Islamism, Secularism and the Woman Question in the Aftermath of the Arab Spring: Evidence from the Arab Barometer

The uprisings that led to regime change during the early period of the Arab Spring were initially inclusive and pluralistic in nature, with men and women from every political and religious orientation engaging actively in political activities on the street and in virtual spaces. While there was an opening of political space for women and the inclusion of demands of marginalized groups in the activists’ agenda, the struggle to reimagine national identities that balance Islamic roots and secular yearnings is still ongoing in many countries in the region. This paper seeks to deepen understanding of the extent to which the pluralistic sentiments and openness to accepting the rights women have persisted following the uprising. We aim to examine changes in attitudes towards women’s equality in countries that underwent regime change through popular uprisings during revolutionary upheavals of the Arab Spring and in countries where regimes have remained unchanged. Using available data from consecutive rounds of the Arab Barometer survey, we examine changes in attitudes in nine countries with two rounds of Arab Barometer during and post Arab Spring (Egypt, Yemen, Tunisia, Algeria, Lebanon, Sudan, Jordan, Iraq, Palestine).We find that support for “Muslim feminism” (an interpretation of gender equality grounded in Islam) has increased over the period and particularly in Arab Spring countries, while support for “secular feminism” has declined. In most countries examined, relatively high degrees of support for gender equality co-exist with a preference for Islamic interpretations of personal status codes1 pertaining to women. We discuss the implications of these findings for academics and activists concerned with women’s rights in the Middle East North Africa (MENA).


Introduction
For many Western observers, Islamism, and more broadly governance based on Islamic principles (or any theocratic principles for that matter) is viewed as inherently antagonistic to democracy and women's rights. Theocracies are deemed antagonistic to democratic rule, since, rather than giving power to the people to govern themselves, power is granted to self-appointed surrogates of the divine who dictate their own interpretations. Based on this notion, the working assumption has been that to support women's rights, countries must be secular. Muslim-majority countries like Tunisia and Lebanon that have historically chosen the secular path 1 Personal status codes are the legal codes that govern family matters including marriage, divorce and child custody. are viewed as more supportive of women's rights, for instance, through more liberal interpretations of personal status codes regulating marriage and divorce. By contrast, Islamic interpretations of personal status codes (not unlike other monotheistic religions) is generally viewed as antagonistic to women's rights. This view is supported by the fact that countries living under an Islamic theocracy, e.g., Iran, Afghanistan under the Taliban, Algeria under the Islamic Salvation Front, have seen sharp curtailments in women's rights, including dictating the veiling of women and enforcing conservative interpretations of personal status codes.
Political scientists have also viewed the separation of church/mosque and state as necessary for democracy (Hashemi, 2009;Robinson, 1997). Realist foreign policy has backed Western friendly secular dictators under the premise that true democratic competition in predominately Muslim countries would undoubtedly culminate in "one-person, one-vote, one-time" (Blaydes & Lo, 2011;Hurd, 2008;Neep, 2004). In other words, true democratic competition would lead to the election of anti-democratic Islamist parties, unfriendly to the West that would also curtail political pluralism. Advocates for secularism in the region have pointed out that fostering peace, security and social justice and combating sectarian conflicts in religiously pluralistic communities demand secular governance.
An alternative view posits that Islam is not inherently antagonistic to women's rights or democracy, but rather that those Islamist movements that seek to counter a Western model of modernity and throw off colonial chains have advanced a return to 'purer' forms of Islamic observance. Veiling, sexual segregation and conservative interpretations of personal status codes governing marriage and divorce serve as key symbols of such a return. Moreover, theocracy, or religiously-based authoritarianism serves as an oppositional vision to Western liberal democracy. In this way, political Islam has been poised as a means of achieving a distinctly non-Western form of governance steeped in religious traditionalism. Women and men in the Muslim world have increasingly embraced a return to traditional conservative Islamic values and behaviors that had been thrown off by their parents and grandparents, and have seen it as a reclaiming of authentic identities (Hilsdon & Rozario, 2006). In this view, to embrace Western rhetoric about women's rights is to subjugate the mind to mental colonization, or to serve as agents of Western imperialism.
While initially secularism was viewed as the path to achieving gender equality, its association with the West and Western styles of government that generates a sharp division between Church/Mosque and state have become viewed as an imperialist notion and secularism identified with secular-authoritarian regimes. Three dominant schools of thought represent feminist dis-course in the region. Islamist and Reformist feminists suggest that Islam and women's rights need not be viewed as oppositional. They both ground their interpretations of women's rights in Islam and advocate for changes in the current interpretation of women's position in Islam. The first school advanced by Islamist feminists looks at early Muslim societies that embraced an egalitarian form of Islam, and argues that a return to fundamentals will guarantee women's rights. The second school of thought advanced by Reformist Feminists calls for reforms to keep up with the pace of progress in women's position. They argue that the concept of "Ijtihad"-independent reasoning and the employment of one's mental faculty in solving legal questions-means that the position of women (among other issues) was not meant to be static but flexible and open to modification across societies and time. (e.g., Mir-Hosseini, 2011). Secular feminists on the other hand, have advocated for a separation of mosque/church and state and for interpretation of women's rights that are not grounded in any religious ideology, but are based on equal rights and obligations of women and men as citizens of the state (Mir-Hosseini, 2006;Moghadam, 2002).
