Building social capital in post-conflict communities: Evidence from Nicaragua

https://doi.org/10.1016/j.socscimed.2008.12.024Get rights and content

Abstract

Studies of social capital have focused on the static relationship between social capital and health, governance and economic conditions. This study is a first attempt to evaluate interventions designed to improve the levels of social capital in post-conflict communities in Nicaragua and to relate those increases to health and governance issues. The two-year study involved a baseline household survey of approximately 200 households in three communities in Nicaragua, the implementation of systematic interventions designed to increase social capital in two of the locales (with one control group), and a second household survey administered two years after the baseline survey. We found that systematic interventions promoting management and leadership development were effective in improving some aspects of social capital, in particular the cognitive attitudes of trust in the communities. Interventions were also linked to higher levels of civic participation in governance processes. As in other empirical studies, we also found that higher levels of social capital were significantly associated with some positive health behaviors. The behavioral/structural components of social capital (including participation in groups and social networks) were associated with more desirable individual health behaviors such as the use of modern medicine to treat children's respiratory illnesses. Attitudinal components of social capital were positively linked to community health behaviors such as working on community sanitation campaigns. The findings presented here should be of interest to policy makers interested in health policy and social capital, as well as those working in conflict-ridden communities in the developing world.

Introduction

Most of the literature on social capital has focused on defining the component elements of the concept and assessing the relationship between social capital and social and political outcomes, including importantly, health behaviors and health status (Hardin, 2001, James et al., 2001, Kawachi and Berkman, 2000, Putnam, 1993, Putnam, 2000, Rose-Ackerman, 2001, Rose et al., 1998, Scott, 1985, Sztompka, 1999, Warren et al., 2001, Wessels, 1997, Woolcock, 1998). Although there are arguments about the role of different organizational structures in the formation of social capital, few works have demonstrated the effectiveness of attempts to increase the level of social capital in communities which have low levels of social capital (Bacon, 2004, De Sousa and Grundy, 2007, Jarrett et al., 2005; Portnoy & Berry, 1997; Semenza & Krishnasamy, 2007). There is also considerable concern about how to improve social capital in communities that have had histories of social and political conflict. Studies of post-conflict situations suggest the need to develop interventions that can help improve trust and participation in order to reestablish the civic society's capacity to contribute to a stable and economically sound community (Kreimer et al., 1998, Kuroda, 2002; Michailof, Kostner, & Devictor, 2002).

This article reports on one intentional effort to build social capital in post-conflict communities, a relatively simple and inexpensive intervention focused on developing management and leadership (M&L) skills of community members. The intervention was designed to build social capital in post-conflict Nicaraguan communities with low social capital in order to improve social capital, health status and civic participation in broader governance and political processes. The study was based on a baseline and a two-year follow-up household surveys implemented in three Nicaraguan communities, two of which received the social capital intervention and one similar community that served as a control without the intervention.

The research team, consisting of researchers from Harvard School of Public Health, ALVA, a local consulting firm, and Management Science for Health, believed that Nicaragua, a country recovering from a long civil war (1981–1989) but which has achieved some stability, would serve as an ideal location for this kind of social experiment. Following the civil war, many communities were created by resettling pro-Sandinista guerrilla armies and the armed opponents of the Sandinista regime. Issues of distrust and violence were evident throughout the country, particularly in rural communities. Underlying the political tensions was the social history of clientelism which limited avenues for horizontal collaboration necessary for developing social capital (Molenaers, 2003, Ruben and van Strien, 2001).

This project makes several important contributions. First, this is the one of the few studies that rigorously and quantitatively documents the effectiveness of deliberately planned efforts to increase the level of social capital in post-conflict communities in developing countries which have low levels of social capital. Second, this is the first study of its kind that assesses the effect of social capital on public or community health behaviors.

There is considerable debate surrounding the appropriate way to conceptualize social capital. Some, particularly economists, emphasize the individual or ‘networks’ perspective which focuses on the role of individual level interactions in producing social capital (Coleman, 1988). In this context, social capital is viewed as a resource available to individuals that may provide access to other resources and help promote health and well-being. In contrast, the collective or community level perspective views social capital as a resource at the aggregate level that individuals can leverage in order to achieve collective goals that may be more difficult to realize otherwise. In this study, we focus on and analyze social capital at the community level.

