modernization theory is still a powerful tool for explainingdemocratic attitudes

MethodI ran a series of OLS regressions including cross-sectional and separate coun-try estimations. One major drawback of survey research is the high rate ofmissing data because of nonresponses. In the data used, the percentage ofmissing values was 22% for religious service attendance. When all variablesof interest are included in the regression model, the number of cases retainedin the sample drops to 6,409 (only 46% of all respondents) as a result of list-wise deletion.8Political scientists generally either replace the missing values by educatedguesses (e.g., means) or perform listwise deletion, which results in “a loss ofvaluable information at best and severe selection bias at worst” (King, Hon-aker, Joseph, & Scheve, 2001, p. 49). An alternative to listwise deletion ismultiple imputation, which involves imputing m values for a missing cellbased on the existing information (i.e., observed cells). I used Amelia II,developed by Honaker, King, and Blackwell (2007). I ran five imputations ofthe data in Amelia II and later combined the imputed data sets by calculatingthe mean values in Stata 9.

o test these hypotheses, the present study uses the fourth wave of the WVS

Following the footsteps of this research, I examine the individual-leveldeterminants of support for democracy in 10 Muslim-majority countries.

Do Muslims support democracy?

Conclusion

Neither class member-ship nor perceptions of government involvement consistently reaches statis-tical significance. This result is surprising given the well-establishedtheoretical literature about the role of middle and working class attitudes intransition to and consolidation of democracy.

Theeffect of political Islamist views is mixed, as an individual who prefers to seepious individuals in public offices is less supportive of democracy in Turkeyand Bosnia and is more supportive in Pakistan and Indonesia. Turkey, vehe-mently secular, is the only country where this variable reaches statistical sig-nificance in the negative direction for both specific (4.3%) and diffusesupport (13.2%). This may be the result of secularist modernization policiesfollowed in this country. In contrast, the positive and strong effect of political

Finally, the evidence in favor of Islamic values approach is scarce. Religi-osity does not reach statistical significance in 8 of the 10 countries (for asimilar finding, see Hoffman, 2004; Jamal, 2006; Rose, 2002; Tessler, 2002).

The difference between the negative and positive ends of political trustgenerates a 9% increase in the specific support for democracy. Although thisis a large and expected effect, it may be an artifact of measurement. 11 Thesubstantive effects of tolerance and interpersonal trust are modest, and thelatter makes a difference only for specific support. More interestingly, religi-osity has no discernable effect, whereas stronger preferences for religiousindividuals in public offices decrease diffuse support by only 2% and specificsupport by 4%. This result shows that at the attitudinal level, being a piousMuslim does not necessarily have a negative impact on attitudes towarddemocracy (Esposito & Mogahed, 2007; Hassan, 2008; Jamal, 2006; Rose,2002; Tessler, 2002)

Among the indicators of social capital theory, political trust is statisticallysignificant with a positive coefficient, whereas interpersonal trust reachesstatistical significance in the specific support model.

Educationand income have positive and significant effects on both specific and diffusesupport, confirming these hypotheses (Lipset, 1959)

As expected, tolerance andperceptions of gender equality are positively related to support for democ-racy in both models (Inglehart, 1990; Inglehart & Norris, 2003)

Independent Variables

The first three items are about the outcomes of democratic governance,whereas the last two ask the respondents to provide a general assessment ofdemocracy. Factor analysis was conducted with these five items, and twofactors were extracted. The first three items tapping respondents’ opinionsabout specific outcomes of democratic governance load strongly on a firstfactor (factor loadings are .80, .77, and .77, respectively, with a Cronbach’salpha of .75), and the last two items tapping respondent attitudes regardingoverall support for democracy load on a second factor (factor loadings are.77 and .62, with a Cronbach’s alpha of .52). This distinction resemblesEaston’s (1965) typology of diffuse and specific support. 7 I prefer to useEaston’s concepts below; however, I approach the results of the factor analy-sis cautiously as the use of these concepts does not mean that I use specificand diffuse support in a pure Eastonian sense

ecently, Jamal and Tessler (2008) used evidence from the ArabBarometer Survey to demonstrate that Arab citizens understand democracy indifferent terms.

