Study population
We used data from the Positive Outcomes for Orphans (POFO) study, a longitudinal cohort of institution-dwelling OSC, family-dwelling OSC, and non-OSC. For the parent POFO study, five countries were identified from a group of 13 countries in which the study team had prior relationships with community-based organizations interested in the research. Those five countries (including two distinct sites in India) were selected for their ethnic, political, geographic, cultural, and historic diversity. While these sites were not chosen specifically for their civic engagement differences, and while each is contextually distinct, all sites are in South or Southeast Asia or sub-Saharan Africa, and are struggling with a growing population of OSC under conditions of increasingly stretched safety net resources.
The six POFO study sites in five LMIC are: Battambang District, Cambodia; Addis Ababa, Ethiopia; Hyderabad, India; Nagaland, India; Bungoma District, Kenya; and Kilimanjaro Region, Tanzania. Children ages 6—12 were recruited at baseline between May, 2006 and February, 2008, depending on study site. OSC in institution-based care or family-based care were randomly sampled to be statistically representative of orphans in their respective setting at each of six study sites; non-orphans were sampled from the same geographic clusters to be a qualitative referent group. Additional details of the sampling frame are described elsewhere [31, 32].
At 90-month follow-up (approximately 7.5 years after baseline), participants who were at least 16 years old were asked questions about their civic engagement activities and trust in government. This analysis focuses on those older adolescents.
Measures
Civic engagement
Given the dearth of studies on civic engagement in LMIC, we defined civic engagement to be consistent with existing work, especially since common themes emerged across many countries, even if those countries did not represent the resource-poor settings of interest in this study. Specifically, we defined civic engagement as the active involvement of community members in activities that contribute to community wellbeing, foster awareness of and participation in governance, or promote a sense of duty to help others. In the absence of existing work in this setting, we drew on existing measures from the Longitudinal Study of Adolescent Health in the United States so that our results had potential for comparability [33].
In participant interviews, we let OSC know we were interested in their unpaid participation volunteer and community work through organizations through service clubs, church and social action groups; we did not specify that we collectively termed these activities as ‘civic engagement’.
First, participants were asked whether they participated in regular community service or volunteer work between the ages of 12 to 18. They then were asked to specify whether it was strictly voluntary (because they wanted to do it) or whether it was required by caregivers, school, or a religious group. More than one answer to this question was accepted. Community and volunteer work is broadly defined to account for adolescents’ potentially limited opportunity (especially in varying cultural contexts) to participate in more traditional service activities or organizations; the emphasis is on unpaid efforts that serve to help or promote the wellbeing of others. Such activities may include helping a neighbor or community member outside the household with household work, volunteering at special events, participating in service days, providing instruction to other children, or other similar activities.
Additionally, participants were asked the degree to which they agreed with statements about trusting their local, state, and federal governments; the five-point scale ranged from “strongly agree” to “strongly disagree.” Though passive in nature, describing trust in government may be informative for understanding more active assessments of engagement and use of the existing measure enables potential for comparability to other settings.
For all sites, the voting age was 18. Respondents who were at least 18 were asked if they voted in the most recent government election. In addition, respondents 18 and older were asked whether they had ever tried to enlist in the military or militia. We coded this question as “Yes” if either or both were reported and “No” if neither was reported.
Demographics
Gender, age, setting (institution-based care, family-based care, or biological family for non-orphans), and orphan status (single, double, or neither) were ascertained at baseline. Number of years of completed education was obtained at 90-month follow-up. Grade-for-age was used for modeling to better reflect the participant’s educational attainment and to reduce collinearity with age. For those 18 years or older, twelve years of education was assigned as the target grade-for-age; additional years beyond that were considered above target.
Analyses
We calculated observed prevalences of the civic engagement variables described above. In addition, we constructed a predictive model of regular volunteer work (yes or no) using baseline demographics, current age, and current grade-for-age. Model parameters were estimated using Poisson regression with a log link. The model was weighted to incorporate the complex design of the sampling frame for orphans. Non-orphans (who were not sampled to be statistically representative) were assigned a weight of 1. Results are reported as prevalence ratios (PR) and 95 % confidence intervals (CI).
This study was approved by the Duke University Institutional Review Board (IRB) as well as IRBs at the individual study sites. Caregiver consent as well as child assent was obtained for all participants. Site-specific protocols were developed ensure mechanisms for child protection were in place in case of reports or observation of child maltreatment or neglect over the course of the study; interviewers were trained in these protocols.
All estimates were generated using Stata 13 [34].