Partisan Activism and Access to Welfare in Lebanon

Abstract

How do welfare regimes function when state institutions are weak and ethnic or sectarian groups control access to basic services? This paper explores how people gain access to basic services in Lebanon, where sectarian political parties from all major religious communities are key providers of social assistance and services. Based on analyses of an original national survey (n= 1,911) as well as in-depth interviews with providers and other elites (n= 175) and beneficiaries of social programs (n= 135), I make two main empirical claims in the paper. First, political activism and a demonstrated commitment to a party are associated with access to social assistance; and second, higher levels of political activism may facilitate access to higher levels or quantities of aid, including food baskets and financial assistance for medical and educational costs. These arguments highlight how politics can mediate access to social assistance in direct ways and add new dimensions to scholarly debates about clientelism by focusing on contexts with politicized religious identities and by problematizing the actual goods and services exchanged.

Introduction

Declining or underdeveloped public welfare functions leave wide scope for non-state actors such as international or domestic non-governmental organizations (NGOs), religious charities, and even political parties and movements to supply basic social services. For political organizations and ethnic or religious charities in plural societies, service provision can be used as much to gain and reinforce political support or to maintain group boundaries as to fulfill commitments to social justice. The question of who benefits from non-state service provision by sectarian parties sheds light on how welfare regimes function when state institutions are weak and public provision is minimal—a condition common to many developing countries in Asia, Africa, the Middle East and parts of Central and South America.¹ Where the state is virtually absent or citizen social rights are effectively denied, some types of providers may use political considerations to shape access to basic services.

The basic observation that political factors guide the provision of welfare goods begs the question of how political groups decide whom to target with what quantities and types of welfare goods. What shapes the allocation of social assistance and what levels and forms of social aid do individuals and families receive? This paper explores these questions in Lebanon, where political organizations linked to ethnic and sectarian communities as well as other types of non-state providers (NSPs) play a key role in providing basic public goods and social services. Based on findings from an original national survey and in-depth interviews with elites and non-elites, I make two main claims. First, political activism and, specifically, demonstrated partisan loyalties facilitate access to services. Second, increased activism is associated with access to more comprehensive baskets of benefits, including social assistance in the form of food aid or cash as well as financial assistance for health care and schooling. In general, more encompassing packages of benefits call for more political activism in support of a party than required for more limited benefits. The next section situates these arguments in the contemporary literature on clientelism, which tends to overlook the nature of benefits distributed in clientelist exchanges. The third section provides a brief overview of the welfare regime in Lebanon, detailing the role that sectarian political parties play in the system. The fourth section presents the data and findings from an original national survey and in-depth interviews to illustrate some of the main points. The final section highlights areas for future research.

Clientelism, Modes of Politics, and Access to Welfare

Resources are limited and even the wealthiest non-state providers of social assistance cannot serve all nor can they offer the same level of support to all beneficiaries. Furthermore, parties may distribute benefits unevenly to lure in supporters, reward existing supporters, pursue both strategies simultaneously, or promote greater electoral turnout. Providers therefore establish formal and informal criteria for allocating benefits. Based largely on studies of Latin American countries, recent literature on clientelism centers on how patrons select clients, debating the degree to which parties favor “core” over “swing” voters (Diaz-Cayeros et al. 2007; Dixit and Londregan 1996; Stokes 2005) or aim to increase turnout by targeting “unmobilized” voters (Nichter 2008). I move beyond these debates by focusing on the nature of benefits distributed to clients and by suggesting that the overwhelming emphasis on electoral politics over other forms of political engagement helps to obscure the more enduring relationships forged between political patrons and their clients.

The substance of the clientelist exchange deserves more attention in part because it sheds light on the nature of the relationship between a political organization and citizens. Recent studies of clientelism tend to emphasize one-shot transactions for electoral purposes. But even for purely instrumentalist reasons, such as buying or securing the support of voters, parties may distribute more comprehensive packages of benefits or engage in longer-term relationships with voters.² Citizens are acutely aware of the motivations behind the distribution of social assistance by political parties and can easily become cynical about the experience. Furthermore, party officials and their neighborhood or village-level representatives often have more complex and longstanding ties with their constituents than a one-shot exchange around electoral contests would imply. For example, the ethnographic study of Auyero (2001) emphasizes the complicated linkages between the Peronist Party and its supporters in Argentina, suggesting that the relationship is far more protracted and intimate than other research on the same party might suggest (Nichter 2008; Stokes 2005).³ At the local level, party representatives and citizens are engaged in a complex and multifaceted webs of social ties with each other. This is all the more true in societies where ethnic or religious identities are politicized and constitute primary modes of social organization. In these contexts, ethnic or religious organizations tend to have enduring roots in local communities and citizens may look to the leaders of their respective communities for social protection, particularly when public welfare functions are not developed.

Moving beyond electoral considerations suggests additional reasons for exploring the nature of benefits and not only the profiles of beneficiaries. In countries like Lebanon, where democratic governance is only partially institutionalized and electoral politics are not the only mode of political competition, parties pursue a “dual game” of both electoral and non-electoral forms of political mobilization (Mainwaring 2003). Winning elections is an important goal to provide access to state resources, both material and symbolic. But parties also engage in non-electoral forms of politics, such as protests, demonstrations, riots, and, in some cases, the formation of militias.⁴ When the very rules of allocating power in the political system are contested—as is true in Lebanon and other non-consolidated quasi-democracies—these alternative modes of politics can signal party strength and constitute forms of pressure to support preferred changes to the system. Non-electoral forms of mobilization call for longer term, more comprehensive relationships of social protection between parties and more committed, core supporters, who can be relied on not just to go to the polls and mobilize broader electoral turnout but also to attend or lead party meetings, take part in peaceful demonstrations and protests, and even to participate in more high-risk forms of collective action such as riots or militia warfare.

If party representatives and citizens have more complex and protracted relationships around both electoral and non-electoral politics, then parties may distribute distinct levels and types of social assistance to different kinds of supporters. Put differently, access to diverse packages of social assistance entail distinct outlays of political activism. At a minimum, the most loyal supporters and activists are likely to receive higher levels of benefits. Furthermore, those who engage in the least risky shorter-term forms of demonstrated loyalty to a given party, such as voting in elections, are likely to receive the most limited rewards such as food baskets, cash, or financial aid for basic health care, which can be delivered widely and easily. The logistics of distributing food aid are relatively simple because food assistance is often provided out of mobile units that travel to targeted areas during circumscribed time periods, such as holidays or elections. Similarly, non-core supporters are likely to have access to small amounts of monetary assistance for medical care, particularly for routine visits and less serious conditions. In general, more minimal outlays of political activism should be associated with more narrow combinations of benefits or even just a single benefit. More comprehensive and more protracted benefits, such as combinations of all three types of assistance explored in this paper or long-term medical care for serious illnesses would entail greater and longer-term expressions of political loyalty.⁵

In short, in non-consolidated democracies and quasi-democracies, those who exhibit the greatest and most protracted loyalty to a party and engage in riskier forms of behavior such as participation in demonstrations or riots are likely to receive higher levels of social assistance and more iterated benefits. Below, I assess these claims using data from an original national survey as well as in-depth interviews on access to social assistance and political behavior. Before presenting the survey data and results, I contextualize the data with a brief overview of the Lebanese welfare regime.