The second reason for the problematic nature of the teleological interpre-tation of modernization theory concerns its cultural implications. Although,modernization is expected to foster mass democratic values and beliefs as aresult of industrialization, urbanization, and increased wealth and education,it is not clear how these values and beliefs are brought about or whether theyare causes or consequences of democratization. Furthermore, Lipset (1994,p. 3), the strongest proponent of modernization theory, declared that culturalfactors are more important than modernization for the prospect of democracyand that certain faiths, such as Islam, Christian orthodoxy, and Confucian-ism, are fundamentally incompatible with democratic values. Inglehart(1990) argued that economic development leads to cultural changes that inturn lead to democracy. More recently, Inglehart and Norris (2003) arguedthat the real fault line between Islam and the West is not political. Rather,they claimed that what divides the two cultures are values and attitudesrelated to the perceptions of gender equality and tolerance. As individualsmove to urban areas, get better education, and increase their wealth, they arealso expected to become more secular, more accepting of women’s roles insociety, and socially more tolerant in general (Jamal, 2006; Moghadam,2003). In their recent analyses, Hoffman (2004) and Jamal (2006) found evi-dence supporting this argument. 1 According to a cultural interpretation ofmodernization theory, it can be expected that individuals with positive per-ceptions of gender equality and those who are more tolerant will be moresupportive of democracy compared to those lacking these opinions

Furthermore, a mutually beneficial relationship exists between states andcertain segments of societies in the Middle Eastern countries. The state poli-cies especially targeted middle classes (using state-owned enterprises;Richards & Waterbury, 2008) and organized labor (by controlling the tradeunions; Posusney, 1997) to maintain this relationship. Although the develop-ment of a strong private sector independent of the state may have eventuallybegun to undermine the power of the authoritarian state and created a citi-zenry equipped with democratic values, rentierism (Richards & Waterbury,2008) or the “resource curse” (Ross, 2001) provided the state with someautonomy from the public by removing the government reliance on taxes inthe oil-rich Middle Eastern countries and the other parts of the Islamic world(i.e., workers’ remittances and tourism revenues in Morocco and Turkey)

It is argued that members of the middle and workingclass will challenge authoritarian practices if and only if they experience con-siderable independence from the state (Kamrava, 2005). To elaborate, if anauthoritarian state controls a substantial portion of its citizens’ behavior byproviding jobs and benefits (e.g., tax-free income), the modernization pro-cess may not necessarily create a democratically oriented citizenry (Bellin,2002; Kamrava, 2005). This is particularly valid for the authoritarian regimesof the Middle East, where clientelism contaminates the political life andserves to legitimize authoritarian practices (Lust-Okar, 2008). The existenceof a strong government in political and economic life may depress the emer-gence of “critical citizens” (Norris, 1999), particularly those among the mid-dle class. This tendency, combined with the clientelistic nature of the politicalsystem (especially in certain parts of the Muslim world), may strengthenauthoritarian values and prevent the emergence of a democratic citizenry

The relationship of development and democracy, however, has proven tobe more subtle than this linear economic scheme. “Multiple modernities”may emerge as a result of interaction between modernizing forces and theexisting cultural and institutional structures in a country (Eisenstadt, 2000,pp. 13-15).

AlthoughHoffman (2004) used explanations related to religious orientation and politi-cal and civic culture, Tessler (2002) incorporated religiosity, Islamic politicalvalues, and gender in his analysis. Jamal (2006) and Bratton (2003), on theother hand, modeled individual support for democracy as a function of mod-ernization and religious values

These studies,however, have focused only on select regions such as Arab countries (Jamal,2006; Tessler, 2002), Turkic nations of central Asia (Rose, 2002), Africannations (Bratton, 2003), and non-Arabic Muslim countries (Hoffman, 2004)

So far, few students of comparative politics have investigated the determi-nants of individual support for democracy in Muslim countries (Bratton,2003; Hoffman, 2004; Jamal, 2006; Rose, 2002; Tessler, 2002).

Fukuyama, 1992; Huntington, 1999; Kedourie, 1992)

What are the individual-level determinants of sup-port for democracy in the Muslim world?

These countries represent different regime types including elec-toral democracies, authoritarian republics, and monarchies. Table 1 presentsthe rankings of these countries according to the United Nations HumanDevelopment Index (HDI; 2001), their Freedom House status (2001) andGDP per capita (2001), and their percentage support for democracy calcu-lated from the WVS.