DETERMINANTS OF CIVIC ENGAGEMENT AND POLITICAL PARTICIPATION: A CASE STUDY IN YEMEN

Transcription

1 DETERMINANTS OF CIVIC ENGAGEMENT AND POLITICAL PARTICIPATION: A CASE STUDY IN YEMEN A Thesis submitted to the Faculty of the Graduate School of Arts and Sciences Of Georgetown University in partial fulfillment of the requirements for the degree of Master of Public Policy in Public Policy By Montaha Hassan, B.S. Washington, D.C. December 11, 2014

2 Copyright 2014 by Montaha Hassan All Rights Reserved ii

3 DETERMINANTS OF CIVIC ENGAGEMENT AND POLITICAL PARTICIPATION: A CASE STUDY IN YEMEN Montaha Hassan, B.S. Thesis Advisor: Jorge Ugaz, Ph.D. ABSTRACT Yemen s exercise with democracy started with the unification of Yemen Arab Republic and the People s Democratic Republic of Yemen in 1990, where the ruling parties acknowledged that to retain power, they had to share power. This window of opportunity instituted a hopeful democratic process for a region with scant freedom of expression and assembly and fair elections. In the early 1990s, a large number of political parties, civil society organizations and charities were founded. However, socio-political challenges weakened the emerging plurality. As soon as the 1994 civil war ended with the defeat of the Yemeni Socialist Party, true political vying amongst political parties declined and politicians were no longer relying on the support of national constituencies. In this research, I hypothesize that women, younger citizens, and residents of southern cities and residents of Saada home of an imamate insurgency and plagued with armed conflicts with the central government since have different levels of civic engagement relative to men, older citizens and residents of other northern cities, respectively. To explore the different levels of civic engagement and political participation, main voting patterns and likelihood to be a member of a political party, I rely on multivariate regression techniques, including Ordinary Least Squares and Logistics models using data from the Arab Barometer II (2011) and the World Values Survey Wave 6 (2014). My analysis reveals that: 1) In 2011, there was no difference in voting patterns between men and women, whereas in 2014, men are found to be slightly more likely to vote than women. On the other hand, there is a statistically iii

4 significant difference between their likelihood to be members of political parties; and 2) Younger citizens between the ages of 18 and 25 are less likely to vote or be members of political parties relative to their older counterparts; and 3) In 2011, residents of South Yemen were less likely to vote than residents of North Yemen, whereas in 2014, residents of South and North Yemen were as likely to vote. On the other hand, there is a statistically significant difference between their likelihood to be members of political parties in 2011 and It is my hope that these findings bring new insights into the current political debate, especially during this transformative period that initiated the Constitution Drafting Committee prerogatives. The results of this study should also guide political parties to develop inclusionary policies to appeal more to female members and the youth; while the central government works on developing policies prioritizing economic growth, social cohesion and short to medium term exceptional policies to remedy past discriminatory actions endured by residents from South Yemen and encourage their political engagement to express their aspirations to affect the policy making process. iv

5 ACKNOWLEDGEMENTS The research and writing of this thesis is dedicated to Hamido. Me more! MONTAHA HASSAN v

6 TABLE OF CONTENTS INTRODUCTION BACKGROUND LITERATURE REVIEW... 7 HYPOTHESES DATA EMPIRICAL MODELS RESULTS DISCUSSION CONCLUSION APPENDIX BIBLIOGRAPHY vi

7 List of Tables Table 1: How Many People Reported Voting Table 2: How Many People Reported being a Member of a Political Party.22 Table 3: Distribution of Respondents by Region..23 Table 4: How much do Respondents Trust the Following Institutions Table 5: Fair Elections? Table 6: Interest in Politics Table 7: Distribution of Education Attainment 26 Table 8: Voting Patterns (Arab Barometer II).. 28 Table 9: Voting Patterns (World Values Survey Wave 6) Table 10: Membership in Political Parties (Arab Barometer II) Table 11: Membership in Political Parties (World Values Survey Wave 6) vii

8 List of Appendix Tables Table 1: Odds Ratio of Voting Patterns (Arab Barometer II) Table 2: Odds Ratio of Membership in Political Parties (Arab Barometer II) Table 3: Odds Ratio of Voting Patterns (World Values Survey Wave 6) Table 4: Odds Ratio of Membership in Political Parties (World Values Survey Wave 6) viii

9 INTRODUCTION This study aims at exploring the association of gender, age and region of residence on citizens civic engagement in Yemen. Exploring the existence of discrepancies in civic engagement levels between men and women, between younger and older citizens, and among different regions of residence should help shed light on immediate social, political and economic policies that the Government of Yemen should enact. Moreover, knowing the sign and power of association of either of these three factors on civic engagement can allow policy makers to better target their policies to increase civic engagement among all citizens while developing inclusionary policies that aim at increasing women s, youth and regional participation. This is vitally important because there is anecdotal evidence that women, youth and citizens living in certain regions within Yemen have been neglected by urban-male-elite politics, that often utilize the issues of women rights, youth unemployment and rural poverty as targets of their discussions, yet never as actors of these debates. A number of researchers documented the setback expressed by women advocacy groups, youth movements and certain regions by the low levels of civic engagement over the last twenty years (Philbrick Yadav 2009). To better understand the analysis put forth in this thesis regarding civic engagement and political participation levels in Yemen, it is important to explain what these two terms mean and how researchers use them. Tessler et al. (2012) define civic engagement as follows: Scholars have long recognized that stable democracy depends upon having not only the proper political institutions but also a democratic culture. This includes, for example, high levels of interpersonal trust, political interest, involvement in community and civic organizations, and tolerance of others. [emphasis added] 1

