Public confidence in government: empirical implications from a developing democracy

Abstract

This article explores the determinants of confidence in the Turkish government. We question whether confidence-related questions in mass surveys tap specific support for the incumbent government or tap diffuse support for government as a democratic institution. For this purpose, sociocultural, performance, and party explanations are tested. Four waves of the World Values Survey for Turkey are used as the data set. The article finds that performance and party-based explanations are the most relevant. Turkish citizens place greater emphasis on ‘government as the incumbent’ rather than on ‘government as a democratic institution’. The analysis also reveals the changing influences of both performance and party-based explanations across time, which points to the significance of context. Through a cross-country analysis, the viability of the findings in the Turkish case are evaluated against those of other developing democracies.

Keywords

confidence government public opinion political support developing democracies Turkey

Introduction

Democratic regimes are dependent on citizens’ political support. In representative democracies, political parties, elections, and civil society institutions act as agents of this support. In addition to these institutional channels, however, citizens’ support of a democratic regime and its institutions is important for democratic sustainability. It is the crucial component of regime legitimacy, without which the functioning of the political system and the efficacy of its institutions would be undermined.

Citizens’ support can take various forms. In his influential study, David Easton (1975) differentiates between objects, as well as types, of support. He categorizes objects of support as the political community, the political regime, and the political authorities. Types of support, in turn, are characterized as diffuse and specific. Diffuse support corresponds to the sense of goodwill citizens have about basic aspects of the political system. Specific support is concerned with citizens’ satisfaction with, and evaluation of, the regime and the incumbent government’s performance. Within this framework of analysis, Easton regards confidence in political institutions as a form of diffuse support. Confidence, he argues, reflects a general positive orientation toward a given set of institutions. Lack of confidence, then, implies a general disaffection of the citizenry with the political institutions in question.

The relationship between diffuse support and confidence as formulated by Easton has been accepted uncritically by many scholars. Indeed, their research, influenced by Easton, has mainly focused on the determinants of public confidence across the advanced industrial democracies, because declining levels of confidence in political institutions since the late 1960s were associated with public apathy toward representative institutions. This, in turn, was found to be alarming for democratic sustenance. The research conducted in this vein has relied on confidence questions in mass surveys (Norris, 1999; Pharr and Putnam, 2000).

However, Easton (1975) mentions the empirical difficulty of identifying whether confidence is of a diffuse or specific nature. The reason for this difficulty is the problem about identifying if, how, and under what conditions specific support converts to diffuse support over time. Accordingly, specific support is likely to bring about diffuse support in cases in which citizens’ satisfaction persists beyond a specific point in time. In a similar vein, then, the persistence of citizens’ dissatisfaction with democratic performance and political incumbents carries the risk of bringing about wholesale skepticism about the efficiency of a democratic regime. As a result, the question as to whether declining levels of confidence in political institutions are a result of citizens’ temporary dissatisfaction or democratic regime disaffection becomes a serious empirical challenge.

This article addresses this empirical challenge by analyzing whether citizens’ confidence in political institutions is about diffuse or specific support. The article takes government as the political institution to be considered. The reasons for this choice are twofold. First, together with parliaments and political parties, governments are pillars of representative democratic regimes. In addition, the possibility of a change in government on the basis of election results is a fundamental democratic principle. Second, Easton (1975) makes an explicit association between confidence in government and legitimacy. According to Easton, absence of confidence in government would result in rule through coercive means since it would be extremely difficult for any government to make binding decisions and to put them into practice without citizens’ consent for a government’s right to rule. This consent is what accords legitimacy to the government.

Easton’s point is well supported with further analyses. For instance, Chanley et al. (2000) acknowledge citizens’ compliance with enacted laws when confidence in government is high. Moreover, they refer to the direct political implications of confidence in government and argue that once citizens lose confidence in government they become more likely to ‘support non-incumbent and third party candidates,’ and are more likely to ‘express support for devolution of decision making from federal to state governments on issues such as crime, welfare, and the environment’ (Chanley et al., 2000: 240). Kelleher and Wolak (2007) write about the implications of confidence for the government and suggest that when citizens’ levels of confidence in government are high, governments feel more comfortable in taking the risk of introducing policy innovations. Cases of low confidence in government, in turn, are related to low risk taking in policy-making, instability, and loss of legitimacy.

It should be noted that in all of these accounts the term ‘government’ refers both to ‘government as a democratic institution’ and ‘government as the incumbent.’ Yet a discussion of legitimacy would have more grounding in cases in which citizens refer to ‘government as a democratic institution’ when asked about their confidence in government. Hetherington (1998: 792) argues that since the government is made up of institutions that are run by incumbents, ‘feelings about both should explain trust.’ Indeed, it is exactly the anatomy of these ‘feelings’ that the present article aims to explain. In other words, we seek to understand whether citizens refer to ‘government as a democratic institution’ or to ‘government as the incumbent, its performance and support for it’ when they indicate their confidence levels in this institution.

As is noted, the bulk of research on citizens’ confidence in political institutions focuses on advanced industrial democracies. So far, only a limited number of scholars have attempted to explain confidence levels across developing democracies (Lühiste, 2006; Mishler and Rose, 2001). Yet, we believe that studying citizens’ confidence in political institutions in developing democracies is important because the stability and legitimacy of a fledgling democracy largely depends on citizens’ support. In this regard, accounting for aspects that cause fluctuation in public confidence across time in this context is even more essential.

By using World Values Survey (WVS) data from 1990 to 2007, this article aims to account for Turkish citizens’ confidence in government across four different time periods.¹ The choice to use a single-country analysis was made to control for democratic institutional design, which allows for further control of the robustness of findings across time. Turkey is a developing democracy in a long-term process of accession to the EU; hence, further democratic institutionalization is constantly contested. Citizens’ perceptions and evaluations of democratic institutions are of primary importance in this process, but only a limited number of studies have focused on Turkey (Esmer, 1999; Kalaycıoğlu, 2002, 2008). These studies have primarily used a descriptive approach by statistically comparing Turkish citizens’ confidence levels across different types of institutions. In this article, however, we aim to account for the determinants of Turkish citizens’ confidence in government and their variation across time. Moreover, we aim to test the explanatory power of these determinants in a cross-country setting in order to see whether they are also valid in the context of other developing democracies.