Politics and Welfare in Lebanon

Sectarian Parties in the Lebanese Political System and Welfare Regime

The political system established at independence in 1943 institutionalized preexisting patterns of political sectarianism in Lebanon (Makdisi 2000). To this day, top government positions are allocated by sect while the electoral system also institutionalizes sectarian identity by allocating seats by sect at the district level according to formulas derived from the sectarian identities of registered voters.⁶

The structure of the political system as well as the political dominance of elite commercial and financial interests had repercussions for the post-independence Lebanese welfare regime, which involves minimal if any state intervention as well as heavy reliance on private, non-state actors. As a result, the welfare regime is highly fragmented and unregulated, providing ample opportunities for private actors to supply social services and take credit for state-sponsored benefits. Non-state actors, including political organizations and religious charities, both profit from and sustain the underdevelopment of public welfare functions.

The minimal direct state role in assuring social protection has long historical roots. During the Ottoman and colonial mandate periods, religious organizations such as churches, missionary groups, and Muslim organizations were important providers of social assistance and basic services (Fawaz 1994: 135; Hanssen 2005: 116).⁷ After Lebanon gained independence from the French in 1943, little progress was made in constructing national institutions of social protection until the presidency of Fu'ad Shihab (1958–1964). Under Shihab's tenure, al-Damaan al-Ijtima'i of the National Society Security Fund, a major source of health financing in Lebanon (Ammar 2003), was established and public infrastructure was expanded.

The Lebanese civil war (1975–1990) led to the partial and, for some programs, total breakdown of the few public social welfare institutions established in the 1960s. In part as a response to state failure, confessional groups in all of Lebanon's religious communities initiated or further developed their own social welfare programs while NGOs proliferated in the domain of social provision (Corm 1991; Hanf 1993; Harik 1994; Traboulsi 2007).

After the Ta'if Accords formally ended the war in 1989, some militias transformed themselves into political parties with associated networks of welfare agencies and social centers. Other political parties also launched their own social welfare wings. The experiences of Muslim and Christian parties and parties-cummilitias diverged in the post-war period. Unlike Sunni, Shi'i, and Druze organizations, the Christian militias did not reinvigorate their wartime welfare activities until after 2005 as a result of repression by Syria, which maintained forces in Lebanon until that year.⁸

The breakdown of health and educational institutions according to public and private affiliations illustrate the importance of non-state providers and, especially, sectarian parties in social provision in Lebanon. Religious charities and sectarian parties play a particularly critical role in primary health care. About 17% of medical centers and dispensaries are run by Christian charities and 11% by Muslim charities. Among political parties, Sunni and Shi'i political groups also run important health care networks, such as the Sunni Future Movement of the Hariri family as well as Shi'i parties such as Hezbollah and the Amal Movement. Sunni and Shi'i parties account for about 7% and 8% of all basic health care institutions, respectively. Although public institutions ostensibly account for about 9% of total primary health care centers and dispensaries; in reality, many of these institutions are linked to or controlled by different political factions. The importance of NSPs is also apparent in the educational sector. In 2006, about half of Lebanon's approximately 2,800 primary and secondary schools were privately run, with the Catholic School System accounting for the bulk of private schools and important school networks linked to political organizations. Although it impossible to quantify the provision of short-term material assistance, political parties also run programs to distribute food and other forms of short-term material aid, particularly during electoral cycles.

The role of sectarian organizations in social provision far exceeds the numbers of their centers. Beyond the direct provision of services in affiliated institutions, sectarian parties provide aid for health and educational needs, food and financial assistance, and act as intermediaries to facilitate access to citizen “entitlements.” For example, citizens with demonstrated need are legally entitled to state coverage of up to 85% of hospitalization costs for the treatment of certain serious conditions. In practice, most Lebanese cannot access this benefit, partly because of soaring budget deficits accruing to the Ministry of Public Health and partly because politicians and parties act as gatekeepers to this and other ostensibly public benefits. Sectarian party representatives use connections to reserve hospital beds and arrange government payment for the hospitalization costs of supporters.⁹ After paying the remaining 15% of fees not covered by the state, politicians and parties often claim credit for the full costs of care for hospital stays.

Identifying and Monitoring Supporters

Parties have elaborate ways of identifying and monitoring the participation of supporters in electoral and non-electoral forms of mobilization, enabling them to dole out or withhold rewards accordingly. Partisan loyalty, or demonstrated support for a party, is determined through multiple actions ranging on a continuum from minimal to more extensive expressions of commitment. Supporters may vote for party-appointed candidates, select the full slate of party or party alliance candidates,¹⁰ display posters and other visual materials signaling their political preferences, whip up voter turnout by shuttling others to the polls or otherwise serving in turnout efforts orchestrated by parties on election day, and regularly attend party meetings and events. In the realm of electoral politics, then, partisan loyalists are akin to “core” supporters as defined in the literature on clientelism, or “voters whose needs or electoral predispositions are well known to the party” (Stokes 2007) through a demonstrated record of commitment to a given party. As argued above, it is important to move beyond an exclusive focus on electoral politics to capture the dynamics of exchange between parties and citizens. The most active supporters may also participate in non-electoral forms of mobilization such as demonstrations, riots and even militia organizations.