10 Scholars use civic engagement and political participations interchangeably. Tessler et al. (2012) acknowledged the necessity to present a clear definition of the two terms, where civic engagement encompasses political participation, community service and collective action. Joakim Ekman and Erik Amna (2012) concur with this definition, where they recognize that some scholars focus on the private aspect of civic engagement that is expressed in political participation, while others emphasize the social aspect of civic engagement, which is often characterized with community service. Drawing from data on 19 Western and 11 Muslim countries gathered by the World Values Surveys Wave 3 and Wave 4 ( ), Yuchtman-Ya ar and Alkalay (2010), state that in Muslim countries (Yemen not included in the analysis), the best predictor of democratic values is education followed by income, while age and gender have weaker association with democratic values with men, and younger individuals being more prone to adopt democratic values than women and older citizens. My thesis does not delve into the effect of education on civic engagement as over 40 percent of the Yemeni population is illiterate (World Bank 2009), and this variable would yield better results if we compare groups that have different educational attainment, not literate with illiterate citizens. Tessler, Jamal and Robbins (2012), on the other hand, note that political participation in Yemen is the lowest in the Arab World, with a rate of only 36 percent, while civic engagement rates are estimated to be 26 percent. Yet, the Second Wave of the Arab Barometer Survey shows that Yemen civic engagement rates increased dramatically by 23 percent post-arab Spring. These findings measure the effectiveness of social, political and economic events impact in defining political participation s shift. Having evidence on the associations between gender, age, and region and civic engagement in Yemen is 2

11 important in order to develop policies that address the root causes of low levels of civic engagement. Such remedial policies could include: 1- To address low female political participation: mandate girls education; set 18 years old as the minimum legal age to marry; 1 and implement awareness campaigns to change attitudes towards women s role in society. 2- To address youth political engagement: implement robust anti-corruption laws; attract investments to generate employment; and engage youth in strategy formulation and civil society development. 3- To address discrepancies between regions: expedite the drafting of the new constitution that should stipulate the formation of the new 6-regions federal system 2 ; transfer budgetary and administrative prerogatives to regional councils; and implement long term policies to overcome the influence of factions - political, tribal, or military - on national politics. 1. BACKGROUND This thesis focuses on three segments within the Yemeni society that have been marginalized in terms of development, labor and political participation. For example, women enjoyed voting rights since 1970, yet these rights have not translated into equal distribution of human development goals between genders; Yemen ranks at the bottom of the 149 countries 1 According to Human Rights Watch, more than 50% of Yemeni girls are married before age 18, while it is estimated that around 14% of girls are married before age Announced by President Abd Rabbuh Mansur Hadi in February

12 measured by the Gender Inequality Index. 3 On the other hand, Yemeni youth have to face increasing living expenses and rising unemployment rates, while witnessing flagrant levels of corruption and nepotism. Finally, the centrally controlled regime, in an attempt to concentrate its power, allowed discriminatory policies against the South of Yemen and Saada due to what started as political confrontations and ended with armed conflicts. The following section describes the characteristics of these three segments. 1.1 Political Participation and Gender: In Yemen, women s low levels of education is known to negatively impact family s health, income, and the country s overall economic growth (Hill and King 1993). It may also impact political participation rates since women who face daily struggles due to poor health and low economic prospects do not have time to participate in political activities (Riphenburg 1999). According to Carol Riphenburg (1999), the proportion of women s votes was only 19 percent and 30 percent in the parliamentary elections of 1993 and 1997, respectively. Women s participation in political discourse is crucial in generating debates around specific societal issues, such as girls education, human rights, economic development and equality. Often times, parliamentarian representation, which is mostly comprised of male members, do not focus on female related issues. As a result, public policies are often biased in male dominated societies. In the wake of the Arab Spring, women in Yemen have played a major role in expressing political opinions; however, this new shift in attitudes may be limited to urban areas, where protests took place, while 70 percent of the Yemeni population lives in rural areas (Caton 2013). Understanding the average Yemeni female political involvement and participation in elections is 3 The Gender Inequality Index is obtained from an adjusted Human Development Index, where Yemen ranks at 154 out of 187 countries. UNDP Human Development Index, The UNDP Human Development Index (HDI) measure a country s positioning in literacy, life expectancy, GDP per capita, and other indicators. 4