Social capital theorists have drawn a distinction between cognitive/attitudinal and structural/behavioral elements of social capital. Cognitive elements refer to attitudinal manifestations and include values, attitudes and beliefs. Trust in other members of the community is an example of ‘cognitive’ social capital. Structural elements are viewed as behavioral manifestations of social capital and refer to participation in groups, community and civic activities, as well as, social networks, roles and norms. Both structural and cognitive dimensions of social capital have been linked with positive health outcomes. Trust and participation in greater numbers of groups have been linked to decreased mortality (Kawachi, Kennedy, Lochner, & Prothrow-Stith, 1997; Kennedy et al., 1998, Lochner et al., 2003, Skrabski et al., 2003; Subramanian, Kawachi, & Kennedy, 2001), improved mental and physical self-reported health status (Gundelach and Kreimer, 2004, Helliwell and Putnam, 2004, Hyyppa and Maki, 2001; Kawachi, Kennedy, & Glass, 1999; Yip, Subramanian, Mitchell, Lee, Wang, & Kawachi, 2007), and decreased violence (Galea, Karpati, & Kennedy, 2002) and higher incomes (Grootaert, 2001). This study evaluates the impact of a social experiment designed to increase both cognitive and structural components of social capital and to assess the relationship between these components and selected health behaviors and participation in governance processes.

The social experiment was carried out over three stages. Initially, the research team enlisted a team of government, community and representatives from non-governmental organizations to assess, based on their experiences, the level of social capital within six communities with similar demographic characteristics. Two communities – El Rosario and Villanueva – were chosen with perceived “high” levels of social capital, while four communities – Cinco Pinos, Pantasma, Rivas and Waslala – were selected with reportedly “low” levels of social capital. In Stage One (August 2003), the research team analyzed the levels of social capital in these six communities. Preliminary quantitative analysis comparing social capital indicators found that the communities with perceived levels of high social capital did in fact rank higher on social capital measures than communities with perceived levels of low social capital. The team decided to implement a management and leadership (M&L) training program intervention in two communities (Pantasma and Waslala) and to maintain one control community (Cinco Pinos). While Rivas also shared similar demographic characteristics, resource constraints prevented the research team from collecting a larger sample size. In Stage Three (March 2005), researchers conducted a follow-up household survey to determine whether specifically-designed activities had an observable impact in raising community levels of social capital and civic participation and in improving health behaviors.

In the first baseline survey, 198 households (HH) were interviewed: Cinco Pinos (87 HH, population 592), Pantasma (55 HH; population 878), and Waslala (56 HH, population 1100). Interviewers interviewed the head of the household and in those cases where the head of household was not present, they interviewed the oldest adult (older than 18 years). In the second follow-up survey, 210 households (HH) were interviewed: Cinco Pinos (92), Pantasma (58), and Waslala (60). For both the baseline and follow-up surveys, we used a systematic random design methodology. In addition, the samples were independent, meaning that the individuals surveyed before and after the intervention may have been different which is appropriate given our interest in assessing the impact of interventions on community levels of social capital. The difference in the number of households interviewed owes to a higher response rate for the follow-up survey which does not pose a problem to our interpretation of results since we are examining community levels of social capital. Within the communities, census data on the number of households determined the sampling frame, with a rate of selection of every second or third household depending upon the number of households within each community. For our sampling frame, we have about a 10% margin of error (for an estimated percentage of 70 and 90% confidence interval).

Following previous studies on social capital which use World Bank methodology and the World Bank's Social Capital Assessment Tool (SCAT) (Bossert et al., 2003, De Silva et al., 2007, Grootaert et al., 2004, Harpham et al., 2002, Krishna and Shrader, 2000, Mitchell and Bossert, 2007, Narayan and Pritchett, 1999), we computed social capital indicators at the community level which capture both the cognitive and structural components of social capital. For the cognitive elements of social capital, we focused on feelings of trust and solidarity, social harmony and sociability. For the structural elements we analyzed participation in groups and “meeting load” (frequency of attendance in meetings). We also explored the impact of the interventions on civic participation in governance processes, which we viewed as an outcome or manifestation of social capital. Many of the variables were Likert scales which invited self-reporting. Information on individual characteristics such as types of housing, living conditions, and education levels was also collected from the SCAT survey. Data on poverty levels was collected from the Instituto Nicaraguense de Fomento (2000).