How do parties monitor the political behavior of citizens? During electoral cycles Mafateeh Intikhabiyya—literally, “electoral keys” or prominent individuals who mobilize voters during elections in rural communities and urban neighborhoods—help to mobilize and monitor local residents. Parties have also devised virtually foolproof and, under the current electoral law, legal procedures for verifying that voters actually fulfill their verbal commitments. After obtaining permission from the Ministry of Interior, parties place representatives at each ballot box in polling stations, which contain separate boxes for men and women that are organized alphabetically by family name. As voters arrive, party workers stand next to the boxes and check off on their own lists whether or not a person has voted. Throughout the day, party workers receive calls on their mobile phones from community-based party workers to see who has and has not voted. If a family has not yet voted by a certain time, a party worker is dispatched to ensure that its members still intend to vote. When the polls close, party representatives are permitted to remain in the precincts to observe the ballot-counting process. In some stations, official election workers use an overhead projector to show the markings on each ballot to all observers. Parties profit from the opportunity to view the actual ballots by distributing distinct ballots to the heads of different families and recording which families received which ballots,¹¹ enabling party workers in the precinct to count up the numbers of ballots received from each family and to ensure that voters did not cross off candidates on the party list. The multiplicity of ballot boxes organized alphabetically by family name also facilitates the monitoring process.¹²

Lebanese political parties also monitor non-electoral forms of political participation. Linkages to community associations enable party officials to gather information on the political preferences and activities of local residents and to distribute material assistance to supporters on the grassroots level. Parties establish close ties with local elected officials such as mayors or mukhtars, or neighborhood and village-level elected officials, as well as with well-known individuals in the community because they are key sources of information on the behavior of local residents. In addition, parties set up their own community level networks: party cadres sometimes go from house-to-house in the urban and rural areas where they are based to “register” families for the ostensible purpose of ascertaining socioeconomic need. Party representatives and their allies come from the communities in their jurisdiction and therefore are well placed to observe local residents, whose political sympathies are well known through information transmitted via local social networks and visual displays of partisanship in the form of posters, stickers, the color of clothing and personal effects, and other symbolic forms on homes, cars and accessories.¹³ The next section describes the survey data used to test the relationship between political behavior and access to social assistance in Lebanon.

Survey Data and Variables

Data from interviews with party representatives, development consultants, government officials, and “ordinary” citizens provide compelling evidence that sectarian organizations assess political behavior in determining whether and to what degree individuals and their families should receive services. But most party representatives deny that their own organizations discriminate according to the political behavior of beneficiaries.¹⁴ To evaluate the linkage between political behavior and welfare allocation therefore requires additional data beyond interviews with the providers themselves. The analysis relies on an original national survey conducted in Lebanon in April and May 2008, supplemented by data from in-depth interviews with party officials, welfare agency employees, government officials, local development consultants, and beneficiaries of social service programs to explore how political behavior is associated with access to social assistance.

The survey instrument includes questions on access to assistance for primary and secondary healthcare, financial aid for health and educational services, and food and other forms of short-term material benefits; religious identity and observance; political preferences and participation in political and religious organizations; and standard demographic questions. After the interview, survey enumerators also noted any political and/or religious posters or other materials displayed in the respondents' homes as a supplementary measure of political and religious affiliations (Corstange 2007).

Based on a nationally representative sample of 2,859 households across Lebanon,¹⁵ the survey sample captures a broad cross-section of the Lebanese population in terms of gender (53.6% female; 46.4% male), age,¹⁶ socioeconomic status,¹⁷ sect, and political affiliation. The last official national census was held in 1932 but, given political sensitivities, it is unlikely that a national census will be held soon. The Ministry of Interior provides data on the sectarian identities of registered voters at the local level based on information supplied by mukhtars.¹⁸ While there is no official data on the current sectarian breakdown of actual residents, it is broadly accepted that Christians from diverse sects now constitute just under 40% while Muslims account for about 60% of the population (CIA World Factbook 2008; US Department of State 2008). The survey sample captures the full array of sectarian groups, albeit with slight overrepresentation of Christians and underrepresentation of Shi'i and, especially, Sunni Muslims.¹⁹

The respondents represent a diversity of partisan alignment with the overwhelming majority reporting support for the main sectarian parties.20 I include measures of both party membership and support because partisanship in Lebanon is generally expressed in terms of “support.” Only party cadres hold official membership in political parties and, as a result, few citizens are actual members of parties.

Data from the survey show that social assistance from non-state organizations is substantial, with almost 52% of respondents reporting that they received some form of aid. Overall, 26% benefited from food and cash handouts of which 63% reported that a political organization provided these items, 32.5% received financial aid for medical care, 27% of respondents with school-age children reported that they received financial aid for school fees, and 25% reported that they received health services from a non-governmental organization. These figures undoubtedly underreport the degree to which respondents received social assistance and, especially, assistance received from or arranged by political parties for several reasons. First, social norms dictate that most families are reluctant to admit to financial need. Second, it is surprisingly difficult to gather information on the affiliations of service providers—political, religious or otherwise—from interviewees. As patterns of non-response to survey questions about provider characteristics suggest,²¹ many respondents were either reluctant to reveal that they received services from political or religious institutions or unaware of provider affiliations. Third, by necessity, the key measures of the dependent variable—access to welfare—do not include services supplied directly by party agencies or charitable organizations but rather focus on financial assistance for a limited array of third-party services. Sensitivities regarding the political and religious affiliations of welfare providers that offer such aid may have deterred respondents from admitting that they had sought social assistance from such organizations. Fourth, the probable underrepresentation of Muslims in the sample suggests that, if anything, the data underestimate access to social assistance by political or religious organizations. In general, welfare institutions are more centralized and more directly linked to political parties in Muslim rather than Christian communities.

Variables

The dependent variables focus on access to welfare benefits, including food and material aid as well as financial support for medical services and schooling. (The appendix lists the questions that provide the data for the key dependent variables.) This array of services captures important components of the social programs and activities of political parties. Parties are renowned for distributing food, household items, and even cash during holidays and, especially, during elections. Financial assistance and discounted medical services and school fees are also central to the welfare activities of political parties, which provide financial assistance or arrange reduced prices for supporters at their own networks of clinics and schools or at institutions run by private, for-profit organizations, or NGOs with sympathetic administrators. Parties also use their influence to facilitate expedited benefits from public programs or pay out-of-pocket fees for services largely covered by state agencies.

I run separate models with two different specifications of the dependent variable. The first dependent variable, WELFARE1, is measured dichotomously with values for no aid received (0) and any type of aid received (1). The second measure of the dependent variable is measured on a four-point ordinal scale and its values range from no aid received (0), one type of aid received (i.e., food, medical or educational assistance) (1), two types of aid received (2), and all types of aid received (3). Thus, the values of the dependent variable range from no aid or the most basic benefit package, such as food assistance only, to the most extensive package of benefits, including all types of benefits covered in the survey. Table 1 depicts the coding and distribution of the two measures of the dependent variable.

Table 1. Coding and distribution of the measures of the dependent variable (WELFARE1 and WELFARE2).

