Democratic commitment in the Middle East: a conjoint analysis

Abstract

Polls from the Middle East/North Africa show high support for democracy. However, the veracity of this support has been called into question. This study uses a conjoint analysis to show that citizens support democratic institutions, as well as favoring an effective welfare state and a state religion. The results demonstrate that support for elected governance is not contingent on the state's providing economic benefits; citizens are more likely to favor participatory government at each level of economic outcome. Interest in incorporating religion in the state, however, is contingent on the political and economic profile described; the contingent effects suggest interest in Islamic governance is, at least partly, instrumental. Although pro-democracy public opinion alone does not secure democratization, it creates fertile ground for future democratization movements.

1. Introduction

At a time when democracy and public interest in democracy appear to be in retreat (Foa and Mounk, 2017), democracy (dimuqratiyya) polls very well in the Middle East/North Africa (MENA) (Robbins, 2015; Al-Ississ and Diwan, 2016). In the 2018 Arab Barometer, 72 percent of respondents said that “democratic” systems are the best form of government. This support has been challenged on two grounds. Firstly, MENA citizens could be overstating their support (Kuran, 1997; Maseland and van Hoorn, 2011; Claassen, 2020b). In this case, they could support democracy to a lesser extent or secretly disapprove of it.

Secondly, they could be imagining a different regime than political scientists do when they endorse “democracy”—a regime that is not democracy at all (Bratton, 2003; Tessler et al., 2012; Ciftci, 2013; Alvarez and Welzel, 2017). The surveys used to identify these preferences, such as the Arab Barometer and the World Values Survey, ask respondents about dimuqratiyya. Surveys suggest that approximately half of MENA residents conceive of dimuqratiyya primarily as a series of economic outcomes rather than political institutions (Arab Barometer V, 2018). Furthermore, citizens who conceive of dimuqratiyya as a favorable economy or social stability, rather than a system of government defined by popular political participation, are less likely to endorse an elections-based system of government (Ridge, 2023). Survey results could then be masking citizens' regime-type preferences. This study uses a conjoint design to clarify MENA citizens' regime-type preferences.

Through original surveys in Egypt and Morocco (August 2019 and January 2020), this study examines citizens' preferences. By incorporating state features related to elections, political participation, religion and religious leaders, the welfare state, and the economy, the conjoint design contains multiple features that could drive popular support. By incorporating diverse elements and presenting them in conjunction with each other, rather than asking about direct structural preferences over elements individually, the conjoint structure identifies preferences and their relative weights. All components may be valued, but the tradeoffs induced by the forced-choice conjoint design can demonstrate which are valued more.

Thus, the analysis addresses the veracity of the public's apparent interest in elected governments through these conjoint task preferences. The survey finds similar preference profiles in both country cases, despite these countries having very different regimes. Egypt has been a militaristic semi-republic since the 2013 coup against the elected Islamist government. Morocco is a constitutional monarchy that, in 2019, had an Islamist government. Nonetheless, in both surveys, the citizens demonstrate preferences for democracy in the form of elections and political participation. This support for elections and popular participation, however, is only one factor for citizens' choices. They also desire a strong economy and a functioning welfare state. Additionally, respondents favor maintaining an official religion while excluding religious leaders from the government. Ultimately, these results are consistent with support for democratic governance. Furthermore, respondents who endorsed elected government in a direct question, compared to those who did not, give more choice weight to opportunities for political participation. This works against the suspicion that survey respondents are concealing anti-democratic sentiment.

The conjoint also investigates instrumental support, as opposed to intrinsic support, by considering changing interest in democracy based on economic outcomes. Conditioning on the described levels of economic deliverables, citizens' preferences for democratic institutions and participation do not change. Thus, support for democracy in the Middle East is more principled than has been feared. In turn, conditioning the responses on the presence of elections shows minimal differences. Citizens prefer a government in which they can participate, but they do not prefer it to the exclusion of economic considerations. While the connection between public opinion and democratization is contested (Claassen, 2020a, 2020b; Tai et al., 2022), robust public support creates more fertile ground for democratization in the future.