In the first survey, information about health behaviors related to child and maternal health was collected by asking specific questions based on Demographic and Health Surveys (DHS) questions to all households with children less than five years of age. These questions asked about: the use of modern medicine for recent cases of child diarrhea and respiratory illnesses and maintenance of child development and growth. These health behaviors are particularly important in Nicaragua where the leading causes of infant mortality are diarrhea, acute respiratory illness, and malnutrition.

An additional set of broader questions was asked in the second survey, including individual perception of their general health their use of family planning methods, and participation in community (public) health activities including efforts to improve the nutrition and health of the children in the community and to participate in community clean-up campaigns. Because this survey containing the larger set of health questions was conducted after the implementation of social capital interventions, we only analyzed the associative relationships about higher levels of social capital in Stage Three of the study and the health variables in the communities.

Stage Two of the project (November 2003 to July 2004) involved a systematic, methodologically-based structured M&L program in two of the three communities with low levels of social capital – Waslala and Pantasma. The project team also monitored the control community of Cinco Pinos to assure that no similar intervention occurred there.

The objectives of the USAID funded M&L Project were to implement and support activities that would a) develop management and leadership capacities with the goal of strengthening community organization and self-management and b) encourage the development of higher levels of household participation in community activities and increase trust among community residents and between the community and local public institutions. While the interventions were dynamic and tailored to the specific needs and contexts of individual communities, they had to meet the following broad requirements:

  • build on existing organizations in the community rather than impose new organizations

  • develop participation mechanisms that encourage increased and continuing attendance at meetings and encourage broad participation in project activities

  • develop communication, consensus building and conflict resolution skills both in the community organization and within the wider community to build higher levels of trust within the community

  • encourage decision making and empowerment of community members especially those who have not participated previously

  • create enduring ties of support with organizations outside the community

The strategies of activities implemented in Pantasma and Waslala involved: 1) technical assistance to support the formation of a Committee for Municipal Development and to assist the mayor and associates with moral leadership and other functional skills; 2) a series of leadership training workshops for leaders of the community, teachers, and others in content areas of moral values, leadership, strategic planning, budgeting, and conflict resolution training; 3) information, education and communication over a community radio program, as well as value and moral leadership campaigns, educational courses, mural paintings and other activities; and 4) outreach which involved the technical team's regular monitoring of families in the communities to motivate participation and trust among community members. The local research team systematically monitored the implementation of the interventions to assist in improving their effectiveness and ensure a degree of consistency across communities.

Given that many of our dependent variables are dichotomous and ordered variables, we use the maximum likelihood logistic regression estimation for survey design and report odds ratios. We have two estimating equations. First, since we want to test specifically whether the M&L program had an impact on community or aggregate levels of social capital, civic participation and health behaviors, we create an “interaction” term between a time dummy variable and whether or not the community received the intervention (i.e. intervention × time)). It is this interaction term which informs us whether the systemic M&L program had an effect on levels of social capital after the treatment period.

The estimating equation for our basic model (Table 2, Table 3, Table 4, Table 5) is:Y=β0+β1Time+β2Intervention+β3Time×Intervention+β4Poverty+e

In this model, Y is the set of cognitive and structural dimensions of social capital (Table 2, Table 3), civic participation and empowerment (Table 4) and health behaviors (Table 5). The sample includes pre- and post-intervention observations from the two treatment communities and the control group (Cinco Pinos). The size of the sample is 408 (198 in the initial survey and 210 in follow-up survey). We analyzed the relationship between the interventions and an extensive list of social capital indicators. With the exception of the participation and meeting load variables, we report only those variables that are statistically significant at 10% level of significance.

Given that the survey questions related to community health practices were limited to post-intervention observations, we were only able to assess the associative relationship between social capital and health behaviors. (Appendix A includes descriptive statistics for variables pre- and post-intervention.) For this set of equations (Table 6, Table 7), our model is:Y=β0+β1Social Capital+β2Poverty+eIn this model, Y is the set of individual and community health behaviors. Social Capital is a set of social capital variables measured by questions in the survey. We ran each of the specifications with a single social capital indicator so as to isolate the effects of our variable of interest. This estimating equation allows us to examine whether higher levels of social capital are associated with positive health behaviors and outcomes in all three communities. In other words, we ask: “Regardless of whether the community received an intervention, is it the case that higher levels of social capital are associated with more positive individual and community health behaviors?” Across all specifications, our estimation models used sampling weights (pweight) in order to arrive at the correct point estimates. Our sampling fraction is 408/1302 (the inverse of this is pweight). We also considered stratification by municipal groups to avoid biased standard errors and control for level of extreme poverty. Finally, given our limited population size, we ran each regression with the finite population correction (fpc). This reduced the size of the standard errors associated with small populations, thereby increasing the precision of our estimates.