The events culminating in the Arab Spring have continued to complicate the simple dichotomies between secularism and theocracy and women's rights and Westernization. In several post-transition countries, moderate Islamist parties consistent with democracy and presenting interpretation of women's rights consistent with Islam are emerging, raising questions about the simple opposition frequently asserted between women's rights and Islam and between Islam and democracy (Boduszynski, Fabbe, & Lamont, 2015;Kurzman & Turkoglu, 2015;Netterstrom, 2015). Moreover, the uprisings initially appeared secular and inclusive in nature, with women having engaged actively in political activities in physical and virtual spaces, alongside their male compatriots (Abu-Lughod & Ferguson, 2014;Retta, 2013;Satterfield, 2013). 2 During the uprising, protesters with diverse political leanings were united against authoritarianism, but without a clear or shared vision of what the postrevolutionary society would look like. There was an opening of political space for women and the inclusion of demands of all marginalized groups in the activists' agenda. Islamists, secularists and modernists for a brief moment came together and collectively accepted a unified vision of a transition to a society that is more democratic.
It is unclear what the impact of the uprisings has been on women's rights. Has the Arab Spring created an opportunity to break down false dichotomies and build democracies with pluralistic religious parties and women's rights grounded in a progressive interpretation of Islam? Or does the same Western/modern versus Islamist/traditionalist logic continue to govern attitudes on women's rights in the Middle East and North Africa (MENA)?
To analyze these questions, we examine the degree to which citizens of countries that underwent recent regime change have altered their views on women's status compared with countries without regime change. We examine the degree to which lack of support of theocratic systems, a sense of political empowerment and having favorable views of Western societies predict more open attitudes toward women's rights. We additionally examine differences in attitudes between the sexes. Do women and men hold similar views regarding women's status adjusting for support for secularism and religiosity? Are women being held back by men or by their own making as has been suggested by some observers of women's embrace of Islamist ideologies 3 (Rozario, 2006)?

Women's Rights in the MENA
While some researchers suggest that the cultural fault line that separates the West and the Muslim world is about gender equality and sexual liberalization more so than democracy and that the apparent dearth of gender equality makes democracy unsustainable in predominately Muslim societies (Inglehart & Norris, 2003), others maintain that women's rights are caught in the crosshairs of deeper political battles (Al-Ali, 2012;Cohen & Enloe, 2003). What constitutes "women's rights" and whose conception of women's rights should apply remains a contentious issue and battles over these definitions reveal how power is constructed in authoritarian regimes, how inequalities are produced and reproduced in public and private spaces, and how the State reconstructs to maintain control of its population.
In much of the MENA personal status codes, deriving from Shari'a (Islamic law), governs the institution of the family and certain aspects of women's social and legal status. Personal status codes embody the symbolic representation of the "woman question" in the Middle East, or the problematic position of women in the modern state (Charrad, 2001;Hatem, 1994). As Charrad (2001) explains, "in the Maghreb, as in other parts of the Islamic world, women's rights as defined in family law are the crux of the matter. They are experienced as fundamental, as is reflected in the use of the expression 'women's rights' in the Maghrib to refer to family law" (2001, p. 5). In this respect, personal status codes represent the archetypal women's rights issue in the MENA, and it is around these codes that the struggle over the fraught "woman question" is frequently fought in the region (Hatem, 1994).
The promulgation of personal status codes and the impact they have on women's position has been an area of great contention in Muslim-majority countries, particularly from the vantage point of Islamist movements (Hatem, 1994). The interpretation of the Shari'a and how women's status is operationalized in these documents can have a profound impact on the status of women, either as a source of greater equality if "liberally" interpreted, or a source of restriction on women's full citizenship rights if "conservatively" interpreted (Ziai, 1997). Yet, studies of how instances of political liberalization in the MENA impact on women's status have received limited scholarly attention (e.g., Brand, 1997). What the public thinks of these codes has received even less attention. A recent Pew Forum on Religion and Public Life poll of Muslims in 39 countries found tremendous variation in attitudes on women's rights, which varied largely along regional lines (Pew Center on Religion in Public Life, 2013). The MENA region, according to the study, is the most conservative regarding gender equality compared with Muslim-majority countries in other parts of the world.
Apart from personal status codes, the issue of women's dress has taken on particular significance in the Muslim world. Initially embracing more modern dress post-colonialism and throwing off the veil, in the postindependence period, women in the MENA began increasingly adopting more modest forms of dress including the more restrictive burqua, which has diffused across the Muslim world and is being adopted in places where this practice is not traditional (Rozario, 2006). Full body covering and face covering was a traditional custom only in parts of the Middle East including Saudi Arabia, but not in North Africa. Head covering has also begun to be adopted in places where this was not previously the norm.
This raises questions about how to understand and interpret the diffusion of these "invented traditions" (Hobsbawn & Ranger, 1983). Should the choice to veil or gender segregate be seen as a form of empowerment, or as a repressive social norm? Does it represent women's internalization of their own subjugated status, or is it an expression of their independence as a modern, but non-Western authentic woman? Is it a political statement, and if so, what is the statement? Should laws banning veiling be seen as anti-democratic and contrary to religious liberty and freedom of expression just as laws requiring veiling are seen as repressive to women? Feminist scholarship has underscored the tension this creates for Muslim women who find themselves torn between their culture/religion on the one hand and their desire for more equality with men on the other hand (Winter, 2006). However, insofar as women's equality is painted as representative of Western individual rights, women are forced to choose between their group rights and their individual rights (Winter, 2006).
With these tensions in mind, we examine the degree to which recent attitudes in the MENA region reflect 1) the "secular-feminist" camp that views women's rights as achievable only through the separation of state and religion; 2) the "Islamist-reformist feminist" camp that views women's rights as consistent with and only achiev-able by the favorable interpretation of religious texts. To what extent can we distinguish support for these different types of feminisms in popular public opinion in the region? How do support for these different interpretations of women's rights vary across countries and over the time period of the Arab Spring?