2. Desiring democracy?

The world is seemingly populated by democrats. That includes the Middle East, which is governed by authoritarians. Surveys in the Middle East have long revealed strong public support for democracy in response to direct questions, although lower than Western Europe (Claassen, 2020a). Citizens identify it as the best form of government and even say they think it is appropriate for their countries. This support pre-dates the Arab Uprisings (Jamal, 2006; Tessler et al., 2012) and endured through them and their (un)successful regime transformations (Robbins, 2015; Al-Ississ and Diwan, 2016).¹

The veracity of this support—or the measurement of it—has been questioned. Firstly, these surveys have a “d-word” problem (Bratton 2003). Scholars of Middle East public opinion have speculated that, in answering questions about democracy, the public holds a different mental conception of the government than researchers have. Survey respondents may imagine a regime that incorporates a role for Islam or Islamic values in the government (Ciftci, 2013). Although democracy need not be secular, scholars are skeptical of incorporating Islam, especially those who blame Islamism for the dearth of democracy in the region (Blaydes and Lo, 2012; Gouda and Hanafy, 2022). Ciftci (2021) proposes that Islamic conceptions of justice (‘adl/adelet) that focus on political justice may breed democratic interest—if that justice is grounded in a theory of free will; if it is associated with a belief in predestination, then an interest in justice would lead to political quietism. The branch of Islamic justice grounded in the social dimension would favor order, security, and the economy, which would predispose toward benevolent authoritarianism. In this schema, it is evident that the role of core constructs like religion and justice in democratic attitudes is a function of how they are construed at the individual level.

Survey-takers may construe questions about dimuqratiyya as questions about the economic policies the state will enact. For instance, they may conceive of a government that has limited corruption while ensuring employment and welfare, which a democracy may or may not deliver (Ridge, 2023).² Apparent support for democracy could reflect instrumental valuation; they value it because they think it will produce good economic outcomes (Bratton and Mattes, 2001), or it could be misspecification, meaning they value it because they think it is the favorable outcome. MENA residents favor redistributive economic policies, and this is separate from their attitudes toward elected government (Ridge, 2022a, 2022b). The most counter-intuitive option is that MENA citizens may answer questions about democracy while conceiving of non-democratic governance structures. They may think of a military takeover or theocracy when formulating their responses (De Regt, 2013; Alvarez and Welzel, 2017; Kirsch and Welzel, 2019). Following this literature, direct questions about dimuqratiyya cannot access citizens’ attitudes about democracy. They may measure the opposite.

Assertions of over-estimated support appeal to preference falsification (Kuran, 1997). Claimants speculate that a social value placed on democracy—or that the respondents think researchers place on democracy—leads respondents to pay “lip service” to democracy without actually valuing it. Alternatively, citizens in an authoritarian society could under-report their interest in alternative regime structures. These would lead to countervailing pressures for mis-specification. Benstead (2018, 536), however, demonstrates that this risk is exaggerated: “worries that the Arab world is a more challenging survey context—or that citizens answer dishonestly—reflect biases of ‘Arab exceptionalism,’ more than fair assessments of data quality.”

Others postulate that the support is an illusion. Anything rare is more valuable. Thus, the citizens are endorsing democracy because they live in non-democracies, rather than because they particularly support it. It is a function of marginal preferences (Maseland and van Hoorn, 2011). If democracy were to increase, support would decrease. Claassen (2020b) argues the depressive effect of democratization is driven by expanding protections for minority rights—a liberal values—rather than majoritarian institutions—the democratic institutions themselves; Ridge (2022a, 2022b) shows that support for liberal values is independent of democratic support in Egypt and Tunisia. According to this school of thought, public interest in democracy is ephemeral and artificial. Although respondents are not actively dissimulating, they are not committed to democracy.