Value	WELFARE1		WELFARE2		Mean PAI
	Frequency	Percent	Frequency	Percent
0=No aid	907	47.41			0.675
1=Any aid	1,006	52.59			0.941
0=No aid			922	48.2	0.675
1=1 type of aid			603	31.5	0.797
2=2 types of aid			326	17	1.141
3=All types of aid			62	3.2	1.200
Total		100		100

As Table 1 shows, over half of the sample received some kind of social assistance. The majority of respondents only received a single benefit. Only about 3% of the respondents reported that they received all three types of benefits.²²

The main independent variable of interest, the Political Activity Index (PAI), is a continuous variable which was constructed through a factor analysis of responses to questions about different forms of political participation.²³ (The appendix provides the questions used to construct the index and the factor loadings.) The survey captures diverse forms of political participation, including support for or membership in political parties, voting behavior, volunteering for party organizations, attending party meetings, and the display of partisan materials such as posters or symbols at the place of residence.²⁴ In the sample population, PAI ranges from 0 to 3.74, with a mean of 0.81 and standard deviation of 0.91. The Cronbach's alpha coefficient for PAI is 0.71.

Table 2 presents the levels of political activism associated with each component of the PAI index associated with different ranges of the variable.²⁵

Table 2. Average levels of PAI per mode of participation by different levels of political activism.

Question	Level of political activism
	Low PAI≤0.6	Medium 0.6<PAI≤1	High PAI>1
Overall PAI	0.340	0.820	2.373
Attended party meetings	0	0	0.652
Frequency of attending party meetings	0	0	3.191
Volunteered in 2005 national elections	0.043	0.171	0.551
Voted in 2005 elections	0.558	0.993	0.858
Voted for full party list	0.279	0.882	0.723
Membership in party	0	0.007	0.454
Support for party	0.198	0.934	0.916
Displays political posters/materials at home	0.058	0.171	0.578
Number of observations	1,247	287	379

For the majority of respondents (1,247 out of 1,911), who reported low levels of PAI (ranging from 0 to .6 out of 3.81 on the index), the primary form of political participation was to cast a ballot in national elections. Low PAI respondents were most apt to vote in the 2005 elections (0.558), with some selecting the full list put forth by their preferred party (0.279). Although some expressed support for a party (0.198), none claimed to be members of a party. Few low PAI respondents displayed partisan posters in their homes (0.058) or volunteered in the elections (0.043) and none reported that they attend party meetings. Among respondents with moderate levels of PAI (ranging from 0.6 to 1), levels of voting in the 2005 elections (0.993) and selecting the full party list (0.882) were significantly higher than reported by those with lower levels of activism. This group was also far more likely to express partisan support (0.934), and a handful even claimed to be members of a party (0.007). Respondents with moderate PAI also exhibited higher propensities to display political posters in their homes (0.171) and to volunteer for a party in the elections (0.171). The most politically active respondents (those with PAI scores of more than 1) engaged in the full range of forms of political engagement in the index, including participation in party meetings as well as forms of electoral participation. Reported attendance at partisan meetings was relatively high in this group (0.652), which entails a greater and more continuous commitment to a party than going to a polling station to cast a ballot or even volunteering as a poll worker during electoral cycles. As would be expected, high PAI respondents reported relatively high levels of membership in a party (0.454) and were most likely to display partisan materials at their homes (0.578).

The average PAI score was higher for men (0.947) than for women (0.702); but in practice, women may receive more credit for activism than their lower scores suggest. PAI is measured at the individual level, yet in Lebanon political participation often occurs at the level of households or even extended families. Women and youth often benefit from political activism carried out by husbands, fathers or other male relatives. As a result, the PAI measure likely underreports the recognition that political parties accord to these demographic groups for political activism. For example, a father may be known as a core supporter of a political party thanks to his attendance at political meetings or community outreach on behalf of the organization. Family members—and not just the father himself—will be perceived as core supporters as well and will likely receive direct benefits and/or discounted services arranged by the party.²⁶

The analyses of the survey data include variables to control for socioeconomic status, age, sex, religious involvement (as measured by participation in religious groups), self-reported piety, whether or not children reside in the household, sectarian identification and the degree of religious fractionalization in the respondent's place of residence, each of which might independently affect access to welfare. Low-income individuals are more likely to seek and obtain welfare goods, yet survey respondents are often reluctant to reveal household income, particularly where tax evasion is widespread and social norms deter people from revealing hardship. I therefore include three different measures of socioeconomic status, including a measure of total household income, an index of ownership of basic household items, and an index indicating whether the respondent faced financial or material hardship in the past year.²⁷ It is also important to include multiple measures of socioeconomic status because most providers, including political parties, claim that they allocate services according to need rather than political or sectarian identity. In addition, families with children may require more social assistance than others and of course are the only subset of the sample likely to seek financial assistance for schooling, if needed. A dummy variable measures the presence of school-age children residing in the respondent's household.

Levels of political activism may also differ by sect. To control for this potential source of variation in access to services, I include dummy variables for sectarian affiliation, including Shi'a and Sunni Muslim with Christian as the reference category.²⁸ In the Lebanese context, different sectarian communities have distinct legacies of welfare provision. A history of Christian missionary social provision dating back to the eighteenth century fostered robust church-based providers of material assistance, health care and, especially, schooling. The Sunni community, too, boasts longstanding welfare institutions dating back to the Ottoman period. Since the 1970s and 1980s, Shi'i political and religious organizations have developed extensive and comparatively centralized welfare programs, prompting some Christian respondents to complain that the Shi'a enjoy privileged access to social services. Furthermore, as noted above, in the post-war period Muslim parties have developed and maintained more articulated welfare networks than their Christian counterparts. Table 3 shows that average levels of socioeconomic status varies by sect and partisanship.

Table 3. Mean socioeconomic status by sectarian identity and Partisan affiliation.

Group		Income	Ownership	Need
Sect	Christian	2.260	4.744	0.891
	Shi'a	1.868	4.222	0.797
	Sunni	1.811	4.345	1.072
	Average	1.947	4.433	0.971
Party	Christian party	2.301	4.727	0.947
	PSP	1.844	4.313	1.117
	Hezbollah	1.977	4.243	0.827
	Amal movement	1.769	4.416	0.718
	Future movement	1.803	4.394	1.068
	Average	1.939	4.419	0.935

The variation in the sample corresponds with general expectations about the population: Christian respondents generally have higher socioeconomic status than Shi'a, Sunnis, and Druze, although Shi'a report even lower measures of financial hardship which may arise from the community's comparatively strong welfare institutions that have developed in recent decades.