Methods
Data Source. With a dearth of systematic and rigorous empirical research on public opinion in the Arab world, far too little is known about the nature, distribution, and determinants of the political orientations of ordinary citizens in the MENA. This study uses data from Waves 2 and 3 of the Arab Barometer Surveys. Data for Waves 2 and 3 surveys was collected using face-toface interviews, with households selected using a multistage area probability sampling approach. Male and female respondents eighteen years of age or older were selected from within households. In-country partners, for the most part based at universities, research centers, nongovernmental or private sector organizations, carried out the data collection process. We used data from all countries that had completed two rounds of Arab Barometer surveys (Waves 2 and 3). These included three Arab Spring countries (Egypt, Tunisia, and Yemen), and six non-Arab Spring countries (Algeria, Iraq, Jordan, Lebanon, Palestine, Sudan).
Although ideally we would have liked to capture the change in attitudes before the political uprisings that led to regime change and after regime change, the timing of Wave 2 surveys only allowed for examination of the period during the political upheavals in Arab Spring and comparison countries and about 2 years subsequently. Wave 2 of the Arab Barometer was collected during 2010/2011 and Wave 3 during 2013, over the period directly following the Arab Spring uprisings. In fact, the political uprisings themselves was what permitted the conduct of surveys that in the past had been prohibited in Egypt and Tunisia. Table 1 (appendix) summarizes the sample sizes and dates of data collection in each country in the sample.

Dependent Variable(s)
The goal of this research was to determine how attitudes towards women's rights and the interpretation of religious codes relating to women's status have changed over the period of Arab awakening in countries that experienced and did not experience political uprisings. Did the uprisings lead to greater support for women's rights, or a more radicalized, strict religious interpretation of women's status? Did other countries in the region that did not experience political change experience changes in attitudes towards women's status? How do Muslim-majority countries vary in their attitudes towards women's rights and support for a secular versus religious interpretations of women's status?
We were particularly interested in what might be termed "Muslim feminism" or the simultaneous support for women's equality, but a rejection of a purely secular interpretation of gender roles and relations. In other words, Muslim feminists are men and women who reject the notion that the only means of supporting women's rights is through state sponsored secularism.
Support for Muslim Feminism. 4 Our primary dependent variable is, therefore, a composite measure of support for gender equality and support for a strict or reformed reading of Islamic law as it pertains to women's status. This measure represents the interaction of individuals consistently holding beliefs in gender equality, but also advocating for a legal system that is based on interpretation of religious texts as they pertain to women's status. Below we describe how support for gender equality and support for a strict or reformed interpretation of religious doctrines scales were derived. Support for Secular Interpretation of Women's Rights (Secular Feminism). Four questions that appear in both Wave 2 and Wave 3 of the Arab Barometer Survey capture the degree to which respondents endorse a strict or reformed interpretation of Islamic law or advocate for secular legal codes related to women's status in terms of women's inheritance rights, the rules regulating marriage and divorce, gender segregation in education and rules regulating women's dress (i.e., enforcing the hijab). 4 There were major limitations in the question wordings for gender-related questions that impede inferences that can be drawn about support for different visions of how to achieve women's rights and gender equality that are worth noting. A number of the questions were double-barreled or awkwardly worded and did not provide enough nuance to identify different categories of feminists or supporters of women's rights. For instance, the question about requiring women to wear the hijab was phrased as "women should wear modest clothes without needing to wear a hijab". Disagreeing with this statement is ambiguous as it might mean that the respondent believes that women should be required to wear a hijab or that the respondent disagrees that women should wear modest clothes. Moreover, the statement that "the government and parliament should enact inheritance laws in accordance with Islamic law" and "personal status laws (marriage, divorce) in accordance with Islamic law" provides too little nuance about what this would actually mean. For example, in some instances, a strict interpretation of Islamic law, which at least guarantees that women inherit half of their husband's assets, may be more generous to women than a law that says that women should get nothing. For most countries in the MENA, the question is not whether Islamic personal status laws should be applied, but how they should be interpreted-liberally to confer more rights and greater gender equality or conservatively, constraining women's rights sometimes even more than a "strict" reading of Sharia law would dictate. Applying Islamic as opposed to customary law may in some cases provide more protections to women's status. The differences across countries in how these status codes have historically been interpreted is wide (Charrad, 1997). Likewise, the questions regarding gender equitable attitudes were somewhat flawed. The question asking to what extent respondents agree that "women can work outside the home," could be interpreted as more of a factual statement about whether women in that country in fact do work outside the home rather than an opinion statement (i.e., "women should be allowed to work outside the home unimpeded"). These limitations in question wording meant that the measure of "Muslim feminism" generated in this analysis may be too blunt an indicator to distinguish between those that endorse the more moderate "modernist" interpretation of Islam and those that might support a more radicalized political interpretation of Islam, or Islamists.
The questions are asked as follows: A. The government and parliament should enact inheritance laws in accordance with Islamic law; B. The government and parliament should enact personal status laws (marriage, divorce) in accordance with Islamic law; C. Gender-mixed education should be allowed in universities; D. Women should wear modest clothes without needing to wear a hijab.