To tap into the multiple potential meanings respondents could be considering in these questions about dimuqratiyya and their democratic preferences, this study employs a conjoint design to examine respondents' preferences.³ The states are described in terms of the opportunities for political participation. If the citizens value democracy, then the demonstrated preferences should favor multiparty or single party elections over no elections (hypothesis 1a). If they actually mean to endorse an authoritarian regime of any stripe (e.g., strong man, military) or if they secretly are opposed to democracy and were effacing that in a direct question, the supposed “support” in the prior surveys would indicate anti-democratic preferences. The demonstrated preferences would show a preference for holding no elections (hypothesis 1b). The same would be expected for their opportunities for political participation; democrats would be less supportive of a system that imposed barriers to political participation (hypothesis 2a). Secretly indifferent respondents or secret autocrats would expose null or negative effects (hypothesis 2b).

The state can also be described in its religious caliber. In direct questions, Arab populations are divided on whether religious leaders and religion should influence politics and the state should regulate Islam (Arab Barometer V, 2018).⁴ If the respondents are imagining “democratic models incorporating Islamic values” as part of the reportedly high levels of support for democracy rather than “strict secular democratic arrangements,” there should be support for incorporating religion into the state (hypothesis 3a). This could be achieved without undercutting their institutional supports. Respondents may, however, prefer to keep religion out of government (hypothesis 3b). The conjoint can evaluate secularist preferences and democratic preferences net of each other and assess their relative weight in the respondents' choices.

Given the role of economic outcomes in some construals of dimuqratiyya and concerns about instrumental interest in democratization, it is necessary to include economic details (Abbott and Teti, 2022; Ridge, 2023). It takes no stretch of the imagination to believe citizens would prefer low unemployment (hypothesis 4a) and a stronger welfare state (hypothesis 4b). Redistributive programs are popular in the Middle East (Ridge, 2022a, 2022b). MENA residents face high rates of poverty and unemployment (Hong, 2019), and regimes regularly use subsidies to ameliorate these conditions and pacify the public (Salevurakis and Abdel-Haleim, 2008; Verme et al., 2014). Threats to those programs can cause protests or riots (Beinin, 2016). Destabilizing the regime can interrupt resource transfers that are immediately beneficial. Potential gains from democratization, which are speculative, have reduced relative utility in practice. After the Arab Spring, publics that were disappointed in how regime change played out (e.g., social turmoil and economic challenges) evinced less faith in dimuqratiyya in subsequent surveys, and democratic interest took years to recover (Kilavuz and Sumaktoyo, 2020; Spierings, 2020). It is expected that, in framing their preferred state, they favor the successful economic condition.

To assess instrumental support, the level of support for democracy across the different specifications of economic deliverables is considered. The voter may value elections or participation differently depending on the economic circumstances in the state. Intrinsic support would still be present in the choice task regardless of the economic circumstances (hypothesis 5a). Instrumentalists' choices will be less informed by the elected institutions than the economic outcomes (hypothesis 5b). This would be consistent with Spierings's (2020, 523) treating “democracy as instrumental to other goals,” at least in part. The conjoint technique can identify such dependent interest.

By and large, citizens live lives in the state structure they already have. As such, a choice task exercise about state structure might seem artificial. In a global sense, public opinion alone may not drive authoritarians to democratize (Claassen, 2020a; Tai et al., 2022). It would make democratic institutions, once instantiated, more durable. However, the MENA region is different in this regard. Middle East/North African countries have recently and recurrently faced the question of what they would want a new government to look like during the Arab Uprisings, civil wars, coups, and demonstrations. These populations are familiar with debating what a state should be like. As such, questions about what is most valuable or necessary in a government are far less hypothetical in the MENA.

3. Materials and methods

For this study, original surveys were conducted in Egypt and Morocco through YouGov's MENA panel.⁵ The Egyptian survey was fielded in August 2019; it includes 1000 Egyptians. The Morocco survey was fielded in January 2020; it includes 991 Moroccans. This panel has been used in other political science studies (Nyhan and Zeitzoff, 2018; Blackman and Jackson, 2021; Ridge, 2022b).⁶ Respondents were able to take the survey in Arabic or English (Appendix A). In total, 94.6 percent of Egyptians and 93 percent of Moroccans took the survey in Arabic. The panel targets age and gender representativeness, but because online surveys require literacy, the sample is slightly more educated than the general population (Arab Barometer V, 2018).⁷