This article is organized as follows. The first section provides a literature review of the competing explanations of confidence-level determinants. The second section will provide a brief review of Turkey and introduce the main hypotheses. In the third section, we present the variables and the statistical model. The last section concludes with a discussion of the theoretical and practical implications of the findings.

Determinants of confidence: Competing explanations

Three schools of thought are influential in the literature on public confidence in institutions. These are sociocultural explanations, performance-based explanations, and party-based or ideological explanations. We consider in turn what each of these contributes to our understanding of confidence in government.

Sociocultural explanations

Sociocultural explanations regard confidence as a type of diffuse support and focus on the nonpolitical factors of deep-rooted social and cultural properties such as interpersonal trust, political interest, civic participation, and post-materialist values. Confidence, then, reflects citizens’ positive orientations toward the regime at large rather than citizens’ evaluations of regime performance (Inglehart, 1999; Mishler and Rose, 2001; Newton and Norris, 2000; Norris, 1999).

Trust is related to confidence in political institutions in both direct and indirect ways. The direct influence of trust is in people’s willingness to cooperate for both political and nonpolitical causes; hence, trusting people are argued to be more positively disposed toward political institutions (Bjornskov, 2006; Zmerli and Newton, 2008). The indirect influence of trust is intermediated through civic participation (Putnam, 1993; Stolle and Rochon, 1998; Uslaner, 2002).

Another strand of research which focuses on deep-rooted social and cultural properties to explain confidence in political institutions is Inglehart’s proposition of post-materialist value change across advanced industrial democracies (1999). Accordingly, declining voter turnout as well as confidence in political institutions is explained by value changes across these countries away from traditional values and state authority and toward individual autonomy and self-expression (Inglehart, 1997, 1999; Inglehart and Welzel, 2005).

Performance-based explanations

Performance-based explanations draw upon the observation that people tend to have confidence in institutions they perceive to be working effectively. A number of studies report a positive association between citizens’ perceptions of national economic performance and their confidence in institutions (Citrin and Green, 1986; Hetherington, 1998; Mishler and Rose, 2001). Economic performance, measured by both individual material well-being and macroeconomic conditions, is found to be important for evaluating both institutions and incumbents (Mishler and Rose, 1997).

Economic conditions, however, are not the only determinants of citizens’ evaluations of government performance. There are also a number of studies that find citizens’ satisfaction with incumbents and their perceptions of a regime’s capacity to rule the country transparently and effectively to be associated with higher levels of confidence in government (Mishler and Rose, 2001; Rohrschneider and Schmitt-Beck, 2002; Rothstein, 1998). These findings are tested and found to be valid for both advanced democracies (Rohrschneider, 2002) and the developing democracies of former communist countries (Mishler and Rose, 2001).

Party-based or ideological explanations

Besides culture and performance, citizens’ political ideology is also observed to have a direct relationship to confidence in institutions. Individuals on the far left are found to be more skeptical about political institutions, which are portrayed as oppressive or unable to represent the real interests of the masses. Alternatively, individuals on the far right, it is argued, develop a close allegiance to political institutions (Listhaug and Wiberg, 1995; Miller and Listhaug, 1999).

Besides political ideology, political partisanship is also found to be significant in explaining confidence levels. It has been shown that the citizens who are supportive of the incumbent party tend to have more confidence in government. In their study, Anderson and LoTempio (2002) empirically confirmed that, for the USA, confidence in government is highest among those people who voted for the incumbent government.

It should be noted that in the literature sociocultural explanations are more frequently used to explain declining levels of confidence in political institutions across advanced industrial democracies, whereas performance-based explanations are utilized more for developing democracies. Studies based on party or ideological explanations are rarer.

Sociocultural explanations may be more relevant for advanced democracies because the stability and predictability of democratic institutions across those countries extend over longer periods of time. A deep-rooted reservoir of experience with democratic rule, in turn, may be reflected in political attitudes and attitudes regarding social participation. However, citizens of developing democracies do not have the advantage of being able to form a certain disposition toward the regime on the basis of its performance. More often than not, democratic institutions are much less predictable across these countries; hence, cultural support for democratic rule is rather fragile. Accordingly, confidence in political institutions is much more likely to be explained by performance-based or by party-based and ideological explanations, which relate better to citizens’ short-term calculations.

Different explanations propose different accounts of citizens’ confidence in political institutions. In Eastonian terms, sociocultural explanations regard confidence as diffuse support; hence, confidence questions in mass surveys are assumed to account for citizens’ evaluations regarding the basic aspects of the political system. In contrast, performance and party-based or ideological explanations treat confidence as specific support, and so confidence questions are related more to citizens’ satisfaction with institutional performance as well as with those political authorities in power.

As already noted, Easton (1975) points to the empirical challenge of whether confidence questions refer more to diffuse than to specific support. When declining confidence levels are of the diffuse support type, as suggested by sociocultural explanations, then a systemic challenge to democratic regimes would be of concern. However, the proponents of sociocultural explanations mention the adaptive capabilities of, in particular, the advanced industrial democracies, where new social values of self-expression and personal autonomy demand new forms of political participation. Yet even these new forms of political participation have to be accommodated by the traditional institutions of political representation as long as representative democracy remains the political regime of these societies. In addition, the viability of sociocultural theories would imply an even more serious systemic challenge to developing democracies because the adaptive capabilities of their institutions in terms of accommodating new social and political movements are much more restricted.

In the case of performance-based explanations accounting better for declining confidence levels, however, the challenge to democratic regimes would be moderated to dissatisfaction with the incumbents and their performance. Accordingly, if the confidence questions measure specific support, then citizens’ evaluations should be of primary concern to incumbents, because they would reflect the citizens’ level of satisfaction with their performance.

Political ideology and party affiliation can also determine citizens’ confidence levels. This type of support can be labeled as too specific, because it would undo both the influence of deep-rooted social and cultural properties and citizens’ performance-based evaluations. In cases of the viability of party-based or ideological explanations, traditional democratic institutions such as political parties and ideologies become particularly relevant. This situation, in turn, would challenge the post-cold-war euphoria about the grand assumption of ideological neutralization and would reveal that the traditional institutions of representative democracies still matter.