Individuals who indicated strong disagreement/disagreement with statements A and B and strong agreement/agreement with statements C and D were coded (2), and identified as Secular Feminists. Individuals who expressed strong agreement/agreement with statements A and B and strong disagreement/disagreement with statements C and D were coded (0) and identified as Islamist. Individuals who somewhat agreed or somewhat disagreed with the statements were coded (1) and identified as Reformists. To develop the Muslim feminist interaction, Individuals with consistently Islamist viewpoints on women's status on three out of four questions were coded as 1 to combine with gender equality measures described below. Gender Equality Norms. Respondents were asked their degree of agreement/disagreement with the following statements on a 4 point scale: A. Women can work outside the home; B. Men are better political leaders than women; C. University education is more important for men than women.
The three measures were dichotomized and summed to produce three categories of respondents-those with consistently gender equitable responses are those who strongly agreed/agreed with statement A and strongly disagreed/disagreed with B and C (these were coded [2]). Individuals with mixed responses did not consistently endorse the more gender equitable responses (these were coded [1]). Individuals with consistently gender inequitable responses are those who strong disagreed/disagreed with statement A and strongly agreed/agreed with statements B and C (these were coded [0]).

Muslim Feminism (Islamist and Reformist).
Those with consistently gender equitable responses were interacted with individuals that consistently reported supporting an Islamic interpretation of women's status on three out of four questions to generate a measure of Muslim feminism. Those supporting Muslim feminism in the sample therefore represent individuals who both consistently endorse the application of Islamic law to women's status but also consistently endorse norms of gender equality. This measure is intended to capture individuals endorsing the Islamist and reformist schools of thought pertaining to women's rights in the region-individuals that believe that women's rights can and should be justified and grounded in religious interpretations of the Quran.
Each of these dependent variables (support for secular feminism, gender equality and Muslim feminism) was analyzed separately adjusting for covariates described below.

Independent Variables
Citizens' views of women's status are likely affected by their overarching political orientations as well as demographic characteristics. We adjust for several variables affecting individual attitudes towards women's status.
Support for Political Secularism. We analyze individuals' support for a politically secular versus theocratic regime using three variables: 1) support for Theocracy "a system governed by Islamic law without elections or political parties" (% believe this is very appropriate/appropriate); 2) support for secular democracy "Support for a parliamentary system in which only nonreligious parties compete in parliamentary elections" (% very appropriate/appropriate); 3) support for a pluralistic system "support for a parliamentary system where nationalist, left wing, right wing and Islamist parties compete in elections" (% very appropriate/appropriate). Each was treated as a separate dichotomous variable.
Support for Legal Secularism. We also included questions about support for a separation between legal codes and religious codes as this is likely to affect support for a secular interpretation of women's status. We used the following question to measure legal secularism: The government and parliament should enact laws in accordance with Islam (% disagree strongly/disagree). We hypothesized that those who were antagonistic towards secularism generally would be more hostile towards secularism in regards to women's rights.
Anti-Westernism. Individuals that have particularly negative views towards the United States or Western powers may reject women's rights or secular interpretations of women's status, which they view as representing Western enlightenment values. We include two measures of anti-Westernism: 1) % that agree/agree strongly with the statement "foreign interference is an obstacle to reform in your country"; and 2) % that agree/agree strongly with the statement "the United States' interference in the region justifies armed operations against the United States everywhere".
Religiosity and Religion. Individuals with a greater degree of religiosity should be less likely to hold secular viewpoints and may view gender equality as inimical to religious observance. We adjust for religiosity using three questions measuring individuals that report always praying daily, attending Friday (or Sunday) Prayer and always/most of the time reading the Quran. Though the majority of the individuals from each country are Muslim (94% of the total sample), we also adjust for the minority of respondents that are non-Muslim.
Gender. We anticipate women being more likely to support women's equality and a more progressive interpretation of religious scriptures on women's status. However, we also recognize that women often internalize inequitable gender attitudes and may also reject secularism as the basis for protecting and advancing women's rights. We therefore adjust for the gender of respondents. We also compute gender disaggregated point estimates for the each of the questions pertaining to women's rights and gender equality variables over the time period.
Age. As older individuals may have more "traditional" beliefs about gender roles and norms, or conversely, the younger generation may have less tolerance for secularism, we adjust for the age of respondents using three categories (18-34, 35-54, 55+).
Marital Status. Married individuals may have different views on personal status codes and gender norms than unmarried individuals. We code individuals as currently married or currently single (bachelor, divorced, widowed, engaged).
Treatment of Missing Variables. Overall rates of missing data were quite low. No single variable had more the 8% values missing or with individuals reporting don't know/refused. For values that were missing, we used multiple imputation to assign values.

Analysis
We first calculated the point estimates for each of the major dependent variables (attitudes towards women's equality, women's personal status) for each country at Wave 2 and 3 overall and disaggregated by gender. To calculate point estimates representative of the population, we applied sampling weights to the data and calculated confidence intervals. This allows for visual inspection of the overall change in attitudes across countries over the time period, disaggregated by question and gender.
Multivariate models were run as ordered logistic regression models for the measures of secular feminism and support for gender equality and logistic regression to examine predictors of Muslim feminism. The study employed a difference-in-difference approach comparing change in the dependent variables between Waves and between Arab Spring countries and non-Arab Spring countries adjusting for covariates. To examine the change between periods, in our models, we included survey Wave as an independent variable, and as an interaction term with Arab Spring and non-Arab Spring countries. If the interaction terms are significant, this shows that the change in outcomes in Arab Spring countries between the two time points is significantly different from the change in outcomes in non-Arab Spring countries. We also included country and Wave fixed effects to account for stable differences across countries between Waves.
