RELIGION NETWORKS AND HIV/AIDS IN RURAL MALAWI
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
The Degree of Doctor of Philosophy in the Graduate
School of Ohio State University
By
jimi adams
* * * * *
Ohio State University 2007
Dissertation Committee: Approved by Professor Kazimierz M. Slomczynski, Advisor
Professor James W. Moody, Outside Member ______
Professor Korie Edwards Advisor Sociology Graduate Program Professor Steven H. Lopez
Copyright by
jimi adams
2007
ABSTRACT
Sub-Saharan Africa’s residents represent approximately two-thirds of the
nearly 40 million global HIV/AIDS cases, while comprising only about one-tenth of
the world’s population. In the rural settings where most inhabitants of SSA live,
religious organizations are the only formal organizations present, and virtually all
residents of SSA participate in a religious organization. Many have theorized a
relationship between religion and HIV/AIDS, suggesting alternately its helpful and
harmful potential in this crisis. The existing research conceptualizes religion, HIV
risk and the connection between them by studying individuals, organizations, or
aggregations of individuals and organizations. In this dissertation, I demonstrate the
adjustments a network perspective contributes to researchers’ ability to understand
religious organizational responses to this epidemic, the nature of HIV-risk and,
perhaps most importantly, how these are linked. The resulting conceptualization
suggests some of the first mechanisms that demonstrate how beneficial and harmful
HIV-related outcomes can arise simultaneously from religious structure or
corresponding individual behaviors.
The first section describes religious organizational networks to demonstrate
how social network structural factors shape the HIV-related messages conveyed
within religious organizations. While many intended models of prevention and
intervention rely on implicit formal organizational hierarchies, little is known about ii how this contributes to the content and effectiveness of subsequent prevention
messages. Therefore, I first provide comparative description of the networks within
which local religious leaders develop the HIV-related messages conveyed in their
congregations. I then compare the HIV-related messages of religious leaders at the
national-denominational and local-congregational levels, to demonstrate the existing
gaps between the intended model and the functional reality of these efforts. I explain
how local clustering of these networks drives the discordant messages. For the
analyses in this section, I draw on a social network perspective as a story-telling
device that provides a better understanding of how religious organizations develop
their responses to the HIV-epidemic.
Conceptualizations of individuals’ HIV-risk also benefit from adopting a
network perspective to more readily capture the epidemiology of HIV. While
individuals’ religious involvement may reduce risk behaviors; HIV risk cannot be
fully understood in these terms. Potential infection is dependent, not only the
frequency of risk behaviors, but also on characteristics of ones’ partners (e.g.,
probability of their infection). With the final analysis in this dissertation, I generate a
series of simulated networks to demonstrate that the differences observed in risk
behaviors associated with religious affiliation do not necessarily translate into
corresponding differences in network-oriented risk properties. I draw on data from the
Malawi Religion Project and the Malawi Diffusion and Ideational Change Project to
investigate these questions.
iii
ACKNOWLEDGMENTS
This project has been a long process of which I have played but one part.
Ultimately, the ideas are mine, so only I can be blamed for what’s here, however the
contributions of quite a few people substantially strengthened the work. I particularly
would like to thank my advisors in this project. While a dissertation typically
progresses under the guidance of one person, mine has included the tutelage
(officially and unofficially) of three. Their willingness to work with and defer to one
another during this process has allowed the work to continue, and not get bogged
down in administrative limbo.
James Moody has provided invaluable guidance in framing the project as a
whole, and providing continual support – theoretical, technical and mental, at stages
when each were nearing complete breakdowns.
Kazimierz Slomczynski took me on as a student at a relatively advanced stage
of the project, and provided helpful pushes forward, without which the project, and
my career, probably would have stalled out several times.
Susan Cotts Watkins invited me to join the projects that provide the data for
this project. Her continual offering of suggestions, and unmatched knowledge of what
has already been done have been invaluable resources in this process. She willingly
let graduate students, not only run the data collection projects, but also shape their
content, which is beyond what I could have asked for. This gave me the chance to iv examine questions that interested me in ways that otherwise would have not been
possible. For that opportunity, I am immensely grateful.
I would also like to thank the other members of my committee, Korie Edwards
and Steven Lopez for adding their helpful insights into areas of this project where
their expertise was especially salient, and filled what otherwise would have been
substantial gaps.
Two fellow graduate students have been particularly helpful in my making it
through this process. Jenny Trinitapoli willingly accepted me into what was, in
essence, her project. I thank her for not seeing me as infringing on her territory, but
instead being an incredible sounding-board. The ideas here were often little more than
half-baked before our discussions formed them into something cogent. Joshua
Dubrow has read drafts of, and provided helpful feedback on, virtually everything I
have written over the past five years, and has not laughed at me once. That I know of.
There are numerous others who read drafts of this work, chatted online or over
coffee about the ideas, or simply offered a listening ear when needed. Among those, I
would especially like to acknowledge the feedback and support I received from Sara
Bradley, Agnes Chimbiri, Peter Fleming, Stephane Helleringer, Hans-Peter Kohler,
Ryan Light, Michelle Poulin, Mark Regnerus, Georges Reniers, Irina Tomescu-
Dubrow, Alexander Weinreb and Sara Yeatman. I would also like to note the
MDICP/MRP field teams for their invaluable contributions, not only to the collection
of excellent data, but also to shaping the projects that come out of them. In particular,
my work has especially benefited from the input of Sydney Lungu, James Mwera,
Julius Nyambo and Joel Phiri. v During this project, I have received financial support from several sources,
without which the work would not have been possible. I am grateful for this support
from the National Institutes of Health (grants 1R01-HD050142 and 5R01-HD041713,
Susan Watkins, PI; 7R01-DA012831 and 1R01-HD041877, Martina Morris, PI), the
Society for the Scientific Study of Religion, the Office of International Affairs and
the Initiative in Population Research at Ohio State University, and the Population
Studies Center at the University of Pennsylvania.