13 one objective of this study. Increasing women s political participation should contribute to leveling the debate to address legislative and societal obstacles and will contribute to putting in place policymakers who can address gender inequality gaps in terms of education, health, employment and economic growth. 1.2 Political Participation and Age: The study s second research question will be to explore whether there is a significant generational difference with regards to civic and political engagement levels. In many mature democracies, older constituents often participate more in political debates and have higher voter turnout. However, in Yemen, where 63 percent of the population is under the age of 25, 4 many social grievances and political discontent were mainly expressed by the youth, which culminated with the toppling of President Ali Abdulah Saleh and his government in It is important to examine whether high unemployment rates (World Bank estimates at 60 percent in 2012) among the youth and their longing for equality, opportunity and freedom have been articulated through increased political participation. In 2012, a presidential election was held, and a new transitional government was formed; yet youth membership within the newly formed government was limited since the former ruling party (General People s Congress) and the opposition parties (Joint Meeting Parties) were competing for their shares of cabinet seats. The National Dialogue Conference - a 565-member entity tasked with developing policies to solve Yemen s challenges- had 20 percent participation of youth, although debates with impactful ramifications on the Yemeni society were limited to the participation of established political parties. 4 Estimation for 2014 according to Population Projections ( ) by Central Statistical Organization (CSO) and National Population Council (NPC)- The Government of Yemen,

14 Therefore, it is important to test whether Yemeni youth is civically more engaged than their older counterparts since Yemeni youth were the instigators of the 2011-uprising, and have access to technological means that allow them to observe global political and economic trends. In addition, the level of literacy is different among generations in Yemen, and this distinction may positively impact youths interest in politics and realization of the importance to enact adequate public policies to improve their livelihoods and socio-economic prospects. This research addresses generational civic engagement levels while accounting for differences in educational attainment. It is essential for policymakers to develop policies that address youths concerns in education, labor, and health. It is equally important to assess whether the youths use elections or membership in political parties to express their generation s aspirations. 1.3 Political Participation and Region of Residence: Finally, this research examines the differences among regions in terms of their political attitudes and civic engagement. In Yemen, the central government controls policy trajectories. In 2000, a Local Authority Law was ratified by the parliament, however real self-governing prerogatives were not transferred to the 21 governorates. Additionally, major private and public investments and most public service employments are clustered around the capital, and to a lesser extent, other major cities. This study will examine the difference between individual civic engagement based on region of residence. Some regions witnessed uprisings during the last decade on the basis that the central government had neglected their social and economic development needs. Citizens in South Yemen and Saada, claim that due to armed and unarmed conflicts with the centrally represented regime, regions were punished by ostracizing public figures from these specific regions, and unfair public policies. Others in the western region of Hodeida claim that historical neglect 6

15 continued into the modern republic. This is noteworthy as Hodeida houses over 30 percent of the national electoral constituency, yet does not have a comparable impact on policy trajectory as do areas with tribal powers such as Aljawf, Mareb and Amran (Fattah 2011). It is important to understand how these perceptions justified or not influence individual interest in political debates and civic engagement. Some theories may indicate that grievance could be a good driver for political engagement; it is through better representation that real change can happen; Other theories indicate that if citizens trust in the government is waning, then participation or lack of participation in elections would not influence the lives of those living in regions with lesser education or investments. If discrepancies in civic engagement exist among regions, policy makers should pay more attention to the root causes of this difference when designing policies that aim to decrease the regionally led/caused conflicts (peaceful or armed). 2. LITERATURE REVIEW The political and socio-economic grievances of Yemenis were responsible for toppling the country s 33-year-in-office president in Although the first peaceful demonstrations were instigated by educated youth, with robust female participation, organized political parties soon hijacked the youth movement (Durac 2012). Since the 1990s, political parties have engaged in similar deceptive actions that ignore popular demands in favor of their own parties advancement goals. When the Yemen Arab Republic and the People s Democratic Republic of Yemen declared their unification in 1990, the northern non ideology-based General People s Congress (GPC) and the southern Yemeni Socialist Party (YSP) presented a unique opportunity for political pluralism in the Arabian Peninsula (Williams 1997). Specifically, sharing power 7

16 between the GPC and YSP represented two strong parties competing for political support, while other political parties - the Islamist Islah, the Nasserites, the Union of Popular Forces, AlHaq and AlRashad parties -were active in the same time period. The growing political confrontation between the GPC and YSP led to the 1994 civil war, where the YSP leaders declared the secession of the south and was defeated thereafter. This section explores what the current body of literature has found in terms of determinants of civic engagement and political participation by gender, age and regional factors. 2.1 Gender Determinants: Yemeni women s civic engagement increased considerably after the unification of Yemen in The shift occurred as single-party countries united and additional political parties were founded to counter the influence of the two most prominent parties: the GPC of the former Yemen Arab Republic, and the YSP of the former People s Democratic Republic of Yemen. The competition amongst political parties meant the development of policies, at least rhetorically, to attract voters, especially women s votes. Until the 2011-uprising, women in Yemen had participated in three parliamentary elections, two local council elections and two presidential elections. Yet, a United Nations Development Programme (UNDP) report published in 2008 places Yemen among the world s three worst countries in terms of equal distribution of human development gains among men and women (UNDP 2008). 5 The initial optimism among different women groups quickly dissipated following the 1994 civil war, where the YSP was defeated and its leaders demonized as atheists, (Philbrick Yadav 2009) and the political competition was drastically reduced to the GPC and the newly founded 5 The 2014 UNDP Human Development Index places Yemen at the bottom of the 149 countries measured for Gender Inequality. 8