All analysis was run using Stata version 13 using ordered logistic regression for secular feminism and gender equity scales (xtologit command) and logistic regression for models with Islamic feminism as the outcome (xtlogit command). Results from the ordered logistic regressions are presented as odds ratios (exponentiated coefficients) with confidence intervals for greater interpretability.
Overall, our hypotheses are that: 1) Arab Spring countries will see a larger change in attitudes over time than non-Arab Spring countries due to their political upheavals; 2) Arab Spring countries will see declining support for secular interpretations of women's rights and increasing support for Muslim feminism; 3) Countries with a history of secularism pertaining to women's status will exhibit higher support for secularism.

Results
Weighted point estimates: Support for women's rights and gender equality disaggregated across countries, Waves and gender. Results from the weighted point estimates were summarized in tables 2 and 3. The results show wide variation across countries in support for various aspects of women's rights and gender equality.
Overarching trends. In most countries an overwhelming majority agreed that personal status laws covering marriage and divorce and inheritance laws should be in accordance with Islamic law ( Table 2). The two exceptions to this trend were the historically most secular countries-Lebanon and Tunisia-where a minority and a lesser plurality respectively agreed that women's status codes should be interpreted in accordance with Islamic law. However, bucking this trend, a majority of Tunisians believed that inheritance laws should be interpreted in accordance with Islamic law. In addition to their higher support for secularism, Most countries were divided over whether women should be required to wear the hijab with Lebanon, Tunisia and Sudan most strongly opposing this requirement. Countries were also divided over whether university education should be co-ed with Lebanon and Tunisia showing the strongest support for mixed education over time.
Gender norms were also somewhat contradictory (Table 3). While large majorities in all countries disagreed with the idea that university education is more important for men than women and believed that women can work outside the home, in all countries except Lebanon a majority believed that men are better political leaders than women.
Disaggregated by gender. Female respondents in most countries held attitudes slightly more favorable towards women compared with their male compatriots, but not by much, and differences in attitudes were larger across countries than between the sexes within the same country. Overall, the disagreement between the sexes was much larger on matters of gender equality than the interpretation of women's status codes in relation to Is-lamic laws. Men and women in the same countries were mainly in agreement (within 5% points of each other) on the issues of interpreting women's marriage/divorce, inheritance rights, whether they should be required to wear the hijab and whether university education should be co-ed. There was much more disagreement over questions of gender equality with women in each country having more gender egalitarian norms; i.e., being more likely than men to believe that women can work outside the home, less likely to believe that men make better political leaders, and less likely to believe that university education is more important for men.
Disaggregated by Wave. Changes in point estimates between Waves 2 and 3 (2011 and 2013) were most notable in Algeria and Egypt, but followed somewhat contradictory and not consistent logics across countries. Algeria saw an increase in support for Islamic interpretations of personal status and inheritance laws, but also saw a notable increase in gender equitable attitudes including more individuals reporting that hijabs should not be required and that gender mixed education should be allowed. By contrast, Egypt saw a decline in support for strict Islamic interpretations of personal status and inheritance laws affecting women as well as decline in support for mandating hijabs. Egypt also increased in all categories of gender equitable attitudes except for seeing declining support for gender mixed university education. Tunisia saw large changes in gender equitable attitudes with fewer people reporting that men are better political leaders and that university education is more important for women but only marginal changes in attitudes towards the interpretation of personal status codes pertaining to women.
Multivariate analysis. Table 4 (appendix) summarizes each dependent and independent variable included in the multivariate analysis across all countries. Table 5 (appendix) shows the results of the multivariate analysis with all countries pooled for the three dependent variables-support for secular feminism, support for gender equality and support for Muslim feminism. Table 4 shows that approximately 15% of the pooled sample could be considered secular feminist (consistent support for a non-Islamic interpretation of laws/practices pertaining to women's status), 25% consistently supported gender equitable attitudes and 19% represented Muslim feminists (those who endorse consistently gender equitable beliefs, but also support Islamic interpretations of laws and practices pertaining to women).

Difference-in-Difference: Change in support for secular feminism, gender equality and Islamic feminism over time between Arab-Spring and non-Arab Spring countries. Examining the interaction between Wave and Arab
Spring countries in table 6, Arab Spring countries were less supportive of secular feminism by Wave 3 compared with Wave 2 but more supportive of gender equality and Islamic feminism compared with non-Arab Spring countries. Specifically, opposite trends were observed in Arab Spring and non-Arab Spring countries. Compared with the pre-Arab Spring time period, non-Arab Spring countries increased their support for secular feminism, but decreased their support for gender equality and Muslim feminism.
Predictors of Support for Secular Feminism. Model 1 of Table 5 summarize predictors of support for secular feminism (a preference for a secular interpretation of personal status laws for women). Respondents that consistently supported gender equality and had mixed views on the subject had higher odds of supporting secular feminism (OR = 1.35, 1.55 p < 0.01). Respondents that endorse theocracy in the political and legal spheres were less likely to support secular feminism (OR .75, P < 0.01; 0.21, p < 0.01 respectively) whereas those who endorse secularism in the political sphere were twice as likely to support secular feminism (OR = 2.02, p < 0.01). More religious individuals (individuals who reported praying daily and attending Friday prayer weekly) were less likely to endorse secular feminism (OR 0.85, 0.80 p < 0.01 respectively). Most demographic characteristics were not associated with support for secular feminism. Women were not more likely than men to endorse secular interpretations of women's status codes, nor were individuals living in urban areas, those with higher education or that were more economically secure. The one demographic characteristic that was strongly associated with support for secular feminism was being non-Muslim. Although only 6% of the sample reported that they were non-Muslim, non-Muslims were 9 times more likely to support secular interpretations of women's status (OR = 9.33). Support for secular feminism fell in Wave 3 compared with Wave 2.