My parents long ago taught me (or allowed me) to ask questions in ways that
inevitably paved my path to grad school. For that and the support they provided along
the way, I am thankful. Several friends of mine who have pastored the various
churches I’ve attended over the past decade have pushed (or pulled against) me when
necessary. I want to thank for that: Matt Bartley, Mike Borst, Brian Brooks, Chris
Martin, JR Woodward and Jim Zippay. Each of you helped me to recognize that my
faith and my work do not have to be, and in fact should not be, separate.
The places I have lived while writing this have been an incredible support, and
for being there through it, I am grateful to all of you with whom I’ve shared those
spaces, and considerable chunks of our lives: Kyle Bush, Jed Dearing, Tim & Jenny
Evans, Tim Poindexter, Cisco Sanchez, John & Karen Thomas and Jenny Trinitapoli
& Gregory Collins (and Cassia).
And most of all, I want to thank God for never giving up on me no matter how
many times I try to on Him.
vi
VITA
October 21, 1976…………………….. Born – Lexington, KY
2004…………………………………...M.A. Sociology, Ohio State University
2001 – 2006 ………………….……… Graduate Teaching and Research Associate, Ohio State University
PUBLICATIONS
1. adams, jimi. 2007 “Stained Glass Makes the Ceiling Visible: Organizational Opposition to Women’s Congregational Leadership.” Gender and Society, 21(1): 80-105.
2. adams, jimi and James Moody. 2007. “To Tell the Truth?: Concordance in Multiply Reported Network Data.” Social Networks, 29(1): 44-58.
FIELD OF STUDY
Major Field: Sociology
vii
TABLE OF CONTENTS
Page Abstract …………………………………………….…………………………...….....ii Acknowledgments …………………………………………………………...…...…...v Vita ……………………………………………………………….………..………..vii List of Tables …………………………….…………………………………..……….x List of Figures ………………………….……………………………………..….....xii
Chapters:
1. Introduction: A Relational Approach for a Relational Problem ...….………...1
2. Theoretical Orientation: Moving from People and Variables to Relations and Networks …………………………………………………..20
3. Data: The Malawi Religion Project and The Malawi Diffusion and Ideational Change Project ..…………………….44
4. Examining Policy Location (not Locution): Congregation Leaders’ HIV-Related Discussion Networks …...... 63
5. What’s Lilongwe Got to Do With It? Comparing Religious HIV Discourse between Denominational and Congregational Leaders ….…97
6. Damned if You Do, Damned if You Don’t: Religious Affiliation and HIV Risk Network Structure .……………...…...133
7. Conclusion and Discussion ………………………………..……………….166
viii
TABLE OF CONTENTS cont’d
Page Appendix A: Congregation Leader - Qualitative Training Guide …….…...... …173
Appendix B: Network Component of Congregation Leader Survey .…...... 177
Appendix C: National Leader Interview Guide .…….…………………………...183
Appendix D: Global-to-Local Supply Chain of HIV-Intervention – A Conceptual Model ………………………………………………189
References .………………………………………………………………………...191
ix
LIST OF TABLES
Table Page
1.1 HIV Status by Religious Affiliation MDICP-3 ……………………………..12
2.1 Structure Versus Agency in Actions and Outcomes ………………………...41
3.1 Summary of Religious Affiliation and Participation MDICP-3 …...... 60
3.2 Comparison of Coverage – National and Congregation Leader Interviews ...61
4.1 HIV-Umbrella Organizations by District – Malawi, 2005 ………………….88
4.2 Organizational Interconnections by Religious Tradition ……………………89
4.3 Congregational Co-Participation with Other Religious Organizations …..…90
4.4 Leaders’ Doctrinal Conversation Partners Outside the Congregation ………91
4.5 Leaders’ AIDS Conversation Partners Outside the Congregation ………….92
4.6 Logistic Regression Predicting Doctrinal Conversation Partners …………..93
4.7 Logistic Regression Predicting AIDS Conversation Partners ……………....94
5.1 MRP Congregation Leaders’ Description of the Scope of the HIV Problem ……………………………………….....130
5.2 MRP Congregation Leaders’ Evaluation of the Source of HIV as a Problem ………………………………………..131
5.3 MRP Congregation Leaders’ Suggested HIV-Prevention Strategies …...…132
6.1 HIV Status by Religious Affiliation MDICP-3 ……………………………157
6.2 Static Simulation Parameters ………………………………………………158 LIST OF TABLES cont’d x
6.3 Simulation Parameters – Sexual Partnerships ……………………………..159
6.4 Network Properties from Self-Reported Sexual Partnership Data ……...…160
6.5 Network Properties from AIDS-Discussion-Partners-Reported Sexual Partnership Data …………………………………………………...161
xi
LIST OF FIGURES
Figure Page
1.1 Malawi Map, Highlighting MDICP Research Locations …………………...19
2.1 James Coleman’s “Boat” – Macro-Micro-Macro Linkages and Methodological Individualism ………..….…….……………..42
2.2 Network Rhombus – Meso(-Node) – Meso(-Node) Relations and Methodological Structuralism ………………………………..43
3.1 Malawi Religion Project Sample ……………………………………………62
4.1 Potential Denominational Leadership Relationship Structures ……………..87
6.1 Rumphi Simulated Population Self-Reported Sexual Partnership Data - Exemplar Network ……………………………..162
6.2 Mchinji Simulated Population AIDS-Discussion-Partner-Reported Sexual Partnership Data - Exemplar Network ……………………………..163
6.3 Balaka Simulated Population AIDS-Discussion-Partner-Reported Sexual Partnership Data - Exemplar Network …………………………..…164
6.4 Balaka Simulated Population AIDS-Discussion-Partner-Reported Sexual Partnership Data - Exemplar Network …………………………..…165
xii
CHAPTER 1
INTRODUCTION: A RELATIONAL APPROACH
FOR A RELATIONAL PROBLEM
In mid-July of 2005, I was invited to attend the funeral of Madalitso,1 a young
woman from Mchinji District of central Malawi. In the short walk from where we
were staying to the home where people were mourning, we had a brief discussion
with a local businessman, during which we learned that Madalitso had likely died due
to AIDS-related illnesses. They suspect she contracted it from her former husband,
because he was “movious”,2 and had “gotten thin” before he passed away the year
before.