17 (with GPC and Saudi funds and support) Islamist Islah party. Women then transitioned to associational engagement, rather than political engagement (Philbrick Yadav 2009). Stacey Philbrick Yadav (2009) asserts that women in Yemen were disappointed with political parties since women issues and women rights were often discussed within parties discourses with the sole objective of attracting non-governmental organizations (NGOs) or foreign aid, while excluding women from the debate itself. Yemeni women accuse political parties of being more concerned with intra-parties male-elites conflicts. No true commitment to women s political or socio-economical advancement was undertaken by any of the political parties. Michaelle Browers (2006) attributes the lack of political will to the patriarchal characteristic of both secular and conservative parties in the Arab World. With the exception of election campaigns, women issues are either under the radar or a low priority in political party discussions. Yet, during the civil uprising of 2011, female participation was high which reflected a new shift in women s civic roles. Khaled Fattah (2011) captures this sentiment in his article Yemen: A Social Intifada in a Republic of Sheikhs, where he states that Yemen s social intifada created a new space for women s empowerment, networks, courage and voices. 2.2 Generational Determinants: 75 percent of the Yemeni population is under 35, while the median age is between 17 and 18 years of age (Population Action International 2010). In mature democracies, older generations tend to be more civically engaged, especially in terms of voter turn out, membership in political parties and civil societies organizations. In Yemen, research shows that the youths, who suffer from unemployment and witness flagrant control on the economy by an authoritarian regime, were outraged and took to the streets to demand change in the youth-instigated protests in 2011 (Gallup and Silatech 2009). Salamey and Pearson (2012) 9

18 call the Arab youth a defiant class of citizens. Yemeni youth high unemployment rates reaching 49 percent (Gallup and Silatech 2009) in 2009, and 60 percent in 2012 (World Bank 2012); their exposure to regional democratization trends along with their higher education levels vis à vis their parents may be conducive to higher levels of civic engagement (Staeheli and Nagel 2013). In 2009, Gallup carried out survey research commissioned by Silatech Qatar to analyze Yemeni youth attitudes on employment and the labor market in general. The results of the survey show that compared to other Arab League nations, 16 percent of Yemeni youth displayed the lowest optimism in finding employment, compared to the Arab League median of 34 percent. One of the reasons that led the Yemeni youth to join the demonstrations of 2011 was the high level of corruption in the government, which affects youth livelihoods since 80 percent believe that knowing a high ranking bureaucrat is critical to securing a job (Gallup and Silatech 2009). Vincent Durac (2012) categorizes Yemeni youth into three groups: a) apolitical and not interested in politics, even after the 2011 regime change; 2) youth aligned with established political parties; and 3) youth who realized that change would only come through civic engagement and political participation. This thesis will try to analyze whether the Yemeni youths lack of optimism in their employment and education prospects are reflected in their civic engagement and political participation. Until recently, youth civic engagement was believed to be limited due to low levels of trust in political parties and the government; limited interest in politics and associational participation (Durac 2012). The objective is to analyze how Yemeni youth choose to voice their political and socio-economical frustrations and attain their aspirations through democratic means. 10

19 2.3 Regional Determinants: The victors of the 1994 civil war vowed to implement public and private investments in the southern cities. Pledging to elect southern officials to key public positions demonstrated the new regime s rhetorical commitment to countering the concerns southerners had repeatedly expressed: that the central government would penalize them for supporting the secession war of 1994 (Williams 1997). 6 Yet, following the 1994 war, southern cities have experienced the victor s negligence at best and his punishment at worst. The repetitive discriminatory policies towards residents of South Yemen led to the uprising of the Southern Hirak movement starting from 2007 demanding reparations. In 2011, The Southern Hirak movement started demanding secession. The National Dialogue Conference (NDC) was formed as an element of the 2011 transfer of power deal brokered by the Gulf Cooperation Council (GCC). The NDC is a 565-member body representing different segments within the Yemeni society, who were tasked to formulate policies to address challenges facing Yemen. The NDC recognizes past injustices committed by the central government in southern cities (NDC 2014). Other regions in Yemen also claim that injustices were committed against them based on a confluence of political/tribal/sectarian confrontations, negligence, high levels of corruption, and low accountability at the central government level. For example, government military forces were deployed to Saada starting in 2004 (Fattah 2011). Saada is home to an imamate 7 insurgency whose early demands in the 1990s were to be able to freely exercise their Zaydi beliefs in light of the growing influence of Sunni political Islam, especially after the founding of the Islah party. These demands quickly escalated to establishing an imamate à la pre-revolution Mutawakilite imamate, where only Hashemites are allowed to govern. Between 2004 and 2010, the 6 Remarks made by then Prime Minister, Mr. Abdulaziz Abdulghani, Imamate comes from the word imam, which means revered religious leader in the Zaydi-Shia sect of Islam. 11