Religiosity might also facilitate access to welfare provided by non-state organizations, many of which are linked to religious institutions and may favor those who seem more pious or at least demonstrate a greater commitment to their place of worship by participating in religious events outside of regular services and celebrations. Some religious institutions also have informal ties to political parties, which may use their influence to obtain favorable treatment for supporters in religious charities. I therefore include variables measuring self-reported piety as well as religious commitment, or participation in religious meetings and organizations organized by the respondent's religious institution.²⁹ Religious commitment and piety have separate measures because individuals may be privately observant without taking part in events organized by their places of worship or religious leaders and communities.

Finally, I include a measure of religious fractionalization in the respondent's place of residence based on voter registration records.³⁰ An extensive literature in the social sciences holds that ethnic and religious heterogeneity is associated with decreased public goods provision (Alesina et al. 1999; Easterly and Levine 1997; Habyarimana et al. 2007). To account for the ways in which community characteristics shape the respondent's access to services, the variable fractionalization measures the degree of ethno-religious diversity of the population at the neighborhood or village level.³¹ The variable ranges from no fractionalization or perfect homogeneity (0) to extreme heterogeneity (100).

Results

Political Activism and Access to Welfare

A bivariate comparison of means provides an initial assessment of the relationship between political activity and access to welfare. I first divided the sample into two parts to create categories of low political activity individuals, or people who may or may not vote in national elections but otherwise do not participate in politics, and high political activity individuals, who vote in national elections for full party lists and openly and repeatedly display their partisanship through participation in political groups associated with their preferred party. The results of two-tailed t tests show that respondents who reported higher political activity had greater access to different types of welfare goods. More specifically, politically active individuals had more access to food aid as well as financial assistance for medical care and schooling than their less politically active counterparts (differences significant at p<0.05). An analysis of variance test further supports the relationship between political activity and access to welfare (F=30.09; p=0.00).

Many other factors are likely to shape access to welfare such as socioeconomic status, gender, and age (depending on the type of service received). To control for the independent effects of these and other factors on access to social assistance, I estimate a series of logistic and ordered logistic regression analyses. Table 4 reports the results of all of the models.

Table 4. Political activity and access to welfare: logistic and ordered logistic regression estimates.

Variables	Model 1 (WELFARE1)	Model 2 (WELFARE1)	Model 3 (WELFARE2)	Model 4 (WELFARE2)
PAI	0.382*** (0.070)	0.385*** (0.077)	0.428*** (0.068)	0.429*** (0.0671)
Age	−0.012 (0.043)	0.008 (0.048)	0.012 (0.041)	0.0383 (0.0442)
Sex (1=female)	0.349*** (0.120)	0.450*** (0.125)	0.321*** (0.116)	0.446*** (0.114)
Income	0.013 (0.075)	0.045 (0.076)	−0.001 (0.992)	0.0372 (0.0735)
Own	−0.435*** (0.093)	−0.290*** (0.089)	−0.449*** (0.089)	−0.309*** (0.0842)
Need	0.144 (0.118)	0.173 (0.108)	0.167 (0.121)	0.190** (0.0927)
Religious participation	0.070 (0.069)	0.121* (0.069)	0.091 (0.058)	0.150** (0.0601)
Piety	−0.027 (0.063)	−0.009 (0.064)	−0.052 (0.063)	−0.0419 (0.0633)
Child	0.696*** (0.194)	0.561*** (0.200)	0.864 (0.175)	0.742*** (0.186)
Fractionalization	−0.004 (0.004)	−0.002 (0.005)	−0.003 (0.005)	−0.00244 (0.00441)
Shi'a		0.365 (0.241)		0.398* (0.233)
Sunni		0.941*** (0.282)		1.144*** (0.266)
Constant	−0.757** (0.375)	−1.221*** (0.428)	0.883** (0.363)	1.382*** (0.396)
Observations	1,641	1,418	1,641	1,418
Pseudo R2	0.0517	0.0674	0.0451	0.0630

In model 3, all values of the dependent variable are compared to non-receipt of welfare. Standard errors in parentheses

^*p≤0.1,

^**p≤0.05,

^***p≤0.01

Model 1 uses the binary measure of the dependent variable, WELFARE1, and includes basic demographic control variables such as age, sex, income, ownership, financial need, participation in religious associations, piety, the presence of children in the household, and religious fractionalization in the respondent's place of residence. The results show that ownership of fewer household items is associated with greater access to welfare. These findings are as expected given that low-income families are the most obvious candidates for social assistance and the poor and needy are most likely to be targeted for clientelist exchange (Calvo and Murillo 2004).³² Furthermore, non-state providers, including those linked to political parties uniformly emphasize the importance of socioeconomic status in the eligibility criteria for their services.

The results also show that women are more likely to receive social assistance, a logical finding given that women are more likely than male household members to take charge of family needs, accompany children to medical visits, and seek assistance from welfare agencies in Lebanon and in many societies. Respondents with children residing in their homes were more likely to receive social assistance undoubtedly because their families have greater need for aid and are the only segment of the sample that would seek financial assistance for schooling.

In this basic model, which does not account for sectarian identity, neither participation in religious associations nor piety affect access to welfare, even though religious organizations play an important role in the Lebanese social safety net. Contrary to many studies of ethnic politics and public goods provision, which would predict a negative relationship between heterogeneity and access to welfare, religious fractionalization in the respondent's community of residence exhibits no association with the receipt of social assistance. This non-finding may arise because of measurement issues, specifically the fact that the variable captures fractionalization based on voter registration records and therefore does not always reflect actual residential patterns, or because of the types of social assistance measured in the survey, which do not fully correspond to the types of public goods usually captured in this literature such as infrastructure or aggregate social spending.

Most importantly, the findings suggest that political activity is strongly associated with the likelihood of receiving social assistance. An individual who actively supports a particular party and participates in political demonstrations and meetings is more likely to receive food assistance as well as financial aid for healthcare and, if applicable, schooling for her children than a person who is not politically active. Political activism gains access to welfare goods, even after accounting for socioeconomic need, religious participation or personal piety, and other factors noted above.

Model 2 shows that political activity retains its strong association with access to welfare even after controlling for sectarian identity. Again, in this expanded model, individuals with fewer household assets also have increased access to welfare goods, as expected. After accounting for sectarian identity, religious commitment, or participation in religious associations and meetings, is associated with receiving more social assistance while piety is still not significantly linked to access to welfare. This finding suggests that religious activism—but not the strength of one's personal commitment to her religion—enables an individual to benefit from more social assistance. In a context where religious associations are important welfare providers, participation in religious organizations and events beyond standard religious holidays and festivals can enable people to secure or at least supplement household welfare needs. Furthermore, because religious associations are often (although not always) informally linked to the major sectarian parties, participation in religious groups can also constitute a de facto form of political participation.