These countries are quite different in history and structure. Egypt, the largest Arab country, with over 100 million inhabitants, played a substantial part in the Arab Uprisings. It removed long-term president Hosni Mubarak and instituted an elected government. That government was ultimately removed by a military coup that some citizens referred to as dimuqratiyya for enacting the public will and removing Islamists. The military is composed of “citizen-soldiers” and had largely focused on external threats, rather than the citizenry, which left it popular good will, and it offered “a symbol of stability and continuity at times of uncertainty” (Hassan et al., 2018, 281). In many ways, Egypt has reverted to the pre-revolutionary circumstances (Stacher, 2020). Although Islamist groups, especially the Muslim Brotherhood, are highly regulated in Egypt, the Egyptian constitution treats Islam as the state religion and principal source of legislation. It allows Christian and Jewish law to be the main source for their personal status laws and religious affairs.

Morocco is a parliamentary monarchy, and the king plays an active role in its politics.⁸ The parliament is elected, but key ministries are considered loyal to the king (King, 2019). After the 20th February Movement—the Moroccan arm of the Arab Uprisings—political reforms were instituted that reduced some of the king's powers and a new constitution was passed. While some predicted the reforms laid the groundwork for the monarchy's removal, this has yet to transpire (King, 2011). Protests have recurred in the face of unresolved economic issues. The monarchy's “depoliticization” of the reform process and Moroccan politics in general have allowed it to diffuse protests and offset the potential for true structural change (Maghraoui, 2002; Hibou, 2011). The 2021 general elections even replaced an Islamist government with parties more favored by the king. Morocco's official religion is Malaki Islam.

Conjoint analysis has proliferated in political science. Researchers use it to study voters' candidate preferences (Hainmueller et al., 2014b; Franchino and Zucchini 2015; Carnes and Lupu, 2016; Shockley and Gengler, 2020; Badas and Simas 2022), party leaders' candidate preferences (Doherty et al., 2019), partisan discrimination (Theodoridis et al., 2022), disaster response preferences (Bechtel and Mannino, 2022), issue importance in policy bundles (Horiuchi et al., 2018), Brexit strategies (Hobolt et al., 2020), immigration preferences (Hainmueller et al., 2014a, 2014b), and bureaucrats' constituent preferences (Adam et al., 2021). Graham and Svolik (2020) look at Americans' democratic commitment by comparing the effect of candidates' ideologies and their anti-democratic policy preferences on voters' choices. While candidate choice experiments are common applications of conjoint, they presume the occurrence of free and fair elections; it would also not admit describing an absence of elections or poor economic conditions, for which candidates do not campaign. Assuming elections also removes an important dimension of analysis.

The technique is increasingly used in MENA. Shockley and Gengler (2020) examine candidate preferences in Qatar to identify the influence of co-sectarianism in vote choice. Blaydes et al. (2021) studied Qatari attitudes toward women working. Cammett et al. (2022b) consider security policy preferences in Lebanon, and Cammett et al. (2022a) address the role of ethnicity in attitudes toward clientelism in Lebanon. Shamir and Shamir (1995) applied a conjoint to a study of democratic commitment in Israel. This project contributes to the application of this survey technique in the Middle East.

Direct questions about individuals' preferences or rank orderings with regard to features of a regime artificially separate these elements. The conjoint approach more readily “simulates real-life decision tasks, when people have to trade off features or attributes, one against another” (Shamir and Shamir, 1995, 109).⁹ The conjoint design also helps reduce intentional bias, such as social desirability bias, and cheap talk (Hainmueller et al., 2014b; Theodoridis et al., 2022). It is easy for a respondent to say that every feature is valuable. Ranking requires a hierarchy, but overt rankings “are probably more affected by social desirability bias” than conjoint choices (Shamir and Shamir, 1995, 121). By forcing the respondent to make a choice between systems with different elements, the conjoint ensures that he cannot claim that they are all (equally) valuable. At the same time, it is “agnostic about how respondents reach their observed decisions” (Hainmueller et al., 2014b, 3). Not only does the test not make assumptions about how the decision is reached, it does not require respondents to acknowledge the reason they prefer one government system over another or to be consciously aware of the precise weight that they put on regime features. The respondent chooses bundles and can deceive himself of the reason or have plausible deniability with regard to choices he might fear were unpopular, just as real-life voters choosing among platforms can.

