3.2 Measurement of variables
3.2.3 Measuring the dependent variable: support for a political system
Support for the political system is operationalised in the three empirical chapters according to the three dimensions of support mentioned in Chapter 2. Linz and Stepan’s (1996, chap. 1) distinction of behavioural, attitudinal, and constitutional dimensions of political consolidation, are applied to distinguish three categories of support. Behavioural support is defined as a lack of anti-system behaviour such as political violence. Attitudinal support refers to perceived legitimacy of the political system. Constitutional legitimacy refers to cooperation with the institutional frameworks and procedures of the political system. Support was measured in close temporal proximity to the institution-building moment under examination, to mitigate challenges noted by Carey (2009, 161) of linking inclusivity to support for the political system, if they are temporally separate (Carey 2009, 161). In Chapter 5, support is
28 See appendix B for a breakdown of parties per communal-group alignment and relevant sources
29 Lindberg (2004) uses the winning party’s share of seats and the second largest party’s share of seats as
Jessica Genauer, Doctoral Thesis, School of Politics and International Relations, Australian National University, November 2018
operationalised as support for elections, in Chapter 6 as support for the new constitution, and in Chapter 7 as support for the elected assembly. Each dimension of support is discretely measured and can exist in isolation from the other two dimensions. However, the more dimensions that are present, the higher the evaluation of confidence in the political system. The indicators used to measure the three dimensions of support in Chapters 5, 6, and 7, are shown in Table 3.12. For Chapters 5 and 6 the indices range from zero to four, and in Chapter 7 from zero to two. Behavioural support is measured as an absence of anti-system behaviour, in particular political violence, during institutional design moments. Violence and protest tend to ‘cluster around election time’ (Hoglund 2009, 415) focusing on electoral contests, referenda or key pronouncements during political transition (Lindberg 2004, 69; Mansfield and Snyder 2005). For Chapter 5, behavioural support for elections is measured as an absence of electoral violence. Electoral violence ‘refers to physical violence and coercive intimidation directly tied to an impending electoral contest or to an announced electoral result’ (Strauss and Taylor 2009, 8); such violence demonstrates that ‘major actors do not see elections as legitimate’ (Lindberg 2004, 104 note 40). Election-related violence includes violence on election day targeting polling booths, polling officers, or voters, and violence during the campaign period that is focused on the up-coming electoral contest.30 In Chapter
6, behavioural support for a new constitution, is measured as an absence of political violence in the lead-up to and on the day of approval of a constitution by referendum or in a constituent body. For Chapter 7 behavioural support for an elected assembly is measured as political violence, mass protest and other behaviour intended to undermine the elected assembly, in the three months following transitional elections. Strauss and Taylor (2009) similarly measured political violence occurring within three months after an election.
It is important to note that, similar to all indicators measured in the study, electoral violence is measured only as a dichotomous variable based on whether or not it was present; the specific groups involved in electoral violence and the severity of the violence is not captured by the study’s operationalisation of this variable. This does not deny that the severity of electoral violence and the groups involved can impact the extent to which electoral violence should be considered an indicator of a lack of political support. However, this study
30 See Lindberg (2004, 69) for comparable operationalisation as ‘cases where there have been no reports of any
follows Strauss and Taylor (2006) and Lindberg (2004) who operationalise electoral violence as a dichotomous variable and do not consider the severity or the identity of the perpetrating groups. This study chooses to sacrifice some degree of qualitative depth for parsimony in determining whether or not electoral violence was present, which lends greatest comparative power.
Table 3.12 Operationalisation of support for the political system
Behavioural support
Absence of anti-system behaviour
Attitudinal support
Perceived legitimacy Constitutional support Cooperation with institutional frameworks and procedures
Index Ch. 5 Support for elections 1. Absence of election- related violence (Yes=1, No=0) 1. Perception of legitimacy: intention to vote in elections (<50 percent) (Yes=1, No=0)
1. Absence of a boycott of elections by political actors associated with a key communal group
(Yes=1, No=0)
2. Absence of low turnout overall or on behalf of one communal group (£33) (Yes=1, No=0) 0-4 Ch. 6 Support for the constitution
1. Absence of violence and mass protest against new constitution in the lead-up to and on the day of referendum OR in the lead-up to and on the day of promulgation by constituent assembly (Yes=1, No=0)
1. Absence of high NO vote for new constitution in referendum OR constituent assembly overall or on behalf of one communal group (>50 percent) (Yes=1, No=0) 1. Absence of boycott of referendum OR of constituent assembly (Yes=1, No=0)
2. Absence of low turnout in constitutional referendum OR abstentions on approval in constituent body (£33) (Yes = 1, No = 0) 0-4 Ch. 7 Support for the elected assembly 1. Absence of anti-system activity such as political violence, mass protest and other activity to undermine the elected assembly within three months of election (Yes=1, No=0)
1. Absence of Low support for elected assembly overall or on behalf of one communal group (£33) (Yes=1, No=0)
0-2
Attitudinal support refers to perceived legitimacy of the political system, and is measured using public opinion surveys; for studies that use survey data to assess support for a political system see, for example, Chu et al. (2008), Dowley and Silver (2002), Inglehart (2003), and McAllister and White (2017). Attitudinal support for elections (Chapter 5) is measured using survey data that questions respondents regarding their intention to vote in
Jessica Genauer, Doctoral Thesis, School of Politics and International Relations, Australian National University, November 2018
the first transitional elections. The excitement surrounding first transitional elections is typically high, with strong intention to vote expected.31 It is expected that between 60 and 80
percent of a population will intend to vote in founding elections (Kostadinova and Power 2007, 367). As a consequence, anything less than 50 percent of respondents claiming an intention to vote is considered low.
Attitudinal support for a new constitution (Chapter 6) was measured by the absence of a high ‘no’ vote, overall or on behalf of one communal group, either in a referendum or a constituent body. Given that a simple-majority is usually required to pass or reject a constitution in a referendum (Lorenz 2005, 346-347), a ‘no’ vote of above 50 percent overall or on behalf of one communal group was assessed as rejection of the constitution. Attitudinal support for the elected assembly (Chapter 7) was measured using survey data and opinion polls that asked participants whether they had confidence in the elected assembly. Given that confidence in institutions ‘is a declaration by citizens that institutions are reliable’, it is typically measured using individual-level perceptions of legitimacy reported in nationally representative surveys, see for example, measurement of political trust by Catterberg and Moreno (2006) and Torcal (2014). Confidence in political institutions in established democracies ranges widely, from as low as 20 percent to close to 80 percent (Catterberg and Moreno 2006, 36-37). During political transition below a third (33 percent or below) confidence in political institutions can be considered problematic, indicating a challenge to political legitimacy.
Constitutional support, that is, cooperation with political institutions and frameworks, is measured in Chapters 5 and 6 as an absence of low voter turnout, and the absence of a boycott of elections or referenda. Turnout is often ‘used as an indirect measure of popular legitimacy’ (Lindberg 2004, 66) with low turnout demonstrating a lack of support (Birch 2010; Gilley 2006), as it is associated with ‘voter apathy and mistrust of the political process (Solijonov 2016, 13). Turnout in established democracies ranges from above 85 percent in Scandanavian countries, to just above 42 percent in the United States (47-48). This means that below a third (33 percent or below) voter turnout can be considered a challenge to the legitimacy of elections. A boycott, defined as active non-participation in political procedures is considered non-cooperation with a political system, as discussed by Beaulieu and Hyde
(2009). If political actors representing one communal group publicly announce a boycott of elections, referenda, or a political body, then boycott was considered present.32