Finally, Sunni religious identity is strongly associated with access to welfare. An analysis of the dynamics of social provision in distinct religious communities is beyond the scope of this article; however, this finding likely reflects the relationship between the predominantly Sunni Future Movement and Sunni citizens.³³ Many of the components of the PAI involve activities related to electoral contests and the Future Movement party machine is renowned for distributing social assistance largely, although not exclusively, to Sunnis during electoral cycles.

To assess the substantive relationship between political activism and access to welfare, I calculate the predicted probability of receiving social assistance for a man in his forties with average levels of income, ownership of household items, financial need, children residing in the household, religious participation, piety, and community level religious fractionalization.³⁴ The man's probability of receiving benefits rises with political activism. For example, he has a 41% probability of receiving benefits at the minimum level of activism and nearly a 60% probability at a mid-level of PAI. At the highest level of activism, his chance of receiving social assistance jumps to 73%.

Figure 1 depicts the predicted probabilities that the man will receive benefits and includes corresponding point estimates and confidence intervals.

Fig. 1.

Predicted probabilities of receiving social assistance by levels of political activism (model 1). Reference category: 40 to 50-year-old man with average values of income and ownership characteristics, financial need, children in household, religious participation, piety, and community level religious fractionalization

In Fig. 1, the x-axis shows gradations of political activism, ranging from no or minimal activism to the highest level of activism (again, PAI scores range from 0 to 3.74, with a mean of 0.81 and standard deviation of 0.91), and the y-axis depicts the man's probabilities of receiving assistance. The figure confirms that increased levels of activism are associated with increased probabilities of receiving aid, particularly with a rise in activism by two standard deviations or more from the minimum level.

These findings are consistent with data from in-depth interviews with a broad range of informants. Elite interviews with local civil society organizations as well as providers linked to religious and political organizations attest that politics is critical in shaping access to social benefits, even if socioeconomic need also influences eligibility criteria. The director of a local NGO commented, “The main link between politicians and constituents is service provision. This occurs in many different ways: Through their own institutions, through their political parties, through the direct distribution of money and by siphoning off public services to private ends.”³⁵ A hospital worker in the predominantly Shi'i city of Nabatieh in South Lebanon commented on the politics of access to health services at the Hezbollah-run hospital in her city. She noted, “Ninety percent of the patients who go to the Hezbollah Ragheb Harb Hospital in Nabatiyyeh are in the party. If you don't have papers from Hezbollah or connections to it, then you don't get help from Hezbollah and you go to the Nejdeh hospital instead.”³⁶

Citizens also point to the ways in which partisan loyalty facilitates access to social assistance. A Maronite Christian man in his early thirties who resides in the predominantly Shi'i Muslim city of Tyre in South Lebanon described how politics seem to influence Hezbollah's distribution of social benefits in his community. Although he stressed that Hezbollah is not corrupt and provides extensive assistance to the community, the informant claimed that it favors its own supporters and “makes life difficult” for those working against them in Tyre.³⁷ A Shi'i man from the South corroborated this assessment, noting that a sheikh affiliated with Hezbollah, who is in charge of dispensing social assistance in his region, met with him and concluded that the respondent did not sufficiently support the party nor was he sufficiently outwardly pious to merit access to benefits. The informant claimed that he was a long-time employee of another Shi'i organization, the Imam Musa al-Sadr Foundation, and therefore the Hezbollah official did not regard him as a supporter of the party.³⁸

These claims by ordinary citizens do not only apply to Hezbollah. Other respondents emphasize that partisan loyalty plays a key role in obtaining privileged access to benefits provided by the Hariri Foundation, the social wing of the predominantly Sunni Future Movement. A Sunni woman from the Future Movement stronghold of Aisha Bakkar in West Beirut commented, “In addition to merit, connections help in getting children enrolled in good schools, such as those of the Hariri Foundation.” Sitting in her living room, which was plastered with posters and photographs of Rafiq al-Hariri as well as pins and scarves with Future Movement slogans, a middle-aged Sunni woman in the Hamra neighborhood of West Beirut noted that representatives from the Hariri Foundation registered her family in their database many years ago in order to receive social assistance. Her family members, who are known as core supporters of the party, frequent the Hariri Foundation's health institutions and benefit from additional material aid from the Foundation on a regular basis.³⁹

Political Activism and Levels of Social Assistance

The previous discussion shows that political activism is associated with the receipt of social assistance. In this section, I assess whether different levels of political activism are linked to distinct quantities of welfare, which would indicate that more active supporters receive larger amounts of assistance. Model 3 evaluates this proposition.

In model 3, the dependent variable measures how many benefits, if any, the respondent received (WELFARE2). The model, which uses an ordered logistic regression analysis, includes basic demographic control variables as in model 1 and the same results obtain, notably that women are more likely than men to receive benefits and respondents with higher socioeconomic status as measured by ownership of household items are less likely to receive welfare than those with fewer assets. Again, in model 3, the key variable of interest, PAI, is strongly associated with the receipt of social assistance.

Model 4 includes control variables for sectarian identity and similar results obtained, with a strong linkage between political activism and access to welfare. As in model 2, Sunni religious identity is positively associated with the receipt of welfare. Interestingly, with the more fine-grained measure of access to welfare used in model 4, Shi'a identity exhibits a weak positive association with access to welfare, indicating that Shi'a are more likely than Christians to obtain social benefits.⁴⁰ In addition, in model 4, financial need exhibits a positive relationship with access to welfare, which corresponds with the expectation that poorer families are more likely to receive charitable benefits, and religious participation has a stronger association with the receipt of benefits than in model 2.

To explore the substantive relationship between political activism and the receipt of different quantities of welfare, I calculate the predicted probability of receiving social assistance for a man in his forties with average values of income, ownership of household items, financial need, children in the household, participation, piety, and community level religious diversity.⁴¹ Figure 2 illustrates the predicted probability that the man will receive different quantities of benefits at varied levels of political activism and includes corresponding point estimates and confidence intervals.

Fig. 2.

Predicted probabilities of receiving different levels of benefits by levels of political activism (model 3). Reference category: 40 to 50-year-old man with average values of income and ownership characteristics, financial need, children in household, religious participation, piety, and community level religious fractionalization

In Fig. 2, the x-axis shows gradations of political activism across the four possible levels of benefits, which range from none to all three types of social assistance captured in the survey. The y-axis depicts the man's probabilities of receiving benefits.