Two profiles showing the six attributes and their randomized levels were presented to the respondent (Figure 1).¹⁰ Each attribute had two or three levels for 216 potential profiles.¹¹ The two profiles were never identical, though they could share features. For instance, both presented regimes could have an official religion. Hainmueller et al. (2014a) find that paired conjoints, presenting two options side-by-side, are closest to behavioral benchmarks from real-life choices. Respondents were presented with five profile pairs sequentially on separate screens.¹²

Figure 1. Conjoint choice task screen. NB: The tables are randomly generated. The English-language text is not a translation of the Arabic text.

To identify democratic support, this study draws on a minimalist conception of democracy by focusing on elections and opportunities for political participation (Dahl, 1971; Przeworski et al., 2000). Although political science does not have a singular definition of democracy, the “most popular definition of democracy equates it with regular elections, fairly conducted and honestly counted” (Schmitter and Karl, 1991, 78). For elections, the options were no elections, elections with only one recognized party, and elections with multiple recognized parties. Party structures are not strong in the Middle East. Many Egyptian parliamentarians are independent; independent parliamentarians are less common in Morocco. Sometimes parties are restricted and their supporters run as independents; others are just unaffiliated. Political participation is described in terms of the barriers the state introduces to participation: many barriers, some barriers, or few barriers.¹³

Religion's position is included in two forms. One is whether or not the state has an official religion. The other is whether or not there is an official role for religious leaders in government. In terms of economic outcomes that the government might generate, descriptions of the welfare subsidy system and the level of unemployment were included. The levels of high and low unemployment—specified at 14 and 7 percent respectively—are realistic for the Middle East. Prior to the Arab Uprisings, unemployment in North Africa was slightly above 10 percent, heavily concentrated in the youth (Hong, 2019). The rates increased in the immediate aftermath of the uprisings—a noted potential threat to the durability of the democratization efforts—but they returned to the prior levels. The unemployment rate in 2019 was 10.76 percent in Egypt and 9.02 percent in Morocco. These rates are thus high and low relative to the local conditions. The state could be said to offer good provision of necessities, some provision, or little provision. Based on these features, citizens chose which of the randomly generated state institutional structures they would prefer.

4. Results

In a conjoint analysis, the causal quantity of interest is the average marginal component effect (AMCE).¹⁴ It is the “marginal effect of attribute l averaged over the joint distribution of the remaining attributes” (Hainmueller et al., 2014b, 10). It averages over both direction and intensity of preference (Abramson et al., 2022).¹⁵ Functionally it is the change in the likelihood a profile is chosen if that characteristic is included in the profile instead of the baseline characteristic. Because of this averaging, the results are described in terms of “average” Egyptians and Moroccans. The results for Egypt are shown in Figure 2 and for Morocco in Figure 3. Despite the many differences between these countries, the citizens' preferences are not substantially different. The only feature for which the country estimates are significantly different (p < 0.05) is the response to having no elections. The effect is stronger in Egypt.

Figure 2. Average marginal component effect (Egypt).

Figure 3. Average marginal component effect (Morocco).

Elections are highly influential in the respondents' regime-type choices. The average Egyptian is more likely to favor a regime that has elections and still more likely to favor multi-party elections. The negative effect of not having elections is significantly stronger than that of having elections in which only one party is recognized. Similarly, the average Moroccan is less likely to choose a system that has no elections or elections with only one recognized party than multi-party elections. The negative effect of not having elections is stronger than that of having elections in which only one party is recognized. The AMCEs are smaller for Moroccans than for Egyptians. This is consistent with hypothesis 1a. It is not consistent with hypothesis 1b. There is a demonstrated preference in favor of democratic institutions.