As we entered the small two-room, sun-burned brick home with a tin roof (a
marker that this family was comparatively wealthy), we found approximately 15
women sitting seated shoulder to shoulder around the walls of the front room in the
house. Two more women sat in the middle of the floor next to Madalitso’s body
which was wrapped in several old chitenjes. Almost as soon as we took our seat in
the open spaces they’d created for us along the wall, one of the women reached down
and pulled away the chitenje covering Madalitso’s face, which set off a new chorus of
1 cries through the women in the room. That’s when I realized that the woman two
people to my right was Madalitso’s mother.
What struck me most about Madalito’s funeral was how much it reminded me
of the same rituals we observe in visitations and funerals in the U.S.; and yet how
very different it was. At the time, however I couldn’t put my finger on the difference.
“AIDS is becoming like a common song that people are talking over whenever they meet with one another.” (Baptist Pastor, Balaka)
When I read this quote from one of the Malawi Religion Project interview transcripts,
the difference struck me. In the funeral I attended in Malawi, there was no pretense to
the "ceremony" - attempting to distance Madalitso's death from the rest of her
family's life. In the U.S., of the handful of funerals I have attended, not one was in the
deceased family's home. Every time the corpse has been neatly primped and adorned
in a way that tries to hide the reality. Conversely in Malawi, a setting where virtually
every carpenter's sign advertises "coffin-making" because the reality of death makes
it economically advantageous, similar practices would be nonsensical. Death is a
reality that Malawian's face regularly. Their proximity to death, and its causes,
enables a much more direct recognition, both of death’s reality, but also of how it is
linked to life. By removing death into the confines of specialized parlors in the U.S.,
we have resituated death as an event that disassociates it from every other component
of our lives.
In Madalitso’s case, where her body lay in the center of the room was likely a
similar spot where she spent hours suffering in the days before she died. Her mother,
and most of those in the room that afternoon knew what this disease had done to her
2 body, and it was still evinced in these moments shared together. Pretenses allow for
distancing death from life, but Madalito’s family wasn’t attempting to separate them.
In this dissertation, I aim to take a cue from what I learned from this funeral.
Rather than focusing on the comfortable, the familiar, or the distanced, hopefully in
this dissertation I can strip away some of the pretenses that frequently get attached to
the coverage of topics that garner as much attention as the present HIV-pandemic.
Rather than taking the words of organizational figureheads as representative of what
religious organizations are doing to respond to the epidemic, I will examine the
messages presented by varying levels of religious leaders. Instead of assuming
religious organizational hierarchies exist and are useful for disseminating prevention
strategies, I investigate the relational structure within religious organizations. As
opposed to assuming that observed behavioral changes necessitate protective effects
against contracting HIV, I will consider how those same changes may actually
generate as many problems as they solve. Time to pull back the cloth and see what we
find.
In the twenty-five years since the discovery of HIV/AIDS, the number of
people infected in the world has grown rapidly, and estimates suggest that as many as
40 million people are infected today (UNAIDS 2006a). Levels of infection vary
greatly across and within many regions of the globe, an observation that has driven a
recent boon in research attempting to understand and explain how those differences
arise, and how understanding them could aid prevention efforts and help to reverse
the rise in new infections (Aral and Roegner 2000; Grassly et al. 2001; Green 2003b).
Longstanding scholarship establishes the role of networks both in the epidemiology 3 of infectious diseases, and in the diffusion of medical innovations and information
(Friedman et al. 1997; Morris 2004). To date however, the important contributions of
network theory and analysis have not been readily incorporated into the numerous
subfields that explore the relationship between their topic of interest and the
epidemic. For example, scholars have recently become interested in explaining the
role religious organizations may play in HIV prevention and intervention efforts
(Gray 2004; Trinitapoli 2006b; Trinitapoli and Regnerus 2006a; Trinitapoli and
Regnerus 2006b; Watkins and Chimbwete 2004). In this dissertation, I show how
taking a network approach to studying this relationship between religion and HIV can
fundamentally alter the way this relationship is addressed, and force some
reformulation of the subsequent findings and interpretations within such studies.