Since all three explanations appear to be viable in accounting for confidence in political institutions, it is important to designate the relative explanatory powers of each of them. Mishler and Rose (1997), for instance, found that across post-communist societies the influence of citizens’ evaluations of economic and political performance more strongly related to confidence in institutions than the influence of socialization. Likewise, in their analysis of advanced industrial societies, Newton and Norris (2000) identified evaluations of government performance as the key explanatory variable, while interpersonal trust, civic activism, education, and other socio-demographic variables explained only a minimal percentage of variance in institutional confidence.

Although these studies focus on the implications of the absolute influence of each of these explanations for democratic rule, they do not account for the implications of their relative influence or influences. This point leads us to ask a number of related questions. What should be inferred about the quality of democratic regimes in cases in which confidence levels are explained better by performance than by party-based or ideological explanations? Similarly, does a greater viability of cultural explanations signal disaffection with representative democratic institutions? Alternatively, in cases in which partisanship better accounts for confidence levels, should declining voter turnout as well as declining membership of political parties be alarming for representative democratic rule? Notwithstanding the viability of prior research, this article strives to unravel the implications of the relative importance of sociocultural, performance-based, and party-based or ideological explanations for developing democratic regimes.

Turkey: a case study and hypotheses

As a developing democracy Turkey’s recent history is characterized by two direct military interventions (in 1960 and 1980) and three indirect military interventions (in 1971, 1997, and 2007). Moreover, weakly institutionalized political parties that lack bottom-up solidarity networks exemplify the political system. Military interventions have brought in sporadic and unpredictable periods of political crisis and bred suspicion between the state and the political elite (Keyman and Heper, 1998). Given the uncertainty of political institutional performance, the implications of citizens’ confidence in democratically elected governments are of vital importance for democratic sustainability and institutionalization.

In the Turkish case, one would expect more relevance from performance and party-based or ideological explanations than from sociocultural ones, based on the literature findings above. This is because democratic political institutions in Turkey are not stable and so citizens have yet to form a general disposition toward the regime. Furthermore, the relative explanatory powers of performance and of party-based or ideological explanations are expected to change across time. This change is likely because, due to weak party institutionalization, parties are born and die within short periods of time. Military interventions, weak performance, and lack of solidarity networks are reasons for this rapid party turnover. The consequence, in turn, is the emergence of catch-all parties of short duration (Sayarı, 2007). The Motherland Party (ANAP), which was established after the 1980 military intervention and which had waned by the 2002 elections, is a case in point. Although the duration of the Justice and Development Party (AKP) is likely to be longer, its victory in 2002 owed a great deal to the weak performance of the coalition composed of ANAP, the Nationalist Action Party (MHP), and the Democratic Left Party (DSP).

The discussion leads to the formulation of two main hypotheses, as follows.

H1. Confidence questions in mass surveys in Turkey refer more to either ‘government performance’ or ‘incumbent support’ than to ‘government as a democratic institution.’

H2. The relative explanatory power of performance-based and party-based or ideological explanations changes across time due to contextual variations.

In the case of both hypotheses being verified, it would be safe to conclude that confidence levels refer more to satisfaction or dissatisfaction with regime performance and incumbents; hence, at least for Turkey, confidence cannot be used as a proxy variable for regime legitimacy. If this is so, then the literature on diffuse and specific support will be ‘nuanced’ to take account of the restricted experience of citizens in developing democracies with democratic government.

For the analysis a brief presentation of the parties in power during the period that the WVS was conducted in Turkey (1990–2007) is necessary, since this is the time period our data set covers. Accordingly, ANAP was in power for the 1990–91 period; the Welfare Party (RP) and True Path Party (DYP) coalition was in power for the 1996–97 period; the ANAP-DSP-MHP coalition was in power from 2001 to 2002; and the AKP was in power in 2007. Table 1 reports the means and standard deviations of Turkish citizens’ confidence in institutions for these time periods.

Table 1.

Turkish Citizens’ Confidence in Institutions across Time (Means and Standard Deviations)

	Government	Armed forces	Religious institutions	Press	Police	Parliament	Civil service	Political parties	Courts
1990–91	2.44	3.48	2.92	2.39	2.78	2.67	2.57	.	2.81
	(1.04)	(0.72)	(1.03)	(0.94)	(1.03)	(1.06)	(0.96)	.	(0.97)
1996–97	2.30	3.59	2.76	2.46	2.84	2.35	2.68	1.93	2.86
	(1.07)	(0.68)	(1.05)	(0.97)	(1.04)	(1.05)	(0.86)	(0.89)	(0.98)
2001	2.28	3.42	2.84	2.05	2.87	2.27	2.59	1.87	.
	(1.09)	(0.89)	(1.04)	(0.98)	(1.08)	(1.05)	(0.94)	(0.95)	.
2007	2.74	3.46	2.89	2.06	2.97	2.65	2.50	2.13	3.03
	(1.02)	(0.87)	(0.92)	(0.86)	(0.99)	(0.99)	(0.88)	(0.88)	(0.93)
Total	2.38	3.48	2.82	2.20	2.87	2.41	2.59	1.94	2.90
	(1.08)	(0.82)	(1.03)	(0.97)	(1.05)	(1.05)	(0.92)	(0.93)	(0.97)

Notes: Figures for standard deviations are shown in parentheses below those for means.

The overall results for all time periods show that, among the mentioned institutions, Turkish citizens have highest confidence in the armed forces (a mean of 3.48) and the lowest confidence in political parties (a mean of 1.94).² Regarding confidence in the government, it should be noted that for all time periods Turkish citizens’ confidence in government (a mean of 2.38) is moderately above the midpoint of the confidence scale. Moreover, within each time period there is considerable variation in confidence in government since the standard deviations are high. Hence, although most people cluster near the midpoint of the scale, substantial numbers are found at each extreme.

Figure 1 focuses on Turkish citizens’ confidence in government in more detail and shows that the periods of single-party rule (ANAP in 1990–91 and the AKP in 2007) attracted higher levels of confidence from Turkey’s citizens. This is especially notable in 2007, under the AKP government, when 62 percent of Turkish citizens indicated high confidence in government. The confidence level is around 45 percent on average for previous time periods.

Figure 1.