20 government led 6 military campaigns in Saada that caused material destruction, human losses and dislocation of whole communities. The Saada issue was also recognized during the NDC deliberations as one where previous governments implemented unjust policies, and committed human rights abuses (NDC 2014). Regions such as Tihama, which represents the country s highest concentration of poor, and Taiz, which has the country s highest percentage of educated adults, also complain of systemic negligence in terms of investment in infrastructure and implementation of development projects. The disparity among different Yemeni regions was clearly reflected in the dynamics surrounding the formation of the new six-regions federal system during the NDC. NDC prerogatives were completed in January 2014 (Al-Akhali 2014), resulting in the formation of six federal regions: Azal, Tehama, Aljanad, Aden, Hadramut, and Sheba. The grievances of different regions were echoed in their desire to be separated, at least administratively, from the central government. This research aims at analyzing how individuals from different regions in Yemen engage civically. Do individuals surveyed in different parts of Yemen respond differently to questions related to voting patterns, and the levels of trust in government or political parties? This research will analyze political attitudes in regard to different civic engagement actions among regions, while emphasizing discrepancies between advantaged and disadvantaged regions. There are few studies that have explored, in a systematic and quantitative manner, the relationship between voting and being a member of a political party with gender, age and region of residence in the Arab World. Mark Tessler and Amaney Jamal (2006), the co-principal directors of the Arab Barometer, note that unlike other regions such as Eastern Europe, Latin 12

21 America or Sub-Saharan Africa, the Arab World did not benefit from an analysis of civic engagement based on quantitative studies. The authors attribute this tendency to researchers reliance on cultural and religious characteristics to predict and explain political attitudes of Arab citizens, but also due to logistical challenges in conducting surveys in non-democratic states. Political expert Iliya Harik states that: Political attitude surveys are possible only under conditions of political freedom. (as cited in Tessler and Jamal 2006) For example, the World Values Survey Waves 1, 2, 3, and 4 covered up to 85 countries, but not a single Arab country. Wave 5 covered five Arab countries, and only Wave 6, implemented in 2014, includes Yemen. The Arab Barometer on the other hand, included Yemen since its initiation in Furthermore, research conducted by Tessler et al. (2012) (details in the INTRODUCTION Section) only describe levels of support for democracy, role of women in the work force, and tolerance towards political plurality with a focus on political Islam. Another survey was conducted by Abu Dhabi Gallup Center amid the unrest in 2011 to study Yemenis confidence in the government. The number of Yemenis saying they do not have confidence in their government increased from 39 percent in late 2010 to 48 percent in Yemenis who do not have confidence in their government are more likely to be male (63 percent) than female (37 percent) (Abu Dhabi Gallup Center 2011). This may be attributed to the fact that men are pressured to earn a living, while most Yemeni women do not work. In addition, Yemenis who do not have confidence in the government tend to be more educated than those who have confidence in the government. Regarding the impact of income on the levels of confidence in the government, individuals who have more economical concerns and find it very difficult to get by with present income are more likely to have less confidence in the 13

22 government than those who can get by with present income. Among the government supporters, only 56 percent believe in the integrity of the electoral process, while faith in the electoral process is non-existent among government detractors. Another interesting finding of the Gallup Abu Dhabi survey is the absence of generational difference among government dissidents. 8 The conclusion regarding determinants of levels of trust in government described in this survey will be compared with findings of my thesis analyses. It is worth noting that no study has performed any quantitative analysis to explore the possible disparities in civic engagement levels based on gender, age or region of residence in Yemen. This also means that no multivariate analysis has been conducted, where one can control for all of these variables at the same time and obtain a more precise relationship. Having evidence on the correlation between gender, age and region of residence and civic engagement in Yemen is important in order to better understand the complexities of the socio-political attributes of the country s political arena. Most importantly, this evidence will guide policy makers and development partners to advance policies that address the root causes of societal grievance and low levels of civic engagement in Yemen. 3. HYPOTHESES The U.S. Middle East Partnership Initiative (MEPI) criticizes the central Government of Yemen prior to the regime change in 2011, for not allowing the growth and dissemination of civil society organizations (CSOs) activities. The same can be said for indirect restrictions on political party activities. Yet, there is a reported 7,000 non-governmental organizations registered in the country, whose main activity is to advocate for their members interests, be they 8 Gallup survey was based on a face-to-face interview with 1000 Yemenis, ages 15 and over. 14