Barriers to political participation also inform citizens' choices. In the Egypt sample, the introduction of barriers to political participation reduces the likelihood of support relative to having few barriers to participation, but having many barriers does not matter more than having only some. In the Morocco sample, systems with many barriers to citizens' political participation were significantly less likely to be chosen than those with few barriers. Having some barriers, however, is not significantly different in choice propensity than having few barriers. A preference for lower barriers to political participation is consistent with hypothesis 2a and not consistent with hypothesis 2b. There is a preference in favor of citizen engagement. This is consistent with democratic ideals.

The involvement of religion in the state informs citizens' preferred state; however, the effect depends on the form that relationship takes. On one hand, having no official religion reduces both the average Egyptian and average Moroccan's likelihood of choosing that state structure compared to having an official religion. On the other hand, having no role for religious leaders in the government increases the average Egyptian and average Moroccan's favorability toward that system relative to having a role for religious leaders. In both samples, the AMCE spread is larger for official religion than for participation of religious leaders. This indicates that having an official role for religion is, on average, more influential on the survey takers' responses and/or that the public is more unified in this regard. This distinction between official religion and involving religious leaders speaks to a nuanced view of the relationship Islam should have to the state. Citizens can desire formal recognition for religion without supporting religious leaders' direct involvement.

One view of this is that the citizens want a patina of religion but that they mean its involvement to be superficial, which is why they do not want a role for religious leaders. Another way to construe it could be that they really do want a position for religion in the state but that they are skeptical of involving religious leaders. These men could exercise independent power in the state, and they might push a conservative agenda. They could be corrupt, rather than true representatives of religion and religious law. One college-educated Cairene Muslim man's comment reflects this principle: “Religion is the basis of the country, but there are people who govern the country in the name of religion and offend the religion to which they belong.”¹⁶ The citizens can value religion for their country while being circumspect about how it should be included or represented. Citizens' making distinctions on these dimensions shows a considered response to the potential involvement of religion in government, not knee-jerk Islamism. It also shows that these perspectives can co-exist with support for elected government.¹⁷ Hypotheses 3a and 3b are not fully supported due to the split. Future research could evaluate the many ways in which religious law and leaders inform MENA governance to identify more nuances in public support.

Economic issues are also important in citizens' preferences in both countries. For both Egyptians and Moroccans, a system's producing lower rates of unemployment makes it more likely to be chosen relative to one that has higher levels of unemployment. A potential government is less likely to be favored if it has inferior welfare systems. The average Egyptian is less interested in a regime that has only some necessities provision than in good government provision and prefers some state provision to little state support. The average Moroccan is also less likely to choose a government with limited provision of personal necessities. While having some provision is not significantly different from having good provision, having little provision significantly undermines support. The AMCEs are smaller, though, for goods provision than for unemployment. The unemployment rate then is more important for the average Egyptian and Moroccan's choice than service provision. This is consistent with H4a and H4b.

To distinguish intrinsic and instrumental support for democracy, citizens' preferences net of a particular level of unemployment are considered. The AMCE for each trait among respondents who saw a state with high unemployment and who saw a state with low unemployment, in each country, are shown in Appendix E. In neither case does support for democracy—either in terms of popular participation or electoral institutions—change significantly based on the economic condition. The same holds for conditioning the analysis by welfare state efficacy described (Appendix F). While Egyptians' preferences toward having an official religion changed with the level of service provision—having an official religion matters less when the provision of private necessities is good—their preferences for elections and participation did not; Moroccans were not sensitive to it either. These results support hypothesis 5a, not hypothesis 5b.

The conjoint results are consistent with citizens' valuing elections-based government. In fact, they place a relatively high value on it. Multiparty elections compare favorably to unemployment considerations and religious appeals. Elected government, however, is not the breakaway favorite. Other factors, especially unemployment, have nearly as great an influence. Thus, the expressed preferences seem honest—in fact they appear to be principled, rather than strictly instrumental. They hold distinct from the level of economic deliverables. They are not, however, the singular drivers of regime preferences.