When related to the global HIV pandemic, religious organizations are
frequently painted as heavily entrenched monoliths, which either impede the potential
progress of prevention efforts (Caldwell, Orubuloye, and Caldwell 1999; Pfeiffer
2004), or are recognized as frequently the only organizations existing in the most
heavily afflicted areas (Green 2003a), and are therefore targeted as strategic partners
in the ongoing fight (Jenkins 1995; Liebowitz 2004; Parry 2003). To date, the
research that does empirically address the association between religion and HIV
proceeds largely via one of two approaches. In one approach, researchers have
examined the content of religious messages that address religious prescriptions for
avoiding HIV, such as avoiding pre- and extra-marital sexual partnerships (Garner
2000). Alternately, other researchers examine the multivariate (or sometimes
bivariate) relationship between differences in religious affiliation, or levels of
4 religious participation, and individuals’ HIV “risk behaviors” or actual HIV status
(Rankin et al. 2005; Takyi 2003; Trinitapoli 2006a; Trinitapoli and Regnerus 2006b).
Organizational Messages
Studies that focus at the organization level and those interested in individual
outcomes each approach the religion-HIV link by framing the question in terms of the
potential harmful or helpful potential of religion in stemming the HIV-tide, whether
conceptualized at the organizational or individual level. They select a particular factor
to investigate, such as an individual’s number of sexual partnerships, or the
willingness of religious organizations to promote condom use as a protective
measure. These outcomes are then compared across religious affiliations to see which
organizations are comparatively better at protecting their members than others.
Alternately, researchers compare HIV risk factors across levels of religious
participation to examine the effect of being more actively engaged with one’s faith
community. By framing the question in such either/or terms, researchers do not
consider potential mechanisms that may simultaneously produce beneficial and
detrimental effects coinciding with the same factors.
The HIV-related messages within religious organizations are frequently
described as singular perspectives that rarely vary within denominations (Gray 2004),
and are sometimes painted with such broad strokes as to apply to all religious
organizations (Caldwell et al. 1999). Congregations, for that matter – any level of
religious organization, do not however exist in a void, and this approach leaves little
room for the importance of congregational context. This fails to incorporate the
negotiation, fluidity and potentially complex nature of such messages within any
given congregation. Recent research demonstrates some of the faults in such 5 treatment - describing some observed variations that do exist, particularly between
formal and informal policies within particular religious organizations (Parry 2003;
Trinitapoli 2006b). To date however, researchers have not devoted much attention to
explaining the salience and source of such differences.
I demonstrate how studying the relationship between religion and HIV
through a structural approach reframes the nature of the questions asked and the
approaches taken in their study. In particular, in the sections of this dissertation
focused on organizational networks, I provide a model that accounts for some evident
sources of such differences, which are also explicitly tied to their importance, by
addressing the location more than the content of such messages. This approach lends
itself to simultaneously account for some of the differences that do arise across, but is
also better equipped to explain the differences that arise within, particular religious
organizations.
Individuals’ Risk Behaviors
Similarly, studies that are interested in the link between individual religiosity
and HIV-risk behaviors (Garner 2000; Takyi 2003) frequently address risk behaviors
in ways that do not incorporate many of the important recent advances of network
epidemiologic literature (Friedman et al. 1997; Morris 1993; Morris 2004). I build a
model that demonstrates the weak association between individuals’ “risk behaviors” –
which are known to vary by religious affiliation and participation – and network
properties of risk – whish has not gained much attention to date in the religion
literature. The result is an examination of the ways that religious organizations can
simultaneously protect most of their members against HIV while putting others at
extreme risk. The startling implication of this chapter is that the same “protective” 6 behavioral changes that have been modeled in previous literature are the very
mechanisms generating both potential protective and risk-increasing effects in the
models presented here.
Framing the Problem
The existing HIV/AIDS epidemic is one of the worst health crises facing the
world today. Estimating HIV’s prevalence is notoriously difficult, and historically has
been plagued by problematic methods, such as relying almost exclusively on
antenatal clinics in many developing countries (Boerma, Holt, and Black 2001; Obare
2005; World Health Organization and UNAIDS 2003). Noting these difficulties,
current estimates suggest that approximately 39.5 million people are infected with
HIV globally (UNAIDS 2006a). In 2006 alone, nearly 2.9 million people died of
AIDS-related causes, while approximately 4.3 million new cases arose (UNAIDS
2006a). The nature of the epidemic varies greatly by region, and has changed
substantially with time.
The epidemic disproportionately afflicts sub-Saharan Africa (SSA), with the
region accounting for roughly 24.7 million (62.5%) of the global AIDS cases
(UNAIDS 2006a), despite being home to only about 11% of the world’s population
(PRB 2006). National prevalence estimates vary widely within SSA, with some
countries estimated to have over one third of their population infected (Swaziland,
Botswana), ranging to a low of about 1.0% in Senegal. In total approximately 6.1% of
the region’s population is infected with HIV (UNAIDS 2006a; 2006b). Across the
region, prevalence rates in most countries are increasing, or remaining roughly stable,
though a few countries, e.g., Uganda, Kenya and Zimbabwe, have recently shown
preliminary evidence of prevalence declines (Green 2003b; UNAIDS 2006b). The 7 epidemic effects not only those who are infected or die, but also has produced a
substantial AIDS-orphan population, encompassing as much as 20 percent of the
child and adolescent population in high-prevalence countries in SSA (Case, Paxson,
and Ableidinger 2004; UNAIDS 2006b).
[Insert Figure 1.1 about here]
Malawi is a small landlocked country in southeastern Africa; see Figure 1.1.
Like most of SSA, Malawi has a generalized epidemic, in that it affects more than
one percent of the population, the gender ratio approximates 1:1, most infections are
through heterosexual sexual contact, though perinatal transmission is also common
(NAC 2004; NSO 2005). Malawi’s epidemic is estimated to be eighth highest in the
world, with approximately 14.1% of the population presently infected, and recent
estimates suggest prevalence has stagnated since 2000 (NAC 2004; NSO 2005;
UNAIDS 2006b). Infectivity varies substantially within Malawi, with higher rates in
the southern region than others, and in urban areas than in rural (Crampin et al. 2003;
Obare 2005). Estimates suggest infection among approximately eight percent of rural
residents in Malawi (Obare et al. 2007; NSO 2005).