Predictors of Support for Gender Equality. Support for equitable gender norms followed similar but also different patterns than support for secular feminism (Model 2, Table 5). Those who endorsed secular feminism also endorsed greater norms of gender equality. They were less likely to support theocracy but also less likely to support secular democracy (OR 0.75, 0.81 p < 0.01 respectively) and more likely to support a mixed political system that allows both religious and secular parties to freely compete. Those who regularly attend Friday prayer were less likely to endorse gender equality but those who regularly read the Koran were more likely to endorse gender equitable attitudes (OR 0.88, 1.18 p < 0.01). While women were not more likely to endorse secular interpretations of women's status, they were twice as likely to endorse gender equitable attitudes. More educated individuals but not those that were financially more secure were more likely to endorse gender equitable attitudes. Non-Muslims were more likely to endorse gender equitable norms but not to the same extent as supporting secular interpretations of women's status. Wave 3 respondents were more likely to endorse equitable gender norms compared with Wave 2.
Predictors of Support for Muslim Feminism. Model 3, Table 5 summarize characteristics of support for Mus-lim feminism, or support for Islamic interpretations of women's status coupled with support for gender equity. Individuals supportive of mixed political systems and an Islamic interpretation of the law were more likely to support Muslim feminism (ORs = 1.22, 1.40 p < 0.01), but individuals supportive of theocracy and secular democracy were less likely to endorse Muslim feminism (ORs = 0.83 0.89, p < 0.01). Those who think that foreign interference is an obstacle to reform and those who always attend Friday prayer were less likely to endorse Muslim feminism (OR = 0.71, 0.85 p < 0.01). Women were twice as likely as men to support Muslim feminism and more educated individuals, whereas non-Muslims were substantially less likely to support Muslim feminism (OR = 0.21, p < 0.01). By Wave 3 there was higher overall support for Muslim feminism compared with Wave 2.

Discussion
Regional attitudes towards gender equality and secularism. A primary goal of this study was to examine regional attitudes in the MENA towards issues of gender equality and support for a secular interpretation of women's rights versus more reformist views that gender equality can be achieved through a religiously based interpretation of women's rights. Attitudes towards women's rights and gender inequality has been understudied across the MENA region and particularly popular support for what the feminist literature from the region has described as Muslim feminism. We identified different orientations towards women's rights across the MENA region and observed variations in attitudes across countries over a period of widespread social upheaval.
We observed relatively low support for a traditional secular vision of women's rights whereby women's status issues are adjudicated according to civil law rather than Islamic law with only 16% of respondents dependably endorsing views consistent with keeping religion out of civil life as it pertains to women. There was relatively strong agreement across most countries that marriage, divorce and inheritance rights for women should be adjudicated according to Islamic law with the sole exception of Lebanon, a religiously mixed country with strong secular legal traditions, and Tunisia, though to a lesser extent. There was more variation across countries in attitudes towards requiring the hijab and gender mixed education at universities.
While respondents largely rejected secular interpretations of women's rights, we observed relatively high gender equity norms. Nearly 25% of the sample consistently endorsed equitable gender norms. In particular, large majorities across all countries disagreed with the statement that university education is more important for men than women. Large majorities also felt that women could work outside the home, though it is unclear to what extent this reflected a value judgment (i.e., women should be allowed to work outside the home) versus an empirical reality, particularly for many low-income households where women must work outside the home.
We identified relatively prevalent support for what might be considered a Muslim feminist view of women's rights with nearly a fifth of the sample nearly consistently supporting gender equitable attitudes but grounded in an interpretation of Islamic law and custom. The largest block of the sample, however, fell into neither a purely secular feminist nor Muslim feminist camp. We did not explicitly examine the prevalence of what might be considered a "Muslim fundamentalist" position on women's rights, namely those who are not supportive of women's rights and that believe in a conservative interpretation of religious law that would impose constraints on women's freedoms. Presumably some individuals fell into this category and others had more mixed and inconsistent ideological views.
Change in Arab-Spring versus non-Arab Spring countries. We predicted that Arab Spring countries would experience declining support for secular interpretations of women's rights. We suspected that revolutionary sentiments that rejected the authoritarian regime would also reject the persistent support of the West of that regime at the expense of the "Arab Streets". We anticipated that such resentment would manifest itself in the rejection of Western symbols including secular interpretations of women's rights and that we would see growing support for a Muslim feminist model that grounds gender equality in an interpretation of Islam. We largely found support for this thesis, though examining the toplines, more of the change in attitudes seems to be driven by Egypt than either Tunisia or Yemen.
We observed overall increases in support for attitudes consistent with this Islamic feminism and declines in support for secular feminism, between the two Waves and in Arab Spring countries compared with non-Arab Spring countries. On average, non-Arab Spring countries increased their support for secular feminism and decreased their support for Islamic feminism and gender equality between the two Waves whereas support for gender equity and Islamic feminism was higher in Arab Spring countries in Wave 3. These findings indicate a potential growing acceptance of distinctly non-Western, religiously-grounded interpretations of women's rights in Arab Spring countries. However, with the exception of Egypt, the changes in point estimates of attitudes between the two Waves in the Arab Spring countries (Tunisia and Yemen) were relatively small. Egypt and Algeria saw the largest changes in attitudes over the period. Egypt saw a 10 percentage point decrease in support for Islamic interpretations of women's status codes regarding marriage and divorce and inheritance, but also a 10 percentage point increase in support for requiring the hijab and almost a 20 percentage point increase in support for gender segregated university education, while also seeing a growth in equitable gender norms.