Differences in Turkish Citizens’ Confidence in Government

Figure 1 shows that confidence levels are higher for single-party governments. For the period 1996–97 the RP-DYP coalition was in power and, according to the 24 December 1995 election results, their aggregate vote percentage totaled 40.56 percent. Alternatively, in 2000, the grand coalition of ANAP, the MHP, and the DSP was in power. According to the 18 April 1999 election results, their vote percentage totaled 53.39 percent. In comparison with these percentages, the vote percentages for the single-party governments of both 1990–91 and 2007 were lower. For the period 1990–91, the vote percentage of ANAP was 36.3 percent and for 2007 the vote percentage of the AKP was 34.43 percent.³ These figures reveal that coalition governments do not indicate higher confidence levels due to their larger aggregate percentage of the constituency. Hence, party affiliation cannot be the sole determinant of confidence in government.

The above descriptive statistics of the aggregate confidence levels signal the viability of performance-based explanations. Notwithstanding the high aggregate vote percentage of the coalition governments under consideration, both of these governments’ periods in office were hardly stable. The victory of the RP in the 1995 elections opened up a series of discussions on the condition of the democratic regime in Turkey. The state establishment, in particular the military, was alarmed by the RP’s electoral victory due to its pro-Islamist roots. This skepticism was aggravated further during the RP-DYP coalition government. The result was the so-called ‘post-modern coup’ of 28 February 1997, when the coalition government resigned due to increased tension between it and the military (Çarkoğlu, 2007a).

The country’s government remained turbulent during the coalition government of ANAP, the MHP, and DSP. Turkey witnessed the most serious economic crisis of recent decades in 2001 under this coalition. During the course of 2001, GDP fell by 9.4 percent and social unrest mounted (Öniş, 2003; Tunç, 2003). In response, the government put in place reforms aimed at economic restructuring, which strengthened the institutional capability of the Turkish economy. In addition, during this period Turkey was given EU candidate status at the Helsinki Summit of 1999 and a series of political reforms were put in place in order to fulfill the Copenhagen criteria (Müftüler-Baç, 2005; Tocci, 2005). However, these latter improvements, on both the economic and political fronts, could not save the government and all three parties were unable to pass the 10 percent threshold needed to enter parliament following the 2002 elections.

Thus, although coalition governments attract higher percentages of aggregate votes, they do not automatically secure higher levels of confidence. The performance of these governments in office matters as well. The unstable rules of both the RP-DYP government and the ANAP-MHP-DSP government seem to have influenced citizens’ confidence levels. In contrast, the AKP’s rule from 2002 to 2007 proved successful on the economic front. On the political front, although the state elite sustained its skepticism of the AKP on the basis of its roots in former pro-Islamists parties, the AKP came out stronger from the political crises during this time.

The brief analysis above reveals two major points. The first is the relevance of both performance and party-based explanations of confidence in government. The second is the relevance of contextual analysis in making sense of the variables under consideration. However, it should also be noted that a simple descriptive analysis falls short of contesting the magnitude of the influence of both performance and party-based variables. In addition, it does not account for the influence of the competing sociocultural variables. Moreover, the determinants explaining the variation in individuals’ confidence levels within each time period cannot be understood from this simple descriptive analysis. As a result, the next section will present the method and the model. The subsequent section will, in turn, present the research findings.

The method and the model

In line with the aforementioned theoretical discussions, we suggest an explanatory model in which differences in Turkish citizens’ confidence in government are conceived as the consequences of (1) sociocultural factors, (2) citizens’ evaluation of government performance, and (3) their ideological positions and party affiliations. Accordingly, the dependent variable of our model is confidence in government. Data on this variable are derived from the World Values Survey question which asks individuals to rate their confidence in government using the following scale: 1 for ‘a great deal of confidence’; 2 for ‘quite a lot of confidence’; 3 for ‘not very much confidence’; and 4 for ‘no confidence at all’. Finally, the dependent variable is recoded as a binary variable, whereby 1 corresponds to ‘high confidence’ and 0 to ‘low confidence’ in government.⁴ Binary logistic regression is used as the analytical method of the study.

Independent variables

The Turkish sample of four waves of World Values Survey data provides multiple indicators for each theoretical framework. The first group of independent variables comprises sociocultural variables that measure interpersonal trust, political interest, post-materialist values, and subjective religiosity.

The second group of independent variables includes performance-based variables that appraise public evaluation of the government’s economic and political performance. In the literature, citizens’ satisfaction with the government’s economic performance is measured through sociotropic and egotropic evaluations. Sociotropic evaluations refer to individuals’ satisfaction with national economic performance and egotropic evaluations refer to individuals’ evaluation of their own well-being (Citrin and Green, 1986). WVS data do not include any question that measures sociotropic evaluations. However, the question in the WVS about one’s satisfaction with household income measures egotropic evaluations. Hence, citizens’ satisfaction with their household income is used as a proxy for citizens’ perception of the government’s economic performance. This variable is labeled as subjective economic performance. Citizens’ satisfaction with the people in national office is used as a proxy for the public perception of incumbents’ political performance. This variable is labeled as perceived political performance.

The third group of independent variables aims to measure the influence of individuals’ political ideology and partisanship on confidence levels. In this regard, we use citizens’ ideological self-positioning on the left–right scale and their intention to vote for the incumbent as the explanatory variables. Finally, sex, income , and age are included in the analyses as the usual background demographic variables (see the Appendix for details of the coding and measurement of all explanatory variables).

We repeat the empirical model presented above for four different time periods. However, data for the perceived political performance variable are available only for two time periods (the WVS’s third and fourth waves). Accordingly, for these two time periods we conduct the analysis both with and without the inclusion of the subjective political performance variable in order to account for the relative impact of the performance-based variables.

Results

Table 2 shows the relevance of the three schools of thought; hence, we suggest that confidence questions in mass surveys tap both specific and diffuse support. Thus, the implications of declining confidence levels are not easy to decipher. The decline may reflect dissatisfaction with the incumbent government or its performance, or both, at a given point in time. Alternatively, randomly distributed sociocultural variables may reflect the greater skepticism of certain citizens, such as post-materialists.

Table 2.