Approaching the Study of HIV/AIDS
In the past few decades research has demonstrated the important role networks
play in the propagation of infectious diseases through a population (Friedman et al.
1997; Morris 2004), including HIV and other STDs in particular (Aral, Padian, and
Holmes 2005; Klovdahl 1985; Rothenberg and Narramore 1996; Woodhouse et al.
1994). Other work has further shown how network structure can contribute to
stagnating such disease diffusion (Darrow et al. 1999). Similar models have also been
8 applied to demonstrate the importance of indirect connections in the dissemination of
information both in general (Granovetter 1973; Valente 1995), and about health
promotion strategies in particular (Coleman, Katz, and Menzel 1966; Valente et al.
1997). To date however, other subfields that study the relationships between their
given interest and disease risk or prevention messages proceed largely disconnected
from the insights gained in such network studies. For example, these ideas, and the
subsequent reframing of questions that they infer have not been addressed in the
substantial literature linking religion and health related trends, beliefs, practices or
outcomes (Chatters 2000; Ellison and Levin 1998; Levin 1994).
The literature linking religion and health outcomes has a long history, largely
focusing on the beneficial impact of religious participation on risk-related behaviors,
and subsequently on particular health related outcomes (Chatters 2000; Ellison 1991;
Hummer et al. 1999). These changes in health outcomes result from changes in
particular behavioral differences between religious groups, such as decreased
likelihood of smoking (Strawbridge et al. 1997) or alcohol consumption (Ford and
Kadushin 2002). Others explore the development of social support and its outcomes
that coincide with religious participation (Idler 1995; Krause, Ellison, and Wulff
1998). Recent scholars point out that the link between religion and health related
outcomes is not necessarily unidirectional, and that continuing to frame them this way
potentially limits researchers’ ability to understand this important relationship. Idler
(1995) raised this possibility, particularly suggesting that if health outcomes are
studied over time, different associations may be observed according to religious
participation at different points in time. To date, however, no studies have shown that
these pathways can occur simultaneously. 9 Of more direct importance to studying HIV-risk, still others have observed
that religious affiliation is associated with declines in extramarital sexual partnerships
(Hill, Cleland, and Ali 2004) and delayed sexual onset (Rostosky, Regnerus, and
Wright 2003), but also coincides with lower likelihood of condom use in sexual
encounters (Agha, Hutchinson, and Kusanthan 2006; Bearman and Bruckner 2005).
Some researchers also focus directly on the role of religious organizations in the
development of HIV prevention and intervention messages (Hearn 2002; Liebowitz
2004; Parry 2003; Pfeiffer 2004). For example, in an attempt to capture the broad-
scale influence, Gray (2004) showed that, among the few indicators he modeled,
proportion Muslim was the only one negatively associated with HIV prevalence for
38 countries in SSA. The vast majority of studies explore the link between religious
participation and particular "risky" behaviors. Several studies show the association
between religious affiliation and delayed sexual onset (Agha et al. 2006; Hill et al.
2004). Garner (2000) for example shows that among four churches in Kwa-Zulu
Natal, South Africa, the Pentecostal church studied has lower levels of extra- and pre-
marital sexual partnerships than other congregations. Others question whether there is
a relevant connection between religious participation and declines in risky behaviors
(Lagarde et al. 2000).
In many areas of SSA, individuals report virtually ubiquitous religious
participation (Bedaiko 1995; Bedaiko 2000; Green 2003a). With roughly two-thirds
of the population in SSA living in rural regions (PRB 2006), and religious
organizations frequently being the only formal organizations existing in such rural
locations (Green 2003a), congregations play an important role in many aspects of the
lives of people in SSA (Agadjanian 2001; Englund 2003; Green 2003a). Individuals’ 10 relationships are frequently centered within their religious congregations
(Mkandawire 2000; Watkins and Chimbwete 2004) such that while content of many
religious messages frequently don’t vary substantially across congregations,
individuals’ level of embeddedness within them is a strong predictor of adherence to
these religious doctrines (Agadjanian 2001; Garner 2000). Similar to patterns
observed elsewhere in SSA, in Malawi, while the majority are affiliated with
“Mainline” Christian congregations (Jenkins 2002; NSO 2005), Pentecostal
denominations have seen recent rapid increases, particularly in urban areas (Englund
2003), and a small proportion (a majority in select districts) are Muslim (Feidler
2004; Jenkins 2002).
Resulting from the above observations, and recent research finding some
evidence for differences in HIV-prevalence by religious affiliation (Gray 2004; NSO
2005), researchers have recently begun to explore whether differences in religious
organizations’ prevention strategies can account for these observed differences. In
finding that Muslims have lower HIV-prevalence across 38 countries in SSA, Gray
(2004) attempts to extrapolate backwards to suggest religious doctrines of the Islamic
faith that should generate less risky behaviors, and may generate such differences.3
Religious organizations are well-equipped to promote messages of abstinence and
fidelity (the A & B of the popular “ABC”s of HIV prevention popularized in SSA;
missing only the promotion of condom use) (Green 2003a). Religious organizations
in SSA have been described as the “sleeping giant” and as “Invisble NGOs” with
great capacity to engage the HIV epidemic (Hearn 2002; Parry 2003), but evidence is
mixed about what they are actually doing, despite stark accusations of their failings
in the fight against HIV (Caldwell et al. 1999). Regardless of researchers’ particular 11 stance in the role religious organizations play in preventing or supporting prevention
efforts, existing evidence points to substantial variation in HIV prevalence estimates
by religious affiliation and levels of participation. Table 1.1, for example
demonstrates the differences by religious affiliation observed in HIV prevalence
estimates across the three research locations in the Malawi Diffusion and Ideational
Change Project (described in detail in chapter 3).