23 agricultural, charitable, income-generation societies or professional syndicates (Mitchell 2012). The low membership rates may have been the result of the restrictions on CSOs to reach out to Yemeni citizens. In this thesis, I will use data gathered by the Arab Barometer II and the World Values Survey Wave 6, where 2,200 Yemenis were interviewed during the period before the regime change in 2011 and in 2014 (presidential elections were held in 2012). The issues covered by the questionnaires used in the face-to-face surveys are substantive in terms of measuring public and personal civic engagement trends. As explained earlier in this thesis, public civic engagement relates to charitable and community-service-like organizations, while private civic engagement relates to voting and political participation. Yet, in light of some limitation due to a large number of missing observations, the questions regarding associational engagement (public civic engagement) are not covered in this thesis. According to Yuchman-Ya ar and Alkalay (2010), income is the second most powerful determinant to measure support for democracy in the Muslim World, which is covered in the analysis of the World Values Survey Wave 6 data set, but not covered in the Arab Barometer II data set since over 50 percent of the observations for Yemen are missing. In the same article, Yuchman-Ya ar and Alkalay (2010) demonstrate that education attainment is a stronger determinant of the levels of support for democracy than income. Therefore, in this thesis, education levels are controlled for in the analysis for both data sets used. Income and education attainment variables are not part of the main covariates due to high illiteracy rates in Yemen and the limitations of data gathered regarding respondents income, but they are controlled for in the regression analyses. 15

24 Based on the available data and scholarly articles consulted, I propose the following hypotheses: Hypothesis 1: Yemeni women are less likely to vote than men. Similar to the conclusions reached by Stacey Phillbrick Yadav (2009) and Sheila Carapico (1993), women s increased interest in political parties, elections and the process of policy making diminished after the 1994 civil war, where the ruling GPC found itself with no strong political competitor after the defeat of the YSP. Moreover, women s disengagement was also due to the crackdown on other political parties by the GPC and its temporary new ally, the Islamist Islah party, whose policies advocate for traditional gender roles. Hypothesis 2: Yemeni women are less likely than men to be members of political parties. Political party membership depends on a number of factors, particularly levels of trust in political parties. Yet, as Michaelle Browers (2006) suggests, both secular and Islamist parties in the Arab World failed to advocate for women s issues and women s rights for fear of alienating their members with strong patriarchal backgrounds. This policy inaction is especially true for secular parties such as the YSP whose by-laws favor women s empowerment, but whose political actions were limited to survive in an autocratic environment. Hypothesis 3: year old Yemeni youths are less likely to vote than the older cohort. Yemeni youth grievances started before the 2011 regime change due to the high unemployment rate, prevalent corruption and nepotism, when combined diminishes their prospects for better livelihoods. Youth grievances culminated with the instigation of the 2011 uprising that ended up toppling the incumbent who had been in office for 33 years.. 16

25 Hypothesis 4: year old Yemeni youth are less likely to be members of political parties than older cohorts. Salamey and Pearson (2012) argue that decades of social grievances caused by the unequal distribution of wealth, and tribal, sectarian and ethnic disparity resulted in the youth-led protests in the Arab Spring countries, which include Yemen. Durac (2012) on the other hand classifies the Yemeni youth in three categories: a) apolitical; 2) members of political parties whose goal is to bring about change from within by reforming the political parties they are members of; and 3) independent youth who believe that change can only be realized through increased political participation. Hypothesis 5: Residents from the regions of the former People s Democratic Republic of Yemen (here referred to as South Yemen), and Saada are less likely to vote than their compatriots in other parts of the country. NDC concluded in January 2014, resulting in the announcement by President Abd Rabbuh Mansur Hadi of the formation of six federal regions: Azal, Tehama, AlJanad, Aden, Hadramut, and Sheba. For this thesis, observations collected in Saada city were separated from other observations collected in Azal, 9 while Aden and Hadramut regions were combined. The issues of South Yemen and Saada were highly debated during the deliberations of the NDC as having legitimate grievances that the current government, political parties and civil societies acknowledge as requiring immediate and specific attention to remedy past discriminatory policies. Hypothesis 6: Respondents from the regions of the former People s Democratic Republic of Yemen (here referred to as South Yemen), and Saada are less likely to be members of political 9 In the new 6-regions federal system, Saada is part of Azal region. 17

26 parties than residents of other regions. In the same way that political parties disenfranchised women, they have also been labeled as urban-male-elitist entities preoccupied with internal conflicts and a constant search for government leadership positions or contracts. Fifty eight percent of Members of Parliament (MPs) attest that they do not identify with their political parties (Mitchell 2012), which may indicate that their loyalties are driven by self-interest. These political parties leadership characteristics push away the rank and file, and may appear opportunistic, especially if they claim to represent citizens experiencing discrimination. Such as in the case of the YSP low popularity in southern cities for the party s leaders position on secessionist demands or Islah party s decline in popularity since the 2011 regime change due to the party s prioritization for securing cabinet seats in the transitional government over public interest. 4. DATA Data sets and sample For this thesis, I used individual-level data obtained from two face-to-face surveys, with a total of 2,200 observations from the years 2011 and 2014 as explained in more details here: 4.1 Arab Barometer II (2011) The first dataset I will use for my analysis is the Arab Barometer II (ABII from now on). This dataset includes observations obtained through face-to-face interviews with 1,200 individuals in all regions of Yemen. The survey was conducted in February 2011, with observations disaggregated through the following categories: regions, gender, and age. The levels of political participation will be measured by variables for voting and membership in political parties. Other 18