Binary Logistic Analysis of Confidence in Government in Turkey

Variables	1990−91			1996−97			2001			2007
	β	s.e	exp(β)	β	s.e	exp(β)	β	s.e	exp(β)	β	s.e	exp(β)
Sociocultural explanations
Interest in politics	−0.12	0.09	0.88	−0.07	0.07	0.93	−0.13**	0.04	0.88	0.00	0.09	1.00
Interpersonal trust	−0.41	0.32	0.66	−0.10	0.30	0.91	0.24**	0.11	1.27	−0.29	0.45	0.75
Subjective religiosity	0.55**	0.21	1.74	0.44**	0.17	1.55	0.39***	0.11	1.48	0.50**	0.22	1.65
Post-materialism index	−0.33***	0.08	0.72	−0.15**	0.06	0.86	−0.06*	0.03	0.94	−0.17**	0.07	0.85
Ideological/political explanations
Ideological scale	0.10**	0.04	1.10	0.10***	0.03	1.11	0.03**	0.02	1.03	0.19***	0.03	1.21
Vote for incumbent	0.81***	0.22	2.24	1.32***	0.17	3.74	0.75***	0.09	2.12	1.95***	0.20	7.07
Performance-based explanations
Subjective economic performance	0.09*	0.05	1.09	0.10**	0.03	1.10	0.08***	0.02	1.09	0.09**	0.04	1.10
Demographics
Sex	−0.26	0.18	0.77	−0.12	0.14	0.89	−0.25**	0.08	0.78	−0.17	0.18	0.85
Income	−0.22**	0.08	0.80	−0.14***	0.03	0.87	−0.06**	0.03	0.94	−0.11**	0.04	0.90
Age	0.01	0.01	1.01	0.01**	0.01	1.01	0.01**	0.00	1.01	0.00	0.01	1.00
Constant	0.35	0.55		−1.24**	0.42		−0.89***	0.24		−1.56***	0.47
N	630			1060			2536			851
p > χ²	0.000			0.000			0.000			0.000
Pseudo-R²	0.130			0.150			0.054			0.2735

Notes: * significant at the 0.10 level; ** significant at the 0.05 level; *** significant at the 0.001 level.

Notwithstanding the relevance of the three schools of thought for the Turkish case, ideological or party-based explanations are noteworthy. Both ‘ideological position’ and ‘vote for the incumbent’ variables are observed to be highly significant across all time periods, and can be seen to have a large magnitude of impact showing considerable change across time. For the year 2007, for instance, the impact of incumbent support is observed to be greater than in previous years. While prior to 2007 incumbent support increases the likelihood of having high confidence in government by two or three times, in 2007 citizens’ support for the AKP appears to increase the likelihood of having high confidence in government by seven times.

As noted, citizens in Turkey witnessed a series of political and economic crises under the coalition governments prior to the 2002–07 period of AKP rule. None of the parties in these coalition governments entered the parliament in the 2002 elections. Çarkoğlu (2008) notes that the AKP emerged as the alternative party to the center-right parties that collapsed in the 2002 elections. Besides having a political vacuum to fill, the AKP was found to be quite successful in government, especially during its initial years (Çınar, 2006; Toprak, 2005).⁵ As a result, it is reasonable to suggest that the relative stability and success of the AKP government during the 2002–07 period resulted in strong partisanship among citizens. The regression analyses show that ‘confidence in government’ equates to ‘confidence in the incumbent party’, especially for AKP supporters of the 2007 period compared with party supporters in the three prior periods.

Given the partisanship variable’s relatively lower magnitude of impact for periods other than 2007 and the AKP’s relatively short presence on the Turkish political scene, it is too early to comment on the potential endurance and institutionalization of the partisan strength of the AKP’s constituency.⁶ Yet we can comment with some confidence on the merits of analyses conducted across time that reveal partisanship as one of the most influential determinants of confidence in government. Moreover, it is likely that the relative impact of this variable will increase when the incumbent proves successful in office.

Besides the partisanship variable, ideological positioning is also significant across all time periods and its impact shows significant change across time. According to Table 2, citizens become more likely to have confidence in government as they move toward the right of the ideological position scale. In his analysis of this scale in the Turkish context, Çarkoğlu (2007b: 264) finds that ‘Sunni Turks with less education and who are more trusting … are associated with right-wing self-positioning, while those of Alevi or Kurdish background with relative higher levels of education and tolerance … are more prone to self-position themselves on the left-wing end of the L-R [left–right] spectrum.’ Accordingly, due to the greater status quo orientation of those citizens holding right-wing ideological positions, on the one hand, and the predominant presence of right-wing parties in power for the periods under consideration, on the other, it is hardly surprising to find a positive relationship between right-wing leanings and confidence in government.⁷

However, this variable’s magnitude of impact also changes across time. Especially noteworthy is its substantial decline for the 2001 period. According to Table 2, in 2001 the respondents who situated themselves on the far right of the political scale were only 1.3 times more likely to have high confidence in government compared with those who were on the far left of the ideological scale. This can be explained by the economic crisis that Turkey experienced in 2001. It is likely, then, that performance-based evaluations overwhelm ideological commitments at times of crisis and ensuing instability.

Besides ideological or party-based explanations, we also observe performance-based explanations to be significant determinants of citizen confidence in government. We find that respondents who have the highest satisfaction with their economic well-being are approximately 2.5 times more likely to have high confidence in government compared with those who have the least satisfaction with their economic situation. The influence of performance-based explanations, however, becomes more visible once the incumbents’ perceived political performance variable is also included in the analysis. As noted, this variable is available for only two waves of the WVS.

Table 3 shows that once the perceived political performance variable is included in the analysis, this variable becomes the strongest predictor of high confidence in government. For the coalition governments of 1996–97 and 2001, the results indicate that those who are very satisfied with the incumbents are approximately 15 times more likely to have high confidence in government than those who are very dissatisfied. As noted, these two periods were characterized by considerable political and economic crises. The analysis reveals that at least for these time periods, the influence of the perceived political performance variable is the greatest of all the variables. Similar to ideological or party-based variables, performance-based variables reflect citizens’ specific support; hence, citizens in Turkey are more likely to refer to ‘government as the incumbent’ when they are asked about their confidence in government.

Table 3.