Inversely, the level of elections can be distinguished (Appendix G). Within the Egyptian sample, the AMCEs are not heavily dependent on the electoral opportunities that the state structure includes. Only the official role for religious leaders is significantly different across levels. While the average Egyptian considering a state with elections would prefer a government with no official role for religious leaders, when evaluating a state without elections, this attribute is not significant. For the average Moroccan, none of the attributes is valued significantly differently when paired—or not paired—with a democratic government structure. They support democracy, but it is not an overriding demand. What matters in democracy matters in non-democracy too. This is consistent with both real interest in democracy and real interest in economic development.

That it is the religion-associated features of the state that are shifting in Egypt based on these economic and political elements recalls Berger (2019) and Fair et al.'s (2018) discussions of the meaning of shari'a. If citizens' interest in incorporating religion in the regime is instrumentalist—such as viewing it as a motivator for redistribution or a constraint on authoritarian will—the religious character of the state would matter less if the services are assured or the public will has avenues for direct expression. The secular or religious bent of MENA democracy may have highly contingent support.

It is also possible to compare these results to direct questions about democratic interest. The survey did not ask about support for dimuqratiyya, but it did ask whether choosing the government by election is the best form of government, whether unelected governments are sometimes better, or if it does not matter—a common measure of democratic support modified to avoid the word dimuqratiyya (Appendix H). Egyptians who indicated that democracy is best were more likely to support multiparty elections and to favor few barriers to participation, while being marginally less tolerant of having no elections and many barriers (p < 0.055, p < 0.077); they were not significantly different on other dimensions, though they were marginally more tolerant of high unemployment (p < 0.063). Moroccan respondents who thought democracy was best were more interested in having multiparty elections and few barriers to political participation and were less accepting of regimes with many barriers; they were more tolerant of weak services provision. These results are consistent with the premise that questions about democracy can tap into public support for democracy. However, those questions do not necessarily expose attitudes toward other state features, like (non)secularism.

5. Discussion and conclusions

Conjoint analysis can demonstrate what citizens most value. In these cases, the preferred condition for the average Egyptian and Moroccan has multiparty elections, few barriers to political participation, an official religion but no role for religious leaders, low unemployment, and good baseline public welfare. Of the 216, the choice that was most likely to be chosen by Egyptians and Moroccans—the case that was selected 86.1 percent of the time it was shown in Egypt and 85.2 percent of the times it was shown in Morocco—had multiparty elections, some barriers to political participation, an official religion but no role for religious leaders in the government, some provision of individual necessities, and low unemployment. In Egypt, a regime similar state that had good necessities provision was equally likely to be chosen. These preferences align despite the differences in the current Moroccan and Egyptian states.

That any of the regime options achieves such strong support is noteworthy. In fact, just over half of the potential regimes achieved at least 50 percent support. It is possible that no combination of features in a conjoint study receives majority support as a package, even the case that featured the elements that were positively rated on average (Hobolt et al., 2020). Where the combinations fail, it points to infeasibility in practical politics. In both the Egyptian and Moroccan cases, though, regimes could be constructed with respect to these features that could obtain majority—even supermajority—support. That so many options would receive majority support, however, also demonstrates that many paths forward are available and workable in these countries, not just democratic ones. Getting to some of these positions is a matter of choosing those institutions. Achieving economic outcomes would be harder to actualize.

By contrast, the regime that was least likely to be chosen by Egyptians—taken only 16.7 percent of the times it was shown—had no elections, some barriers to political participation, no official religion or role for religious leaders in government, little provision of necessities, and high unemployment. The least likely to be chosen by Moroccans, chosen 13.2 percent of the times it was shown, had only one recognized political party, many barriers to political participation, no role for religious leaders or official religion, high unemployment, and some public provision of necessities.¹⁸ Overall, the pattern is not particularly surprising. Economic success in a participatory regime that recognizes the predominant religion without empowering potentially restrictive clerics would seem straightforward. What is more interesting is what the analysis can demonstrate in terms of potential trade-offs.

Despite the differences between these country cases, the results are quite similar. This suggests that the patterns could generalize to other countries in the region. For instance, as the democratic support in the conjoint results befits the avowed support, scholars should trust that popular sentiment is rather accurately reported in the other countries as well. Future studies could evaluate the extent of replication in other states. They could also introduce additional elements to the conjoint (e.g., security dimensions, liberal values) to provide points of greater popular comparison.