Table 1a presents the baseline values of social capital indicators for the three communities. Table 1b shows the percent change in various social capital indicators following the systematic social capital interventions in two communities, Waslala and Pantasma, and the control community, Cinco Pinos. As shown in Table 1b, most of the social capital indicators have increased over time, however at different rates.

We measured the structural components of social capital with three survey questions: participation in community groups, frequency of attendance at meetings (“meeting load”), and contributions made to the groups to which respondents belong. In the baseline survey, more than half (56.3 and 57.1%) of Cinco Pinos' and Waslala's residents participated in a group. Following implementation of the social capital interventions, participation in groups increased. In Pantasma, group participation rates increased from 48 to 60% and in Waslala, participation rose from 57 to 61%. Cinco Pinos, the control community, showed the smallest increase, from 56.3 to 59.5%. However, the regression analysis results (Table 2, Column 1) showed that while “intervention × time,” is positively related (1.21), it is not statistically significant (p = 0.629).

The variable ‘meeting load’ measures the log of the total number of times summed across groups each month that someone from the household attended groups meetings. As reported (Table 1b), interventions were associated with an almost 2% increase in meeting attendance in Waslala, but a slight decline of 1.4% in Pantasma. The results of the regression analysis (Table 2, Column 2) indicate a surprising statically significant negative relationship between social capital interventions and meeting loads suggesting either weariness with meetings or more efficient meetings. More than half of respondents across all communities reported making a contribution to the groups to which they belonged (Table 1a). Social capital interventions were associated with an increase in contributions: In Pantasma, the percent of respondents contributing rose by 26.2% and in Waslala, there was a 4.5% increase. However, the regression results (Table 2, Column 3) indicate that while the relationship between interventions and increases in contribution is positive, it is statistically insignificant. Overall, the results (Table 2) indicate that M&L activities did not have a positive impact on increasing structural components of social capital.

Trust and solidarity are principal cognitive components of social capital. General levels of trust were very low in Nicaragua as might be expected in a post-conflict situation. Only 13% of all respondents in the baseline survey (Cinco Pinos 17.2%, Pantasma 16.4% and Waslala 3.6%) said that they could trust people generally. Just over one-fourth (26.3%) felt they could trust people in their neighborhood (Cinco Pinos 28.7%, Pantasma 18.2% and Waslala 30.4%).

An analysis of the follow-up survey after the social capital interventions revealed higher levels of trust. Almost one-fourth (22%) of respondents stated that they trusted people generally. The percentage of survey respondents who said they trusted people in general increased by 18.1% in Pantasma, 6.4% in Waslala and 4.8% in Cinco Pinos, the control community. The results (Table 3, Column 1) indicate that while the relationship between the systematic interventions and trust was positive (2.04), it was not statistically significant (p = 0.173). However, the regression analysis revealed a positive and significant relationship between the interventions and our questions about “solidarity” – namely the belief that one's neighbors were ready to assist in times of need (Column 2).

Sociability and social harmony are additional attitudinal components of social capital. The interventions had a mixed impact on increasing social interaction among community members (Table 1b). While there was a 14.6% increase in the percentage of respondents in Pantasma who said that they had met with their fellow residents to talk, there was a 12.1% decline in Waslala. The regression analysis (Table 3, Column 3) indicates a negative although statistically insignificant relationship between the interventions and increased social interaction.

With respect to social harmony, we find a statistically significant positive relationship between interventions and these elements of social capital. Social capital interventions were associated with increased feelings of closeness among community residents (Table 3, Column 4). Moreover, there was a strong perception among respondents that the spirit of participation in the community was high and that people were making fair contributions to communal activities. The analysis suggests that the social capital interventions had a much greater impact on increasing some of the attitudinal elements of social capital than on structural elements of social capital (Table 2, Table 3).