The figure confirms that higher levels of activism are associated with increased probabilities of receiving a given level of benefits. For example, the man's probability of receiving two types of benefits grows from 12% to 35% when his level of political activism jumps from the lowest to the highest level. An increase from the minimum to the maximum level of activism is associated with a fourfold rise in his probability of receiving all three types of benefits. Conversely, the likelihood of receiving no benefits drops as political activism surges: The man has a 59% probability of receiving no benefits at the lowest level of activism and a 24% probability of no benefits with the highest levels of activism. The probability of receiving a single benefit presents an exception to these general trends, as the overlapping confidence intervals for different values of activism indicate. In this instance, an uptick in political activism does not necessarily lead to an increased probability of receiving a single benefit. This apparent anomaly may arise because this relatively low level of social assistance—a single benefit—may not require a substantial outlay of political activism.

Interviews with elites and non-elites support the findings from model 3. Respondents attest that different categories of supporters receive different combinations of services—or access to similar services on more favorable terms—as a result of their political behavior. Commenting on Hezbollah's health institutions, a staff member at a local NGO which has offices near the party's welfare agencies in Beirut and the South noted, “They will not throw someone out on the street if he asks for services and is not a supporter. Instead, they offer better discounts to their own.”⁴² For some of its welfare programs, Hezbollah has explicit criteria favoring core activists. A Red Cross official noted, “[The party] has specific programs that target its supporters. For example, there is the martyrs program.”⁴³ The Mu'assisat al-Shahid, or the Hezbollah assistance program for the relatives of fighters killed in action, offers comprehensive benefits to eligible families, including assistance for health care, schooling, material needs and other benefits.

Other parties are less transparent about the degree to which political criteria shape access to their social programs but both elite and non-elite interviewees claim that all political parties account for political behavior in distributing its benefits. A Lebanese journalist who works for an international press agency—and whose family members are fervent supporters of the predominantly Sunni Future Movement—emphasized the privileged access to benefits for core supporters of the party from its affiliated social welfare organization, the Hariri Foundation. He noted, “The Hariri Foundation provides high-quality medical services and generally serves all communities. The difference is that core supporters receive subsidized and free care.” For example, when she gave birth, the interviewee's aunt obtained a bed in a top private hospital, which usually caters to the wealthy. The foundation wrote a promissory note in which it pledged to pay all expenses associated with his aunt's care. The most ardent supporters may also receive supplemental and more expensive, longer-term benefits, such as educational aid or a place in a school. For schooling, the journalist claimed, “It requires a big connection to get a place in the Hariri Foundation schools. These are good schools and they have limited spots. So you must have an inside connection to secure a spot.” Another respondent, whose family is known to be a hard core supporter of the party, noted that her relatives frequent the Hariri Foundation's health institutions and benefit from additional material aid from the Foundation on a regular basis.⁴⁴ A woman whose son works for Future Security, a security service described as a proto-militia for the Future Movement, detailed the food boxes and other forms of material assistance that her son receives in addition to his regular salary.⁴⁵

Conversely, respondents who do not receive extensive benefits from political parties claim that their relative lack of social support is related to the fact that they are not core loyalists of political parties. A Christian man from the Christian neighborhood of Ashrafieh in Beirut backs Michel Pharaoun, a Christian candidate who has run on Future Movement lists in Beirut since 2000. Nonetheless, he insists that he does not receive his “rightful” share from social organizations linked to the party because he is not considered a core supporter.⁴⁶ Similarly, a Christian woman from the mixed neighborhood of Borj Abou Haidar in Beirut observed that she “stopped pledging support for politicians, which may have affected [her] ability to receive support from these organizations.”⁴⁷

The most active and visible supporters are most likely to obtain multiple forms of assistance. Hard core party loyalists not only receive basic rewards such as food and other small-scale benefits but also benefit from more expensive and longer term forms of support, such as medical and educational assistance.

The arguments presented above posit that political activism increases to access to welfare, but a plausible objection is that the reverse relationship obtains. The provision of social services leads to greater political activism as citizens increase their participation in partisan events in order to obtain benefits. Nonetheless, theoretical and empirical reasons support the claim that demonstrated political support is more likely to result in increased access to welfare goods than the reverse, particularly for more comprehensive combinations of benefits. Theoretically, it is unlikely that parties bestow substantial benefits on those whose support is uncertain or who are firmly committed to opposing parties given limited party resources. Furthermore, in a context where religious identity is not easily malleable and where conflict waged in the name of sectarianism has stoked sectarian divisions, citizens tend to support co-religionist parties regardless of the benefits they receive. In this atmosphere, social assistance serves to reinforce community boundaries rather than create them. As a result, parties aim to increase turnout rather than to woo the uncommitted during electoral cycles. The logic of turnout-buying (Nichter 2008) calls for rewarding established partisans with demonstrated histories of loyalty rather than luring more marginal or non-supporters.

Empirically, in-depth interviews with elites and non-elites in Lebanon indicate that even food and other limited forms of material assistance such as cash handouts, which are often deployed for electoral purposes, are primarily used to reward existing supporters. Political participation, which for most people amounts to casting a ballot in national and local elections, is a virtual prerequisite for access to basic services supplied or facilitated by political parties as well as for the intermediation of elected officials. A representative from a small political party that aims to rival the Future Movement joked about the exchange of votes for services in Lebanon: “In the U.S., [providers] don't ask if you are a Democrat or Republican before they give you a service. That's what they do here.”⁴⁸

An examination of the distinct forms of political engagement that comprise the PAI also supports the claim that activism precedes the receipt of benefits. For instance, a respondent might engage in some forms of political participation such as voting or attending political meetings in order to visibly “repay” a political party for social assistance but it is less likely that she would hang posters and other materials in the privacy of her home after receiving benefits. Moreover, some of the questions in the survey ask respondents to report forms of participation carried out in the 2005 elections. However, the survey, which was conducted in spring 2008, asked respondents to report aid received in the prior 12 months. Respondents did not vote or volunteer in the 2005 elections as a result of benefits received almost 3 years later. Arguably, respondents might demonstrate support for a party in anticipation of the need for social assistance in the future. Yet, people usually do not engage in such long-term planning for the use of social services such as medical treatment and, in any case, this behavior would support the logic of the argument that political activism “buys” social assistance. Thus, for both theoretical and empirical reasons, it is likely that political support is generally established prior to receipt of material benefits.

Conclusion

The case of Lebanon shows how politics can directly mediate access to social benefits—even those that should be automatic citizen entitlements. As interviews and a national survey on access to welfare and political participation suggest, more politically active individuals, who may vote, attend party meetings, participate in demonstrations, and engage in other regular and visible expressions of commitment to a party, are more likely to receive services, even after accounting for differences in socioeconomic status, gender, age, religious participation, piety, sectarian identity, and other potentially relevant factors. The survey and interview data also support the hypothesis that different categories of supporters receive different quantities of welfare goods. In general, varied profiles of political behavior are associated with different levels of benefits. Higher levels of political activism, which are characteristic of more committed supporters, are associated with access to more benefits, ranging from a single benefit to a combination of two or more benefits.