Binary Logistic Analysis of Confidence in Government Including Incumbents’ Perceived Political Performance

Variables	1996−97						2001
	Model 1			Model 2			Model 1			Model 2
	β	s.e	exp(β)	β	s.e	exp(β)	β	s.e	exp(β)	β	s.e	exp(β)
Sociocultural explanations
Interest in politics	−0.07	0.07	0.93	−0.030	0.079	0.970	−0.13**	0.04	0.88	−0.13**	0.047	0.875
Interpersonal trust	−0.10	0.30	0.91	−0.310	0.320	0.734	0.24**	0.11	1.27	0.179	0.112	1.196
Subjective religiosity	0.44**	0.17	1.55	0.34**	0.175	1.411	0.39***	0.11	1.48	0.32**	0.113	1.377
Post-materialism index	−0.15**	0.06	0.86	−0.095	0.065	0.909	−0.06*	0.03	0.94	−0.050	0.037	0.952
Ideological/political explanations
Ideological scale	0.10***	0.03	1.11	0.09**	0.032	1.089	0.03**	0.02	1.03	0.045**	0.018	1.046
Vote for incumbent	1.32***	0.17	3.74	0.79***	0.188	2.198	0.75***	0.09	2.12	0.55***	0.093	1.740
Performance-based explanations
Subjective economic performance	0.10**	0.03	1.10	0.06*	0.036	1.066	0.08***	0.02	1.09	0.05**	0.021	1.051
Perceived political performance				0.96***	0.104	2.621				0.82***	0.061	2.270
Demographics
Sex	−0.12	0.14	0.89	−0.241	0.149	0.786	−0.25**	0.08	0.78	−0.24**	0.089	0.783
Income	−0.14***	0.03	0.87	−0.10**	0.034	0.903	−0.06**	0.03	0.94	−0.05*	0.030	0.948
Age	0.01**	0.01	1.01	0.01**	0.005	1.014	0.01**	0.00	1.01	0.003	0.003	1.003
Constant	−1.24**	0.42		−3.05***	0.484		−0.89***	0.24		−2.25***	0.275
N	1060			1047			2536			2521
p > χ²	0.000			0.000			0.000			0.000
Pseudo-R²	0.150			0.211			0.054			0.111

Notes: * significant at the 0.10 level; ** significant at the 0.05 level; *** significant at the 0.001 level.

To grasp the relative influences of party affiliation and performance-based explanations on confidence levels throughout all time periods, Figure 2 compares the predicted probabilities of confidence in government for those individuals who support or do not support the incumbent over the changing values of subjective economic well-being. These graphs provide results for religious males of average income who position themselves at the center of the ideological scale.

Figure 2.

Predicted Probability of Confidence in Government for those Supporting and those Not Supporting the Incumbent across Values of Subjective Economic Well-being

Regarding those respondents who have a tendency to vote for the incumbent, the first graph shows the relatively high probability of having great confidence in government for the year 2007. Moreover, comparing the predicted probability of the individuals who tend to vote for the incumbent and those who do not, the graphs show the considerable impact of incumbent support. According to the probability calculations, in the 2007 period the respondents who intended to vote for the AKP and had the greatest satisfaction with their economic well-being indicated high confidence in government with a probability of 85 percent. However, for the same period, those respondents who had no intention to vote for the AKP, but had the same level of satisfaction with their economic well-being, had high confidence in government with a probability of 46 percent. Similar to the case of the AKP’s governing period, there is a confidence gap between those voting for and those not voting for incumbent parties for the rest of the time periods as well.

Among the sociocultural explanations, subjective religiosity is significant in both Table 2 and Table 3. It should be remembered that our initial expectation was to accord more weight to both ideological or party-based variables and to performance-based variables than to sociocultural variables. Yet it should be noted that in the Turkish context the explanatory power of subjective religiosity is almost equal to the impact of the subjective economic performance variable. This finding can be justified on the basis of the argument of Miller and Listhaug (1999), who indicate that conservative individuals are more likely to develop a strong adherence to political institutions. In the Turkish context, this situation may be exacerbated further by the predominant presence of right-wing conservative parties in power during the time periods under consideration.

Another significant sociocultural variable is the post-materialism index. As is the case for advanced industrial democracies, citizens in Turkey with strong post-materialist values are observed to be more likely to have less confidence in government. Yet it should be noted that the significance of this variable disappears once perceived political performance is included in the analysis. Moreover, its magnitude of influence is the smallest among the other significant variables. Indeed, confidence questions would have reflected diffuse support in cases in which sociocultural variables would appear to be the strongest determinants of confidence. However, this is hardly the case for Turkey.

The impact of both subjective religiosity and post-materialist values appears to decrease for the 2001 period. For instance, Table 2 shows that in 2001 the respondents who had strong post-materialist values were only 0.27 times less likely to have high confidence in government compared with those with the weakest post-materialist values. Moreover, in Table 3, after adding the political performance variable to the model, post-materialist values lose their significance. Thus, we can argue that due to the devastating economic crisis experienced in 2001 the impact of sociocultural variables decreased.

Given these results, what can we say about the implications of mass survey questions on confidence in government? At least in the Turkish context, incumbent support seems relatively more important in explaining confidence in government. This relative importance is replaced with political performance variables for the periods of political and economic crises of 1996–97 and 2001. Hence, it is reasonable to suggest that in Turkey confidence questions in mass surveys refer more to either ‘government performance’ or ‘incumbent support’ than to ‘government as a democratic institution.’ This finding is in line with our first hypothesis. Similarly, our second hypothesis suggesting the relative explanatory power of performance and of party-based or ideological explanations across time due to contextual changes is also verified.

The curious finding in our analysis is the robust importance of subjective religiosity, although its impact diminishes for the periods of coalition government of 1996–97 and 2001. This finding is most likely to account for the relationship between religiosity and the tendency to preserve the status quo.

Turkey in comparison with other developing democracies

In order to find out how far our analysis on Turkey is generalizable to other developing democracies, we conducted a cross-country analysis on 15 developing democracies in the WVS for the 2007 period. Developing democracies are conceptualized along two dimensions. The first dimension is electoral. The presence of free and fair elections is checked for through the Freedom House data set. The second dimension is regime durability. Democratic regimes are recent in these countries in comparison with regime duration across advanced industrial democracies. Hence, developing democracies are defined as either ex-communist countries or countries that have experienced military intervention in their recent history.