Some theorists believe that civic attitudes and norms are the product of regular social interaction as (Putnam, 1993). As such, civic participation in governance processes and empowerment are often viewed as outcomes or manifestations of social capital. Following this conceptual framework, we examine the effect of the M&L interventions on changes in civic participation and political empowerment and find support for a strong relationship.

Social capital interventions were significantly associated with increased likelihood of working with one's neighbors on a project to benefit the community (Table 4, Column 1). Systematic interventions significantly increased the likelihood to talk to one's neighbors about a community problem (Table 4, Column 2).

The analysis indicates that feelings of empowerment have increased among residents in the communities that received management and leadership interventions. In particular, interventions positively affected the likelihood of community residents to participate in a protest, contact local health officials about a local problem, and donate to a non-profit organization. Some might argue that the significance of these findings is limited by problems with endogeneity. While it is true that the management and leadership programs supported community decisions to organize and develop certain projects (such as Committee for Municipal Development), it did not require participation in certain forms of advocacy (such as voting or contacting local representatives).

Section snippets

Individual health behaviors

Based on previous studies of social capital, we hypothesized that levels of social capital should be positively associated with individual and public health behaviors. Only three out of the eight questions on health behaviors were asked in the baseline and follow-up surveys (Table 5). We find that while the interventions had a positive effect on individual health behaviors, the relationship was not statistically significant (Table 5).

For the five questions asked only in the follow-up

Discussion

In this exploratory assessment of a social experiment to increase social capital, our findings provide some evidence that interventions designed to increase social capital had a positive effect on increasing some aspects of social capital in post-conflict communities of Nicaragua. In particular, M&L program activities were significantly related to increased levels of cognitive social capital, including solidarity, harmony and sociability. Interventions also resulted in higher levels of civic

Conclusion

The goal of this applied study was to determine the nature of the kinds of systemic actions that might positively affect the development of social capital. Specifically, this project assesses whether direct interventions focused on developing social capital (i.e. management and leadership training and support) can help build levels of social capital in post-conflict communities. Based on an analysis of three communities, one of which served as a control community, we find some evidence to

References (43)

  • J. Coleman

    Social capital in the creation of human capital

    American Journal of Sociology

    (1988)
  • E.M. De Sousa et al.

    Intergenerational interaction, social capital and health: results from a randomized controlled trial in Brazil

    Social Science & Medicine

    (2007)
  • C. Grootaert

    Does social capital help the poor?

    (2001)
  • C. Grootaert et al.

    Measuring social capital: an integrated questionnaire. Working Paper No. 18

    (2004)
  • P. Gundelach et al.

    Happiness and life satisfaction in advanced European countries

    Cross Cultural Research

    (2004)
  • R. Hardin

    Trust and trustworthiness

    (2001)
  • T. Harpham et al.

    Measuring social capital within health surveys

    Health Policy and Planning

    (2002)
  • J. Helliwell et al.

    The social context of well-being

    Philosophical Transactions-Royal Society of London Biological Sciences

    (2004)
  • Instituto Nicaraguense de Fomento

    Municipal demographics

    (2000)
  • S. James et al.

    Social capital, poverty, and community health: an exploration of the linkages

  • R. Jarrett et al.

    Developing social capital through participation in organized youth programs

    Journal of Community Psychology

    (2005)
  • Cited by (66)

    View all citing articles on Scopus

    We would like to thank the community members and leaders in Waslala, Pantasma, Cinco Pinos, El Rosario, Rivas, Villanueva and who gave generously of their time - participating in both the surveys and the training activities of the Leadership and Management teams. We want to thank our collaborators in Management Sciences for Health (MSH), especially Mary Luz Dussán and Susana Galdos, and Barry Smith who was so supportive of this initiative. We also relied on the excellent research support of Diana Bowser at Harvard School of Public Health. Thanks also to Andrew Mitchell, Marc A. Musick, S V Subramanian for guidance and advice. We also want to thank the USAID team, led by Alonzo Wind, whose interest and financial support was crucial to the success of this study. The research for this article was funded by the US Agency for International Development under Cooperative Agreement HRN-A-00-00-00014-00 with MSH and a sub-contract with the Harvard School of Public Health. The authors are fully responsible for the analysis and interpretations presented in this article.

    View full text