27 control variables that are included in the analysis are trust in government; trust in political parties; interest in politics; trust in fairness of elections; and education. 4.2 World Values Survey- Wave 6 (2014) The second dataset I will utilize for my analysis is the World Values Survey Wave 6 (WVS6 from now on). This dataset includes observations obtained through face-to-face interviews with 1,000 individuals in all regions of Yemen, except Saada. The survey was conducted in February 2014, and the observations were disaggregated by regions, gender and age. In the same manner as with the ABII dataset, the WVS6 will help me explore the relationship between political participation and gender, age and region of residence. The levels of political participation will be measured by variables for voting and membership in political parties. Other control variables that will be included in the analysis are trust in government; trust in political parties; interest in politics; trust in fairness of elections; education; and in the case of the WVS6, income scales will also be included in the analysis. It is worth noting that the WVS6 does not include observations for Saada, which impacts my fifth and sixth hypotheses regarding voting patterns and membership in political parties in South Yemen and Saada. If the results generated from my analysis of the ABII data allow for inferential statistics for Saada, then it will suffice, as the sample set is large enough to allow for such conclusions. 5. EMPIRICAL MODELS In this thesis, the two datasets are analyzed separately, where Ordinary Least Squares (OLS) regressions are run (and Logistics regressions are attached in Appendix 1-4). The difference between the two regression analyses is that the ABII does not include a vector for income levels, 19

28 while the WVS6 includes a vector for income scale. Therefore the ABII analysis controls for variables of political culture and education, while the WVS6 analysis controls for variables of political culture, education and income. This study tests whether civic engagement trends in Yemen follow the findings of Yuchman-Ya ar and Alkalay (2010) in terms of identifying determinants of support for democracy in the Muslim World. The following regression models control for other factors, such as political interest, trust in government, and political parties: CE! = β! + β! W! + β! X! + ε! In the above equation, CE represents two different outcomes in the form of dummy variables (i) equal to one if the individual voted in the last election, zero if he/she did not, and (ii) equal to one if the person is a member of a political party, zero if he/she is not. Xi is a vector of controls to explore differences, including gender, age and region of residence. The Wi variable represents the following controls of interest: Trust Gov [a vector of dummy variables for levels of trust in government] Trust Pol Parties [a vector of dummy variables for levels of trust in political parties] Elections Fair [a dummy variable for whether the respondent believes the elections are fair or not fair] Pol Interest [a vector of dummy variables for levels of interest in politics] 20

29 5.1 Dependent Variables a. Voting In the ABII dataset, the question regarding voting in past elections was asked; in the WVS6, the survey asked whether the respondent always, usually or never votes. As expected, 72 percent of the WVS6 respondents reported voting as opposed to 64 percent in the ABII dataset. This surge is the result of an increased awareness of the political, economic and democratic influences in Yemen since the regime change in As mentioned above, the ABII was conducted at the beginning of the 2011 uprising, while the WVS6 was conducted in 2014, after the presidential elections of Table 1 provides a detailed comparative overview of voting patterns between the two datasets analyzed in this thesis. Table 1. How Many People Reported Voting Arab Barometer 2 World Values Survey 6 Voted 64% 72% Did not Vote 36% 28% Proportional Men Vote 53% 51% Proportional Women Vote 47% 49% Number of Respondents 1, b. Membership in a political party On the other hand, the descriptive statistics on membership in political parties does not correspond to the expected results. Table 2 reveals that 31 percent of respondents of the ABII survey indicate they are members of political parties. In the WVS6, 3 years later, the number of respondents who are members of political parties declines to 20 percent. This may be explained by the growing political conflict and stalemate blamed on political parties infighting for power at the expense of public interest and stability. 21

30 Table 2. How Many People Reported Being a Member of a Political Party Arab Barometer 2 World Values Survey 6 Member 31% 20% Not a member 69% 80% Proportional Men Member 64% 61% Proportional Women Member 36% 39% Number of Respondents 1,147 1, Independent variables a. Gender In both datasets, the proportion of men to women respondents is around 49.9 percent. I will use the female category as the reference variable to test whether female voting patterns are different from male voting patterns; and also whether female membership in political parties patterns are different from male membership patterns. b. Age The average age of respondents of the ABII survey is 32 years and the average age of the WVS6 respondents is 36 years. To account for any skewness caused by the few observations whose age is above 70 years, the age variable in both datasets was grouped as follows: 18 to 25, 26 to 35, 36 to 50, and 51 years and above. In the regressions analyses, the 18 to 25 group of respondents is used as the reference variable to which other groups characteristics are compared. c. Region of residence In the ABII dataset, surveys were conducted in the following regions: Ibb, Abyan, Amanat Al Asimah, AlBaydah, Taiz, AlJawf, Hajjah, AlHodeidah, Hadramut, Dhamar, Shabwa, Saada, Sanaa, Aden, Lahaj, Marib, AlMehweet, AlMahra, Amran, AlDalee, Raymah. The WVS6 included all these cities except for Saada and AlMahra. For this thesis, cities were combined into regions that were announced by President Abd Rabbuh Mansur Hadi in February 2014, upon the 22