In the first column of Table 4 we pool the data for 15 developing democracies and show the individual-level determinants that significantly account for citizens’ confidence in government across developing democracies. In order to disclose whether levels of confidence in government change across developing democracies, and to determine whether country-level contextual explanations provide better elucidation of confidence in government, in the second column we included ‘country dummy variables’ for the developing democracies used in the model. Due to lack of data we do not contest the influence of incumbent support and perceived political performance on confidence levels.

Table 4.

Binary Logistic Analysis of Confidence in Government across Developing Democracies

Variables	Model 1			Model 2
	β	s.e	exp(β)	β	s.e	exp(β)
Sociocultural explanations
Interest in politics	0.18***	0.02	1.20	0.15***	0.02	1.16
Interpersonal trust	0.41***	0.05	1.51	0.44***	0.06	1.55
Subjective religiosity	0.10**	0.05	1.11	0.20***	0.06	1.23
Post-materialism index	-0.04**	0.02	0.96	−0.02	0.02	0.98
Ideological/political explanations
Ideological scale	0.06***	0.01	1.06	0.05***	0.01	1.06
Performance-based explanations
Subjective economic performance	0.07***	0.01	1.07	0.05***	0.01	1.05
Demographics
Sex	−0.01	0.04	0.99	0.002	0.04	1.00
Income	-0.004	0.01	1.00	−0.03**	0.01	0.98
Age	0.005***	0.001	1.00	0.01***	0.001	1.01
Country dummies^a
Slovenia				−1.58***	0.12	0.21
Bulgaria				−1.08***	0.12	0.34
Romania				−1.63***	0.11	0.20
Ukraine				−1.61	0.13	0.20
Moldova				−1.22***	0.10	0.30
Serbia				−1.57***	0.12	0.21
Cyprus				−0.09	0.10	0.92
Georgia				−1.17***	0.10	0.31
Peru				−2.29***	0.11	0.10
Uruguay				0.08	0.11	1.09
Guatemala				−1.07***	0.10	0.34
Ghana				0.50***	0.11	1.65
Mali				0.21*	0.11	1.23
Indonesia				−0.34***	0.10	0.71
Constant	−1.58***	0.11		−0.92***	0.14
N	10,872			10,872
p > χ²	0.000			0.000
Pseudo-R²	0.021			0.122

Notes: ^a Turkey is the reference category.

significant at the 0.10 level; ** significant at the 0.05 level; *** significant at the 0.001 level.

The results show that, for developing democracies, all sociocultural variables are significantly relevant in explaining confidence in government. Especially noteworthy among these variables is the significant and positive influence of the subjective religiosity variable. This finding is in line with the Turkish case. Hence, although religiosity is not presented extensively in the literature on confidence in government, it seems that, irrespective of one’s political convictions, religiosity renders citizens more compliant toward government.

However, in contrast to the Turkish case, interpersonal trust and interest in politics appear as significant determinants in the context of developing democracies. Though sociocultural variables are used more frequently for advanced democracies, these variables seem to be relevant for developing democracies as well. Indeed, comparison of the Turkish case with other developing democracies points to a civic attitudinal deficit for Turkey. Civic attitudes are argued to be the social glue which not only induces cooperation at the societal level, but also helps democratic sustainability in cases of crisis. Hence, the reasons for lack of interpersonal trust in Turkey should be investigated further.

Cross-country analyses also prove the relevance of economic performance and ideological variables. Besides these variables, Model 2 shows that ‘country dummies’ are also significant. Accordingly, we can argue that the levels of confidence in government change considerably across developing democracies. Also, comparing the pseudo-R² of both models in Table 4, we can confidently argue that country-level contextual factors add to the explanation for confidence in government.

In the analyses for Turkey, we argued that citizens refer more to ‘government as incumbent support or performance’ than to ‘government as a democratic institution’ when they are asked about their confidence in government. The robust and relatively stronger explanatory power of partisanship and performance variables accounted for this conclusion. Yet in the analyses of developing democracies, we observe that sociocultural variables are as significant as the economic performance variable. A caveat for this finding is the absence of partisanship and political performance variables in cross-country analyses. Nevertheless, the relatively stronger explanatory power of the sociocultural variables for these latter analyses is noteworthy.

It should be noted that our initial expectation of the greater relevance of partisanship and performance-based variables for Turkey rested on the assumption that in developing democracies citizens have yet to form a stable perception about a democratic regime and its institutions. Although this assumption seems to hold for the Turkish case, it can hardly be generalized to other developing democracies. A positive and significant relationship between civic attitudes and confidence in government proves that these attitudes reinforce citizens’ perceptions of government as a relevant democratic institution. This is good news for developing democracies because it signals support for democratic rule independent of government performance. This finding also introduces a caution for the Turkish case because it points to an absence of civic attitudinal underpinnings for democratic governments in Turkey. It is likely that for the Turkish case, citizens’ perception of government is very sensitive to incumbent support or performance.

Conclusion

Confidence questions in mass surveys are likely to catch both the specific and diffuse support that citizens have for government. Accordingly, it is difficult to argue that a declining level of confidence is a harbinger of a legitimacy crisis. It is likely that the decline may relate to dissatisfaction with the incumbent, or low confidence levels may simply relate to lack of incumbent support at a given point in time. In a similar vein, confidence may well reflect incumbent support as well as satisfaction with incumbent performance.

The analysis found performance and party-based explanations more relevant for explaining confidence in government in Turkey. Economic performance proved significant for other developing democracies as well. One feature differentiating the Turkish case from these countries is the relative weakness of civic attitudes. Hence, it can be argued that in Turkey civic attitudes, such as interest in politics or interpersonal trust, do not provide a democratic attitudinal foundation that strengthens the viability of the government as a democratic institution.

Footnotes

Appendix

Acknowledgements

The doctoral studies of Aylin Aydın and Cerem I. Cenker are supported by the Scientific and Technological Research Council of Turkey (TUBITAK). Both appreciate TUBITAK’s support. The authors also wish to thank Ass. Prof. Işık Özel, Prof. Dr. Sabri Sayarı, Prof. Dr. Ali Çarkoğlu, Prof. Dr. Ersin Kalaycıoğlu, Gözde Yavuz, and three anonymous reviewers for their critical comments and insights.