Overall, the most important feature for Egyptians and Moroccans in choosing a government is the nature of the elections. Much more weight is placed on the existence of elections than the relative barriers the government could place in front of citizens' methods of participating in politics. Thus, there is no reason to conclude that the expressed preference for democracy that has been identified in other regional studies is disingenuous. Citizens are demonstrating a preference for the opportunity to elect their governments when making their choices, even when they could conceal their reasoning from the investigators or themselves. Thus, the level of expressed support in direct survey questions has grounding in real preferences. These results push back on assumptions that Arab respondents are lying about their democratic preferences.

The next most influential items are the unemployment rate and the presence of an official religion. They would rather a system generate job opportunities than provide welfare. This relative valuation of work and welfare is in contrast to current regimes that provide subsidies in multiple domains without resolving the substantial and long-standing unemployment problem. These policies are useful, but they are not preferable.

The results also show a nuanced view of secularism. Rather than linking an official religion with a regime that incorporates religious leaders, these elements are distinguished. An official religion is valued. Many MENA states currently recognize Islam as an official religion. These results suggest that that at least is not counter to public sentiment. At the same time, an official role for religious leaders is disfavored, though this element has a smaller average marginal effect. Such a system would not be out of step with other democracies. While several Western democracies, such as the USA, renounce national religions, others, like the UK, maintain established churches. Most importantly, the support for an official religion coexists with support for a system of elections. This is consistent with the argument that Muslims may seek a non-secular democracy. This does not mean though that they seek a democracy unlike those present elsewhere in the world. Strident opposition to religion-affiliated governance could still represent an opposition to a potential democratic state structure. That interest in affiliating religion with the state is dependent on the other state features is an indication that at least some of this interest is instrumental. Instrumental considerations would also contribute to the similarity in preferences across two countries with distinct religio-political profiles. Stridently secular organizations, then, could indirectly further their interests by encouraging factors like economic provision to address those instrumentalists' concerns in another way.

Crucially, the support expressed for democracy appears to apply to those elements in their own right. It does not change with the level of economic deliverables. For instrumentalists, those who only value democracy for what it could offer, economic outcomes would supersede participation. That pattern is not demonstrated with respect to either state unemployment or welfare provision. These features are not obviating the interest in democratic participation. As such, the support for electoral democracy appears principled.

While interested in democracy, the respondents are not only interested in democracy. Even accounting for the inclusion and exclusion of elections, economic features continue to inform citizens' choices. Both elements, then, can be important in shaping regional political preferences. Recall Ciftci's (2021) findings with respect to political and economic justice. A focus on economic justice could dispose respondents to authoritarians, and political justice was not an automatic spur to political protest and participation. Supporting democracy in the region may then entail supporting both institutional and economic security to dispel the allure of provisioning and stabilizing non-democracy.

The ability to conduct such analyses is an improvement over the direct questions about specific institutions. This is not to say that the direct questions are uninformative. For instance, citizens' statements about elected government aligned with their preferences in the conjoint task. Direct questions about democracy or any other institutional structure though would only inform on that dimension. Attitudes toward democracy are just that. They do not reveal information about citizens' attitudes toward these other state features. As such, the direct questions provide only a partial view of MENA residents' politics. Furthermore, the use of the multiply construed term dimuqratiyya in most surveys has obscured what is exactly being endorsed (Ridge, 2023). Generating a fuller-picture of citizens' preferences requires addressing these multiple domains—as well as others that researchers may consider important.

A country populated by democrats may not rush toward democratization on that basis alone (Claassen, 2020a). However, political attitudes can become politically relevant quickly (Kuran, 1997). MENA populations have taken to the streets before, and they may again be faced with choices about the forms that their states and futures should take. These will be opportunities to weigh economic security, regime deliverables, and what role Islamists groups should be allowed to play just as much as they are opportunities to knock down barriers to participation. Latent public support for electoral democracy, in such moments, can be turned to democratization.