Balaka Mchinji Rumphi N HIV+ N HIV+ N HIV+ Catholic 110 13.89 156 7.48 90 5.62 Mission 64 11.48 155 7.69 206 6.09 Protestant Pentecostal 45 15.56 43 15.00 74 2.74 AIC 16 18.75 14210.29 179 8.43 Muslim 526 8.82 3 0 6 0 Other 34 8.82 1025.00 87 4.76
Total 795 10.47 6018.28 642 5.86
NOTE: Numbers presented are N – number of adherents tested, and HIV+ – percentage who tested positive according to 2004 MDICP VCT.
Table 1.1 HIV Status by Religious Affiliation MDICP-3
Moving from Either/Or to Both/And
The focus in most present research either examines the beliefs, behaviors and
risk of individuals or the content of messages within religious organizations. Each of
these approaches implicitly addresses both religion and HIV-risk as properties that
individuals (or groups) “possess” to varying degrees. This produces a corresponding
conceptualization of the religion-HIV connection that is fundamentally based an
“either/or” logic, which seeks to determine whether religion is helpful or harmful for 12 HIV-prevention. At the individual level, if researchers in this area instead begin to
think of HIV-risk in terms that move beyond "risky" behaviors, and instead focus on
the structural properties of risk, we can arrive at a mechanism, which simultaneously
explains detrimental group level changes that arise from helpful individual behavioral
adaptations. Similarly, if researchers address the positions and messages of religious
organizations not as something that inhere in the organizations but develop, fluctuate
and are constructed within particular contexts, we can reach a more dynamic picture
of religious organizations’ contributions (and barriers) to prevention efforts.
With the focus on individuals and the beliefs they hold, or the behaviors they
engage in, and organizations and the positions they espouse, present research largely
engages questions of how well (or how poorly) religious affiliation or participation
contributes to particular outcomes that are framed as especially salient to the present
scope of HIV in SSA. Numerous researchers examine how particular religious
affiliations reduce extra- and pre- marital sexual partnerships better than others, or
how increased levels of involvement in those organizations contribute to similar
declines (Agha et al. 2006; Garner 2000; Hill et al. 2004). Researchers have also
examined how readily religious organizations are engaging in care of those already
living with AIDS (Liebowitz 2004; Pfeiffer 2002). Similarly, researchers have
addressed the contributions of religious organizations to proliferating or stemming
stigmatizing beliefs about HIV (Green 2003a; Gunter and Hue 2000; Herek,
Widaman, and Capitanio 2005) and the uptake of condom use (Agha et al. 2006).
By applying a social network perspective, I demonstrate how the questions
change, findings must be reinterpreted and implications should be reconsidered. I
suggest mechanisms by which single actions, and observed trends can each 13 simultaneously serve to improve protection efforts and increase risk. In particular, at
the organization level, I suggest here that it is important to understand the contexts
within which religious congregations develop their HIV-related messages both to
remove some existing presuppositions about the nature of their contribution to the
ongoing fight against HIV in SSA, but also to investigate if the existing prevention
and intervention strategies rely on organizational structures as they exist. This moves
beyond the assumption in present research of strong links between intended and
actualized content of messages; addressing it instead as an empirical question.
Similarly, at the individual level, I show that moving from models based on the
epidemiology of non-infectious disease-risk to network models that more readily
match the epidemiology of HIV, forces reconsiderations of presumed knowledge of
the link between religion and individual-risk.
Outline of Dissertation
In this dissertation, I therefore demonstrate several advantages of investigating
the relationship between religion and HIV/AIDS in SSA as a relational problem,
approached from the perspective and methods available through a social networks
frame. The findings presented here stand in contrast to some of those gained via
individual or variable-centered approaches presently used to investigate the link
between religion and HIV-risk, or knowledge about AIDS-prevention strategies. I
instead demonstrate what changes result from moving the focus from these actors to
the relationships between them.
The chapters that follow develop the above argument further and provide
several empirical investigations of the importance of religious networks in the nature
of the changing HIV/AIDS epidemic as it presently exists in SSA. In Chapter 2, I 14 overview some of the fundamental theoretical and methodological differences that
come with the shift to a relational focus of the questions being addressed. In this
chapter, I alter a longstanding challenge in social science based on methodological
individualism to demonstrate how methodological structuralism is more adept to
answer the questions derived from relational questions about the religion-HIV link.
While this perspective is fundamentally opposed to individual, organizational or
variable-centered approaches to studies of the same topics, I both build on these types
of work, and demonstrate how this approach fundamentally shifts the questions,
results and implications of existing research.
Chapter 3 summarizes the two large-scale data collection projects from
which the data for the empirical investigations of later chapters are drawn. In addition
to simply describing the content of these data sets, I also include in chapter 3
evaluations of the processes used in collecting these data and several analyses of data
quality.
Chapters 4 through 6 are the primary analytic chapters of this dissertation.
Chapter 4 provides a description of congregational leaders’ AIDS-discussion
networks, focusing particularly on the differences between formal and informal
channels of message development. In particular, I summarize these congregation
leaders’ descriptions of three types of network relationships – their co-participation
with other congregations in events like revivals and fellowships, conversations
regarding general doctrinal issues, and conversations regarding HIV/AIDS-specific
topics. I demonstrate the strong local clustering of the networks within which
congregation leaders’ relationships exist, and their policies are formed.