Egypt's larger mood shift could be attributable to the exceptionally turbulent period between 2011 and 2013 when Egyptians became rapidly disenchanted with the newly elected Muslim brotherhood government. Women in Tahrir square in Cairo during the height of the uprisings reported that "they have never felt as safe and been treated as respectfully as during the time of these protests" (Al-Ali, 2012, p. 27). However, the movement that began as a deliberately peaceful "selmiya" was quickly coopted by a regime that was trying desperately to hold on to power and resorted to the use of women's bodies as tools of warfare. Women's equal participation in the uprisings quickly gave way to harassment and violation of women's protesters that were widely employed by the police and their agents. When they assumed the reigns of power in Egypt, for example, the Muslim Brotherhood used women's position to solidify its control. Its pre-occupation with women-centered policies such as lowering the age of legal marriage for women and elimination of parliamentary quotas signaled an attempt to increase its political legitimacy. The shifts in Egyptian attitudes toward a more secular interpretation were potentially facilitated after the fall of the Muslim Brotherhood by a militarized regime that used secularism as a weapon to combat the influence of the Muslim Brotherhood, while still maintaining a tight grip on the population and an authoritarian rule. Additionally, the predecessor of the Muslim Brotherhood, i.e. the Mubarak regime, while authoritarian and repressive, took great pains to bolster its image through passing a number of women-friendly laws including the law of "Kholh" that granted women the right to easily initiate and obtain divorce through the courts. 5 The rapid shift from support of the Muslim Brotherhood to its ousting over this period could account for the larger attitudinal shifts in this country.
Attitudes in Yemen and Tunisia remained relatively stable in spite of the continued political unrest. The low degree of change in attitudes in Tunisia is somewhat surprising given the growth in the Salafi movement in the country following the uprisings and the turn the new government quickly took towards Islamism. Although claiming to be part of a democratic and moderate movement, the Salafis advocated for polygamy, urf or temporary, pleasure marriages, lowering the age of marriage, and female circumcision (Arfaoui, 2014). Proposed reforms included separation of the sexes, drawing a rigid line between the public and private spheres and sending women back to the "kitchens". Restricting women's public participation and solidifying gender division of labor represented their solution against unemployment (Arfaoui, 2014). While ultimately the Ennahdha Movement Party formed a coalition government with the largest secular party following the implementation of a new constitution in January 2014, the period of time captured in this study (2011)(2012)(2013) represents the height of the Salafist movement's reign when there was a growing support for subjecting women's status to a greater scrutiny of an Islamist lens. Yet, Tunisians attitudes towards the interpretation of personal status codes and gender equality changed little over the period perhaps because Tunisians were already accustomed to the concept of a moderate interpretation of religious codes without the need for pure secularism (Arfaoui, 2014;Netterstrom, 2015).
In Yemen as in other countries swept by the Arab awakening, women marshaled rallies, slept in protest camps, went on hunger strikes and covered the unrest as bloggers and photographers (Finn, 2015). Women were reported to be leading from the front lines. Yet, attitudes in Yemen changed little over the period except on the question of whether the hijab should be required. Support decreased by over 10% points from 51% to 41% reporting that the hijab should be required and an even greater reduction among men than women. At the same time, support for Islamist interpretations of status codes increased slightly. Although women's status in Yemen is widely regarded as among the worst in the region with nearly a quarter of all girls being married before the age of fifteen, high fertility rates, no penalties for domestic violence and nearly two-thirds of women being illiterate (Finn, 2015), their attitudes towards gender equality were not out of step with other countries in the region. As with Tunisia, this may represent the unfinished business of the revolution as women's movements have been put on hold due to the Houthis, a powerful Islamist rebel group based in northern Yemen, taking over the government.
Non-Arab Spring countries in the region held similar though also opposing views on issues pertaining to gender equality and women's status and views have changed in different direction over time. Given its unique power structure and political history, Lebanon stood out for its support of secularism and women's rights setting it distinctly apart from other countries in the region. This has changed little over the Arab Spring period. Palestine, Iraq, Jordan and Sudan experienced little substantial change in point estimates on attitudes towards women's status and gender equality, perhaps reflecting the fact that these countries did not experience Arab Spring uprisings, with the exception of Jordan where protests in 2011 quickly faded. Notably, in spite of a lack of political change in Algeria, there were substantial shifts in attitudes towards status laws and gender equality, including over a 10% point increase in support for a more Islamic interpretation of personal status codes, but also a large increase in the rejection of the hijab and support for gender mixed education. Some have speculated that Algeria was able to escape the Arab uprising because it already experienced such an uprising in the early 1990s when its personal status code underwent significant changes (Tlemcani, 2016). Although, Algeria did not experience regime change, the uprising did prompt discussions of reform and ultimately an amended constitution in 2016.
Demographic and other predictors of attitudes towards women's rights. Overall, a large majority of the sample preferred a legal system in which laws accord with Islam (73%) and endorsed a mixed political system in which religious parties compete openly with secular political parties (also 73%) over purely secular democracy or complete theocracy. A mixed vision of governance that accords with Islam while at the same time allowing democracy and women's participation in society appears to be a dominant motif in the region. In particular, women were more than twice as likely to endorse gender equitable norms and Islamic feminism, but not more likely to endorse secular feminism, suggesting that women in the MENA are largely rejecting the false dilemma between embracing either religion or women's rights.