31 completion of the NDC, as the regions of the new federal system (Al-Sakkaf 2014). For my analysis, I combined the southern regions together into a South Yemen variable, while extracting observations collected in Saada from the Azal region data. The reason for this combination and exclusion is to reflect what the Government of Yemen, political parties and civil societies recognize as special regions that were negatively impacted by past discriminatory policies. The NDC has within its working groups, two distinct groups working on the Southern Issue and Saada Issue. No other regions/cities had similar separate working groups in the NDC. Table 3. Distribution of Respondents by Region Arab Barometer 2 World Values Survey 6 Tehama Sheba Azal (excluding Saada) Aljanad Saada 40 0 South Yemen Total Control Variables In order to determine whether other factors impact voting and membership in political parties patterns, I included the following control variables into the analysis: a. Trust in government In both datasets, the majority of respondents said they trust the government to a limited extent or do not trust the government at all. Refer to Table 4 for details. This control variable is important in order to assess whether individuals who trust the government to a great extent have different civic engagement characteristics as opposed to those who do not trust the 23

32 government at all, which in this analysis was used as the reference category. The important observation is that in 2011 (ABII), the number of individuals who do not trust the government at all was over 39 percent, while in 2014 (WVS6) the number decreased to 32 percent. This change may be explained by the Yemeni citizens sustained support to the transitional government whose main responsibility is to act as a shield for civil conflicts. b. Trust in political parties In both datasets, the majority of respondents said they do not trust political parties at all. Refer to Table 4 for details. Contrary to the change that occurred to the trust in government variable, the number of observations who do not trust political parties at all in 2011 (ABII) was 46 percent, while in 2014 (WVS6) the number of individuals who do not trust political parties at all was 56 percent. This shift may be explained by the competition between political parties to guarantee a larger share of official positions at the expenses of the public and without clear communication strategies to reach their rank and file. During the transitional period, political parties acted alone when signing the GCC-brokered deal to transfer power. The deal failed to include clauses for more government accountability as the youth demanded and entailed a controversial immunity clause, exempting the former regime from alleged corruption and abuses of office committed in the past. Table 4. How much do Respondents Trust the Following Institutions Arab Barometer 2 World Values Survey 6 Government Political Parties Government Political Parties To a great extent 7.8% 6.6% 9.8% 1.6% To a medium extent 21.6% 20.6% 19.9% 8.0% To a limited extent 31.3% 26.6% 38.2% 34.1% I absolutely do not trust it 39.3% 46.2% 32% 56.3% Total 100% 100% 100% 100% Number of Respondents 1,148 1,

33 c. Perception on fairness of elections In both datasets, the majority of respondents still think that elections are not fair, but there is a shift in how elections are perceived following the regime change of 2011 and the 2012 presidential elections. Although there was only one candidate running, both Yemenis and international observers viewed this election as a referendum. Many believed the election was instrumental in saving the country from an imminent civil war, as it was and still is the case in some of the Arab Spring countries, namely Libya and Syria. Refer to Table 5 for more details. Table 5: Fair Elections? Arab Barometer 2 World Values Survey 6 Yes 23% 33% No 77% 67% Total 100% 100% Number of Respondents d. Interest in politics This control variable shows the levels of interest in politics expressed by respondents in the two datasets. The interesting distinction is that in 2011 (ABII), individuals who are very to somewhat interested in politics represented 37 percent of the observations. In 2014 (WVS6) on the other hand, the number of individuals who are very to somewhat interested in politics was 49 percent. This type of increase is expected after a major political change, such as regime change and the recent presidential elections that were held in Refer to Table 6 for more details. 25

34 Table 6: Interest in Politics Arab Barometer 2 World Values Survey 6 Very interested 14% 9% Somewhat Interested 23% 40% Not very Interested 34% 25% Not at all Interested 29% 26% Total 100% 100% Number of Respondents 1, e. Education attainment This variable measures the educational attainment of respondents There is a difference in the characteristics of individuals interviewed in each dataset: the ABII observations have higher educational attainment with 36 percent of respondents having a university degree, while 40 percent of WVS6 respondents are reported to be illiterate. Table 7 shows a comparison between both datasets based on education attainment levels. In the regressions, the reference category is illiterate. Table 7: Distribution of Education Attainment Arab Barometer 2 World Values Survey 6 Illiterate 10% 39.9% Basic 17% 23.0% Vocational 14% 10.6% Secondary 23% 11.6% University 36% 14.9% Total 100% 100% Number of Respondents 1, f. Income scale (for World Values Survey Wave 6 dataset only) This variable is only available for the analysis performed on the WVS6 dataset since over 50 percent of the observations are missing in the ABII data set. 26