15 In Chapter 5, I examine the similarities and differences that exist between the
HIV-related discourses of national/denomination leaders and local/congregation
leaders, focusing on the increased importance of informal messages at the local level.
In particular, I summarize the congregation leaders’ descriptions of three HIV-related
topics – their perception of the scope of the HIV problem, their evaluation of the
source of the HIV problem, and the prescriptions they suggest to their congregation
members to protect themselves from contracting HIV. I demonstrate that local
congregation leaders’ descriptions for each of these are more readily shaped by day-
to-day interactions with their parishioners than are formal proscriptions constructed at
the national level. Combining these chapters suggests that religious organizations in
rural Malawi are congregationally organized, as is increasingly becoming the
approach to studying religious organizations in other contexts as well.
In Chapter 6, I draw on observed data establishing the differences by
religious affiliation in numbers of reported sexual partners to simulate a series of
networks to estimate network measures of HIV-risk. I use these networks to evaluate
the possibility for number of partners alone to influence the simulated likelihood of
being in monogamous relationships, smaller connected components, or in
components with individuals who are HIV positive. I find that observed differences in
number of sexual partners – which do vary by religious affiliation – do not produce
corresponding changes in risk-network position, and in some cases can even
correspond to increased risk.
In Chapter 7 I summarize the resulting changes in our understanding of the
relationship between religion and HIV/AIDS in SSA gained by moving to a relational
approach to their study. I suggest a number of implications for how these findings 16 could be extended into other contexts in SSA (beyond the rural-Malawian focus of
the present project), as well as into the study of other theorized relationships – such as
economic or gender considerations for HIV risk. I also address how the findings and
discussion presented here would in-turn inform present organizational and legislative
prevention strategies, which in recent years have increasingly funneled through
religious channels.
17 NOTES
1 This is not the actual name of the person whose funeral I describe here. I selected a pseudonym that is commonly used in Chichewa for both men and women - Madalitso, which means “blessings.”
2 While I am unaware of the actual translation of this term, it is used to summon images of someone who “moves around” a lot, like a mosquito flitting about. While its literal translation has nothing to do with sex, it directly infers sexual promiscuity.
3 It should be noted that many of the risky behaviors which Gray (2004) relates to the observed differentials in HIV prevalence by country are not behaviors directly tied to risk of contracting HIV, with most of his attention focused on alcohol consumption. This adds an additional layer of speculation about the mechanism generating the observed differences to his explanation.
18 19
Source: (Gerland 2006: 9)
Figure 1.1 Malawi Map, Highlighting MDICP Research Locations
19
CHAPTER 2
THEORETICAL ORIENTATION: MOVING FROM PEOPLE AND VARIABLES
TO RELATIONS AND LINKS
Actions have a constrained possibility to affect particular (desired) outcomes.
How those actions are constrained has been the topic of sociological research since
the discipline’s inception. In the Malawian context, when addressing the potential
strategies for engaging the existing HIV epidemic, these constraints have substantial
control over the range of changes that particular behaviors can produce. This
structure-agency debate has long influenced the way that social researchers address
the questions they study. In this dissertation, I examine some of the social network
based structural properties that effectively detach prevention strategies and related
behavioral adaptations – which are influenced by religious organizations and
individual religious participation – from their intended effects of altering the disease’s
present trajectory. In particular, I demonstrate the importance of structural
configurations within religious denominations that constrain the effectiveness in
development and implementation of top-down intervention models; and explain how
sexual marketplace patterns restrict the potential impact of individuals’ religiously-
associated partnering decisions for diminishing HIV-risk.
20 In this chapter, I develop the a strategy used in the rest of this dissertation,
which moves the focus of research from individuals and organizations, or the
characteristics they “possess,” to the spaces that separate them, and the connections
that fill those spaces. By drawing on this structural frame and several of the
extensions available through Social Network Analysis, I ask a series of fundamentally
different questions than those presently at the heart of investigation in studies of
religion and HIV in SSA. As a result, what I find here both reinterprets some of the
existing findings about this relationship, and presents new information about what
role religious organizations are playing in combating the epidemic that were simply
beyond the scope of previous analytic approaches.
In this chapter I describe how a social network frame alters the questions we
address - through shifting our notion of where social research should focus, and
implementing the corresponding methodological and theoretical alterations that result
from this shift. I build on an alteration of a standing challenge made by James
Coleman to better link micro-level studies to the macro-level outcomes that interest
sociologists. I do this however by taking a different starting point, which also alters
the nature of Coleman’s challenge.
Background
One of the fundamental debates in sociological research pits the relative
importance of structure against agency. Building on Weber’s, and later scholars’
interest in understanding behavior, the agency of individual choices and actions has
long been situated in the cross-hairs of social researchers. A strict rational choice
theorist, for example, builds complex theories of interaction that address macro-level
21 trends as aggregations of individual choices. Conversely, there are scholars who start
from the opposite pole, and primarily address societal and structural constraints they
perceive to restrict individuals’ actions – and their potential outcomes – to “options”
that are, in essence, predetermined.
The poles of this debate are clearly delineated – individuals’ actions are either
individually motivated or completely constrained, and those actions can have either
full or no effect on desired outcomes. Table 2.1 highlights the combinations of these
dimensions, including the extremes (high-high and low-low). While the approaches
underpinning most sociological research fall somewhere in the upper-right or lower-
left panels of this table, perhaps the best way to understand the implications of those
positions however is to examine the extremes (in the upper-left and lower-right cells).
