of observations in the first and the third quadrants. This confirms the previous findings regarding de Bélem and Matias, although a north-south divide exists in their electoral dataset the degree of polarization in their spatial dataset is smaller than for the other candidates. Nevertheless Table 3 and Table 4 show that both Matias and de Bélem poses two electoral geographies that are tributary to what we may define as the left geographical pole. A stronger support in the south and a smaller support in the north.

Analyzing the electoral geographies at the level of the 278 municipalities in mainland Portugal at the 2016 Presidential Election has shown the existence of an important polarization that has mostly an ideological driven basis. We conclude with a discussion about the importance of underling these patterns in the context of analyzing the party system and the electoral politics and processes in Portugal.

1. 3. CONCLUSIONS

Our enquiry suggests that, despite the specificity of the 2016 elections, characterized by a campaign designing an increased neutral role for the President, the partisanship, together with its geographical delimitations, played a role in the electoral outcome. Thus, the electoral geography of the main candidates follows the lines of their ideological identity. Marcelo de Sousa, the candidate proposed by the Centre-right Social Democrat Party won his most votes in the conservative North. Sampaio da Nóvoa, on the other hand, a candidate who enjoyed the support of the Socialist Party, secured most of his votes in the Southern rural region of Alentejo. The finding is even more interesting in the specific context of these elections: a rather low involvement on behalf of the parties and an unusual campaign, especially in the case of de Sousa.

Consequently, we can see that albeit the apparent lack of partisan involvement, the vote more or less followed previous electoral geographic delimitations whilst the structuration of dynamics between electoral geographies of candidates resembles the dynamics within the political space at this election and prior developments within the political culture. In spite of the existing audience democracy at this election a clear geographical divide between the urban North and the rural South existed. This suggest that this election must be analyzed within a contextual (political and geographical) framework. While we identified this cleavage further studies should account for its causes. It may be that, as Giovanni Sartori has argued, television can encourage localism more than nationalization46.

Apendix 1. Candidates’ number in the Portuguese presidential elections.

1 Only for the first round. In the second round the turnout was 77,99%.

2 Initially, four candidates entered the competition, yet two redrew afterwards.

3 Though independent, Sampaio da Nóvoa was endorsed by the Portuguese Workers’ Communist Partyand LIVRE/Tempo de Avançar, both extra-legislative.

Bibliography

AGNEW, John, “Mapping politics: how Context Counts in Electoral Geography”, Political Geography, Vol. 15, No. 2, 1996.

AGNEW, John, Place and Politics in Modern Italy, University of Chicago Press, 2002.

AGNEW, John, Place and Politics: The Geographical Mediation of State and Society, Allen & Unwin, 1987.

ANSELIN, Luc, “Local indicators of spatial association – LISA”, Geographical Analysis, Vol. 27, 1995.

ANSELIN, Luc, Spatial Statistics: Methods and Models, Springer Science & Business Media, 1988.

DE MORAIS, Carlos Blanco, “Parte III da Constituição da República: Semipresidencialismo «on probation»?”, A Constituição Revista, Fundação Francisco Manuel dos Santos, 2011.

DE SOUSA, Luís, “«I don’t bribe, I just pull strings»: Assessing the fluidity of social representations of corruption in Portuguese society”, Perspectives on European Politics and Society, Vol. 9, No. 1, 2008.

DE SOUSA, Luís, “Political parties and corruption in Portugal”, West European Politics, Vol. 24, No, 1, 2001.

DUVERGER, Maurice, “A new Political System Model: Semi-presidential Government”, European Journal of Political Research, No. 8, 1980.

FREIRE, André, António Costa PINTO, O Poder Presidencial em Portugal, Publicações Dom Quixote, Alfragide, 2010.

HUNTINGTON, Samuel P., The Third Wave: Democratization in the late Twentieth Century, University of Oklahoma Press, 1991.

JOHNSTON, R.J., Charles PATTIE, Putting Voters in Their Place: Geography and Elections in Great Britain, Oxford University Press, Oxford, 2006.

LOBO, Marina Costa, “The Presidentialization of Portuguese Democracy?” in Thomas POGUNTKE, Paul WEBB (eds.), The Presidentialization of Politics. A comparative Study of Modern Democracies, Oxford University Press, Oxford, 2005.

MAGALHÃES, Pedro C., “Redes Sociais e Partipação Eleitoral em Portugal”, Análise Social, Vol. 43, No. 3, 2008.

MAGALHÃES, Pedro C., “What are (Semi)Presidential Elections About? A Case Study of the Portuguese 2006 Elections”, Journal of Elections, Public Opinion and Parties, Vol. 17, No. 3, 2007.

MAGONE, José, “The difficult transformation of state and public administration in Portugal. Europenization and the persistence of neo-Patrimonialism”, Public Administration, Vol. 89, No. 3, 2011.

MAGONE, José, “The Logics of Party System Change in Southern Europe” in Paul PENNINGS, Jan-Erik LANE (eds.), Comparing Party System Change, Routledge, London and New York, 1998.

MAGONE, José, Politics in Contemporary Portugal: Democracy Evolving, Lisbon, 2014.

MANIN, Bernard, The principles of representative government, Cambridge University Press, Cambridge, 1997.

MAXWELL, Keneth, “Regime overthrow and the prospects for democratic transition in Portugal” in Guillermo O’DONNEL, Philippe C. SCHMITTER, Laurence WHITEHEAD, Transitions from Authoritarian rule: Southern Europe, JHU Press, 1986 and Ronald H. CHILCOTE, The Portuguese Revolution, State and Class in the transition to democracy, Rowman and Littlefield Publishers, 2010.

O’LOUGHLIN, John, Colin FLINT, Luc ANSELIN, “The Geography of the Nazi Vote: Context, Confession and Class in the Reichstag Election of 1930”, Annals of the Association of American Geographers, Vol. 84, No. 3, 1994.

PAASI, Anssi ,“The institutionalization of regions: a theoretical framework for understanding the emergence of regions and the constitution of regional identity”, Fennia, Vol. 164, No. 1, 1986.

SARTORI, Giovanni, “Videopoltica”, Rivista Italiana di Scienza Politica, Vol. 19, 1989, pp. 185-198.

SHIN, Michael, John AGNEW, Berlusconi’s Italy. Mapping contemporary Italian politics, Temple University Press, 2008.

SHUGART, Matthew, John CAREY, Presidents and Assemblies. Constitutional Design and Electoral Dynamics, Cambridge University Press, Cambridge, 1992.

SOBRAL, José Manuel, “O Norte, o Sul, a raça, a nação – representações da identidade nacional portuguesa (séculos XIX-XX)”, Análise Social, Vol. 39, No. 171, 2004.

TOBLER, Waldo, “A computer movie simulating urban growth in the Detroit region”, Economic geography, Vol. 46, No. 2, 1970.

Online resources:

www.dn.pt

www.publico.pt

www.observador.pt

www.sol.pt

www.rtp.pt

expresso.sapo.pt

www.pop.pt

www.washingtonpost.com

www.medium.com