Future research should explore in more depth whether different levels of political activism are associated with greater access to distinct types of benefits. Interviews with elites and non-elites suggest that marginal supporters are more likely to receive rudimentary forms of assistance, such as food baskets or cash handouts, which require low fixed costs and occur only episodically or on a one-shot basis. At the same time, more committed supporters with long-term track records of active support for a party have higher probabilities of obtaining more expensive benefits such as aid for medical care or schooling, particularly on a continuous basis. It is impossible to test these hypotheses quantitatively with the data at hand because the survey only captures access to short-term material assistance for health care and schooling rather than more protracted and expensive forms of medical and educational assistance. Nonetheless, the logic of variable access to services is compelling: providers expend the least amount of resources and administrative capacity to offer food assistance and therefore are likely to offer it more broadly, including to marginal supporters as a reward for one-time actions such as casting ballots in elections. Schooling and some types of medical care, however, are more expensive and imply a more continuous commitment. Given higher investment levels, educational aid, including tuition assistance and access to coveted spots in private schools, as well as more protracted or capital-intensive forms of medical care are likely to be reserved for the most ardent and long-term supporters, who have iterated exchanges with their preferred political parties.

The linkage between political activism and access to basic services has disturbing implications for social justice. The fact that political affiliation and behavior mediate access to social assistance means that individual citizens—and particularly low-income families who need it most—may not be able to meet their basic needs, particularly if political groups do not perceive them as loyal partisans. Social provision by political organizations can also have detrimental effects on a more macro-level. The creation of multiple welfare networks, each linked to a different political organization, leads to fragmented welfare regimes. In the health sector, this is manifested in vast inefficiencies in the overall public health regime as well as stark inequalities in access to medical care (Ammar 2003; OXFAM 2009; Tabbarah 2000). Inefficiencies and inequalities are also evident in the educational sector because non-state providers such as political parties and religious charities have launched their own school systems with little regard for regional disparities in the distribution of educational facilities. Control over schooling by ethnic or sectarian parties also has important implications for socialization and the long-term prospects for fostering a commitment to a national political community. In divided societies, the stakes are particularly acute as such groups may promote distinct understandings of national history, which potentially hinders the construction of a coherent national identity in future generations (Anderson 1991; Doumato and Starrett 2006; Freedman et al. 2004; Kaplan 2006; Podeh 2000).

To be sure, NGOs can play a key role in providing basic public goods and empowering citizens to meet their basic needs (Brinckerhoff 1999; Brown 1998; Salamon 1995). Governments in many developing countries simply lack the capacity to provide for their own populations, whether due to lack of material resources, administrative deficiencies, or corruption. This is especially true in post-conflict societies such as Lebanon, where the experience of war depletes public resources and undercuts state institutions. In such contexts, non-state providers—including local ethnic, sectarian, and other political actors—play a key role in providing social protection. However, if they differentiate along religious or political lines in distributing basic services, they do not necessarily contribute to the broader public good. Discriminatory access to welfare goods can create new inequalities or further entrench existing inequalities and, at the extreme, can even strain social cohesion.

Biography

Melani Cammett is Associate Professor of Political Science and Director of the Middle East Studies Program at Brown University. She specializes in the political economy of development and the Middle East. Her first book, Globalization and Business Politics in North Africa: A Comparative Perspective (Cambridge University Press, 2007), examines how integration in global manufacturing chains reshapes business politics in developing countries. Cammett's current book project, entitled Everyday Sectarianism: Welfare and Politics in Weak States, focuses on how ethnic and religious parties allocate welfare goods in Lebanon with extensions to other organizations in the Middle East and South Asia. With Lauren M. Maclean, she is also completing a co-edited volume which examines the origins and evolution of non-state welfare providers in diverse regions and their consequences for social protection and state-building. She has also published numerous articles in scholarly journals and book chapters, and has consulted on political and economic development in the Middle East.

Appendix.

Survey Questions

The data for the key dependent and independent variables are derived from responses to the following questions:

Dependent variables

Welfare

Food Aid

In Lebanon, many different types of institutions provide gifts such as household supplies and baskets of food [husus ghithaiyya]. These might be voluntary organizations, religious organizations or political parties. In the past 3 years, have you received any gifts or material assistance from such organizations?

Aid for Health Care

(Questions repeated for clinic facility attended most frequently and additional facility that respondent attended less frequently.)

Some health clinics or medical offices offer financial assistance for their services. This might be in the form of a payment waiver, reduced prices, or reimbursement for all or part of the cost. Did you receive any financial assistance from the institution you visited for these medical services?

Some governmental and non-governmental organizations provide financial assistance to help people pay for the costs of visits to the doctor. Did you receive any financial assistance from another organization to help pay for the cost of these medical services?

Aid for Education

(Question repeated for each child below the age of 25.)

Some schools provide tuition fee waivers, reduced fees or reimbursement for all or part of school fees. Did anyone in your household receive any of these types of financial assistance from a school?

Independent variable

Political Activity Index

Volunteering and attending party meetings

Some people volunteer or work for their preferred political party or groups associated with it, perhaps by helping out around elections, serving on committees, planning events or staffing events. Have you volunteered for or worked for a group affiliated with a political party?

If responded answered “yes” to above question:

On average in the past 12 months, how often did you attend meetings of this group?

1. =More than once per week

2. =Once per week

3. =Several times per month

4. =Once per month

5. =Several times per year

6. =Once per year or less

Did you or a member of your family volunteer or work for a political party in the 2005 National Assembly elections?

Voting behavior

Did you vote in the 2005 National Assembly elections?

Did you vote for a complete list of candidates?

Party membership and support:

Are you a member of a political party?

Do you support a political party?

Partisan symbols and posters

The following questions were answered by the survey enumerator after the interview:

Did the respondent's home or attire include any political symbols or messages?

Which political party was represented in these symbols or messages?

Factor Analysis of Items in Political Activism Index

Item	Factor loading
Volunteer for party	0.835
Attended party meetings more than once per week	0.178
Attended party meetings once per week	0.369
Attended party meetings more than once per month	0.273
Attended party meetings once per month	0.341
Attended party meetings more than once per year	0.365
Volunteered for the 2005 national elections	−0.048
Voted in the 2005 national elections	0.313
Voted for the full party list	0.355
Member of party	0.664
Supporter of party	0.480
Displays political posters at home	0.393