Notes

References

Anderson

LoTempio

(2002) Winning, losing and political trust in America. British Journal of Political Science 32(2): 335–51.

Bjornskov

(2006) Determinants of generalized trust: A cross-country comparison. Public Choice 130(1): 1–21.

Çarkoğlu

(2007a) A new electoral victory for the ‘pro-Islamists’ or the ‘new center-right’? The Justice and Development Party phenomenon in the July 2007 parliamentary elections in Turkey. South European Society and Politics 12(4): 501–19.

Çarkoğlu

(2007b) The nature of left–right ideological self-placement in the Turkish context. Turkish Studies 8(2): 253–71.

Çarkoğlu

(2008) Ideology or economic pragmatism?: Profiling Turkish voters in 2007. Turkish Studies 9(2): 317–44.

Çarkoğlu

(2009) Turkey’s local elections of 2009: Winners and losers. Insight Turkey 11(2): 1–18.

Chanley

Rudolph

Rahn

(2000) The origins and consequences of public trust in government: A time series analysis. Public Opinion Quarterly 64(3): 239–56.

Çınar

(2006) Turkey’s transformation under the AKP rule. The Muslim World 96: 469–86.

Citrin

Green

(1986) Presidential leadership and the resurgence of trust in government. British Journal of Political Science 16(4): 431–53.

10.

Easton

(1975) A re-assessment of the concept of political support. British Journal of Political Science 5(4): 435–57.

11.

Esmer

(1999) Devrim, Evrim, Statüko: Türkiye’de Sosyal, Siyasal, Ekonomik Değerler. Istanbul: TESEV Yayınları.

12.

Hetherington

(1998) The political relevance of political trust. American Political Science Review 92(4): 791–808.

13.

Inglehart

(1997) Modernization and Post-modernization: Cultural, Economic, and Political Change in 43 Societies. Princeton, NJ: Princeton University Press.

14.

Inglehart

(1999) Post-modernization erodes respect for authority, but increases support for democracy. In: Norris

(ed.) Critical Citizens: Global Support for Democratic Governance. Oxford: Oxford University Press, 236–56.

15.

Inglehart

Welzel

(2005) Modernization, Cultural Change and Democracy: The Human Development Sequence. Cambridge: Cambridge University Press.

16.

Kalaycıoğlu

(2002) State and civil society in Turkey: Democracy, development and protest. In: Sajoo

(ed.) Civil Society in the Muslim World: Contemporary Perspectives. North Haledon, NJ: Islamic Publications, 247–73.

17.

Kalaycıoğlu

(2008) Attitudinal orientation to party organizations in Turkey in the 2000s. Turkish Studies 9(2): 297–316.

18.

Kelleher

Wolak

(2007) Explaining public confidence in the branches of state government. Political Research Quarterly 60(4): 707–21.

19.

Keyman

Heper

(1998) Double-faced state: Political patronage and the consolidation of democracy in Turkey. Middle Eastern Studies 34(4): 259–77.

20.

Listhaug

Wiberg

(1995) Confidence in private and political institutions. In: Klingeman

Fuchs

(eds) Citizens and the State. Oxford: Oxford University Press, 298–323.

21.

Long

Freese

(2006) Regression Models for Categorical Dependent Variables Using Stata. College Station, TX: Stata Press.

22.

Lühiste

(2006) Explaining trust in political institutions: Some illustrations from the Baltic states. Communist and Post-Communist Studies 39: 475–96.

23.

Miller

Listhaug

(1999) Political performance and institutional trust. In: Norris

(ed.) Critical Citizens: Global Support for Democratic Governance. Oxford: Oxford University Press, 204–16.

24.

Mishler

Rose

(1997) Trust, distrust and skepticism: Popular evaluations of civil and political institutions in post-communist societies. Journal of Politics 59(2): 418–51.

25.

Mishler

Rose

(2001) What are the origins of political trust? Testing institutional and cultural theories in post-communist societies. Comparative Political Studies 34(1): 30–59.

26.

Müftüler-Baç

(2005) Turkey’s political reform and the impact of the European Union. South European Society and Politics 10(1): 17–31.

27.

Newton

Norris

(2000) Confidence in public institutions: Faith, culture, or performance. In: Pharr

Putnam

(eds) Disaffected Democracies: What’s Troubling the Trilateral Countries? Princeton, NJ: Princeton University Press, 52–74.

28.

Norris

(1999) Critical Citizens: Global Support for Democratic Governance. Oxford: Oxford University Press.

29.

Öniş

(2003) Domestic politics versus global dynamics: Towards a political economy of the 2000 and 2001 financial crisis of Turkey. Turkish Studies 4(2): 1–30.

30.

Pharr

Putnam

(eds) (2000) Disaffected Democracies: What’s Troubling the Trilateral Countries? Princeton, NJ: Princeton University Press.

31.

Putnam

(1993) Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press.

32.

Rohrschneider

(2002) The democracy deficit and mass support for an EU-wide government. American Journal of Political Science 46(2): 463–75.

33.

Rohrschneider

Schmitt-Beck

(2002) Trust in democratic institutions in Germany: Theory and evidence ten years after unification. German Politics 11(3): 35–58.

34.

Rothstein

(1998) Just Institutions Matter: The Moral and Political Logic of the Universal Welfare State. Cambridge: Cambridge University Press.

35.

Sayarı

(2007) Towards a new Turkish party system? Turkish Studies 8(2): 197–210.

36.

Stolle

Rochon

(1998) Are all associations alike? Member diversity, associational type, and the creation of social capital. American Behavioral Scientist 42(1): 47–65.

37.

Tocci

(2005) Europeanization in Turkey: Trigger or anchor of reform? South European Society and Politics 10(1): 73–83.

38.

Toprak

(2005) Islam and democracy in Turkey. Turkish Studies 6(2): 167–89.

39.

Tunç

(2003) The lost gamble: The 2000 and 2001 Turkish financial crisis in comparative perspective. Turkish Studies 4(2): 31–52.

40.

Uslaner

(2002) Moral Foundations of Trust. New York: Cambridge University Press.

41.

Zmerli

Newton

(2008) Social trust and attitudes toward democracy. Public Opinion Quarterly 72(4): 706–24.