The finding of gender differences in towards Islamic feminism accords with recent analyses that show that women's groups and female activists that participated in the Arab Spring uprisings have been largely disappointed in their lack of gains (and even possible loss of status) in the aftermath of the Arab Spring. Retta (2013) found, for example, that women have lost ground in their political representation in the post-revolutionary period in Egypt and Tunisia in spite of being heavily involved in the uprisings. For instance, the Supreme Military Council in Egypt revoked the quota of 64 women in parliament that had been put in place under Mubarak resulting in only 9 of the 508 members of the parliament being female (Retta, 2013). In Tunisia, the growth of Salafism has been blamed for a rise in sexual assaults post-revolution (Retta, 2013). These are similar dynamics as have occurred in previous revolutionary periods. For instance, in the revolution of 1919 in Egypt, the Egyptian feminist movement joined the nationalist movement for independence and marched in the streets calling for both the liberation of women and the liberation of the nation. The nationalist movement created a temporary space for women's demands, but soon closed that space and deemed that women's demands should not derail the task of nation building (Khattab, 2016).
Religiosity had mixed effects on attitudes towards women's rights. On the one hand, praying daily and attending Friday prayer were associated with lower support for secular feminism. Attending Friday prayer regularly was also associated with lower support for gender equity and Muslim Feminism, but reading the Quran regularly was associated with higher support for gender equality. If gender inequitable messages are reinforced in Friday prayer, this could undermine support for gender equality, whereas being knowledgeable about the Quran could provide a distinctly Islamic justification for women's equality. While overall trends seem to favor Islamic interpretations of women's status codes, non-Muslims stand out for their particularly strong embrace of secular feminism likely due to their minority status in the region. Non-Muslims also endorsed more gender equitable attitudes overall. This finding underscores an on-going tension-even if Islam can be consistent with gender equality, how can religiously grounded political and legal systems accommodate people of different faiths if they eschew secularism?
More educated individuals were more likely to hold gender equitable attitudes and to endorse Islamic feminism, but not more likely to endorse secular feminism. Whereas higher education has been linked to greater radicalization (Krueger & Maleckova, 2003), greater education in this case is associated with a distinctly Islamic interpretation of women's status codes but that is still consistent with gender equality. Surprisingly though, individuals endorsing Islamic feminism were less likely to hold anti-Western views suggesting that anti-Western political consciousness is not at the root of support for Islamic authenticity in women's status. This finding highlights the distinction between an Islamic fundamentalist interpretation of women's rights, which may be more closely bound to anti-Westernism, versus the Muslim feminist interpretation that yearns to ground women's equality in religious texts rather than ideology.
A limitation of this analysis is the timing of Wave 2 and 3 directly following the uprisings in Arab Spring countries. Although this was the necessary context that enabled data collection in the first place, it is unclear how attitudes might have differed prior to the beginning of the political uprisings. The results are best interpreted as change during the period of the Arab Spring.

Conclusion
Individuals in the MENA, particularly women, are largely rejecting the false dichotomy between religion and women's rights, and feel that Islam is not necessarily antagonistic to women's rights. All else equal, support for uniquely Islamic interpretations of policies pertaining to women's rights increased over the Arab Spring period particularly in Arab Spring countries. Relatively high degrees of support for gender equality seem to co-exist with a preference for Islamic interpretations of personal status codes pertaining to women. With large majorities of individuals endorsing preferences for "mixed" political and legal systems that allow for a greater incorporation of religion into public life, gender equality is not being viewed as inimical to Islam. The region over this time period was characterized by relatively low support for secular interpretation of women's status.
Although the turn towards women's rights embedded in Islam breaks down false dichotomies often assumed between being religious or being feminist, women's status appears to remain a site of conflict in the region. Women and gender remain central to the construction of social, political and religious hierarchical structures and political control of communities, whether it is ethnic, political or religious (Al-Ali, 2012). Our study confirms that it is not religious doctrines, per se, that construct and reproduce gender regimes and ideologies, but the use of specific interpretations of these doctrines to create and recreate controllable communities and citizenry by the state and its operatives. Women's bodies are still used as an effective site of social construction and control. Our study also confirms that attitudes about women's position are never static but are always fluid, and shaped by state institutions in authoritarian regimes. As has occurred far too many times in history, female activists in Arab Spring countries have faced a hurdle confronting all revolutionary politics: how to transform the egalitarian spirit of a brief uprising into a long-lasting revolution for women's equality. rights: Some general political and theoretical considerations. Women 's Studies International Forum, 29, 381-393. Ziai, F. (1997) Her research has focused on gender, sexuality and health in a global context. Her theoretical interests concern the role of public opinion, party ideology and descriptive representation in public policymaking.
Sana Abdelkarim Alzwawi is a fellow at Center for Women in Government and Civil Society (CWGCS) at the University at Albany; she is placed to work in NYS Department of Health, at the Office of Public Health Practice. Prior to her arrival to the United States Sana had been working with the International Medical Corps-Libya as a Gender Based Violence Program Trainer and Officer. In 2012 Sana had earned her Bachelor of Science degree program in Pharmaceutical Sciences from Benghazi Medical University.
Dina Refki researches the interplay of gender and institutional structures. She applies feminist theories in national, subnational and international contexts. Refki studies women's political leadership and has researched women's substantive and descriptive representation in elected and appointed governance in the United States and internationally. Dr. Refki is the author of several gender-focused research publications produced at CWGCS. She holds a doctorate in Humanistic Studies and a Master in Africana Studies from the University at Albany.