[Insert Table 2.1 about here]
In the structural determinism cell, actors are not free to make choices, and
those choices they perceive to exist are false. Additionally, their (in-)actions have no
impact on the predetermined outcomes; actors have no recourse to alter outcomes.
Conversely, in the free-choice cell, actors are entirely free to make whatever choices
they desire, and those actions – and not any other factors – directly generate the
experienced results. Much of social science takes an approach that more readily
reflects the intermediate combinations presented in Table 2.1. This includes the
embeddedness approach, which focuses on the possibility that actors make decisions
that have substantial influence over their experienced results – but do so from options
that are strongly structurally constrained and therefore have a limited range of
possibilities. The final option is an appearance of choice, while the variations in those
22 selections have little-to-no subsequent impact on the actual outcomes to which they
appear to be tied. In other words, in practice, the upper-right corner in Table 2.1, with
respect to how individual agency contributes to macro-level outcomes looks much
more like structural determinism than it does either of the two approaches. To
foreshadow, the approach in this study will largely contend that the contexts and
topics at the heart of this dissertation, at least for rural Malawi, fall closer to this last
“constrained choice” option than any of the others presented in Table 2.1.
This builds from substantial empirical evidence that demonstrates the
constraints around HIV-risk related choices available to rural Malawians. Virtually all
adult Malawians are married (NSO 2005), while the few who are not generally want
to marry, and expect to do so soon (Poulin 2006; Clark et al. 2006), Recent evidence
suggests some increasing delays in timing of first marriage may be increasing the risk
of adolescents contracting HIV (Bongaarts 2007), though to date, very few
adolescents appear to be contracting HIV prior to their first marriage (Clark et al.
2006; Poulin 2006; Obare 2007) – i.e., failure of abstinence messages do not appear
to be substantially contributing to the HIV epidemic in rural Malawi. Furthermore,
dissolved marriages (which may increasingly be used as a method of negative partner
selection, Reniers 2006), almost universally lead to remarriage (Reniers 2003). There
is virtually no evidence for the existence of sexless marriages, despite recent evidence
suggesting a high majority of women’s HIV-infections are taking place within
marriage – within which condom usage is extremely rare (Clark 2004; Reniers 2006;
Trinitapoli 2007). The scripts available therefore leave little room for Malawians to
remain sexual isolates. In other words, the vast majority of Malawian adults are
23 involved in relationships where unprotected-sex is the norm. As such, latter sections
in this dissertation (see especially Chapters 6-7) will address potential prevention
strategies that move beyond simply targeting individual behavior changes, which to
date have proven largely ineffective.
While sociological theory has long incorporated structural mechanisms into its
explanatory models, these have only found their way into methodological analyses on
rare occasions. Sociology, despite its interest in social phenomena, for the past half-
century (or longer) has largely remained methodologically individual-istic.1 Both
qualitative and quantitative studies face this limitation – qualitative work with its
attachment to thick description of individual cases, and quantitative studies, at least
implicitly so, in its requirements of independence between cases – upon which most
statistical analytic techniques are based. James Coleman suggested re-linking this
prevalent methodological individualism in sociology to macro-sociological structures
as one of the core “intellectual hurdles” for modern sociologists (Coleman 1986); see
relation-3 in Figure 2.1. The structural framework that underpins social network
analysis provides one framework for reconnecting sociological empirical work with
its macro-social interests, which can perhaps best be seen through a reorganization of
the Coleman “boat” – presented in Figure 2.1.
[Insert Figure 2.1 about here]
Much existing work in SNA has been interested in meeting Coleman’s
challenge, attempting to complete the Macro-Micro-Macro connection (we can refer
to this as the A-I-A – mAcro-mIcro-mAcro – cycle). They do this largely by
addressing questions from the same direction as existing individualistic studies,
24 namely by placing the individual at the center of focus in research design. Such
studies may conceptualize networks as (1) properties of individuals to be measured as
variables, (2) network-based theoretical explanations for individual gains, or (3)
particular methodological approaches applying network strategies to examine yet-
individualistically motivated questions. In effect, these studies have accepted
Coleman’s challenge, and have proposed networks as the way to satisfy the
completion of the third phase of the A-I-A cycle, with network variables accounting
for the first and third transitions in his description of multi-level relations (“1” and
“3” in Figure 2.1), while network oriented methodological approaches and theoretical
propositions satisfy the second (“2” in Figure 2.1).
I provide a brief overview of the foundations of SNA here, while more
thorough descriptions are available elsewhere (Freeman 2004; Mullins and Mullins
1973; Scott 2000; Wasserman and Faust 1994; Wellman and Berkowitz 1988). Many
scholars in the area of SNA trace their roots back to early work in sociometry,
particularly the work of Jacob Moreno (1934). Many of the earliest works in this area
focused on trying to develop general forms to describe relational structural patterns
across differing kinship systems (White and Jorion 1992; White and Jorion 1996;
White 1963). A similar interest survives today in the way of interest in extracting
common structural traits that can explain a wide range of social phenomena within
parsimonious relational models. Perhaps the most widespread use of such models is
found in research on the diffusion of information, of which Coleman himself is one of
the pioneers (Bertrand 2004; Coleman, Katz, and Menzel 1966; Rogers 1995; Valente
and Davis 1999). There has been vast amounts of other work in areas such as disease
25 transmission (Morris 2004), the development of social capital (Granovetter 1985; Lin
2001), analysis of friendship (Moody 2001; Zeggelink, Stokman, and Van De Bunt
1996) and publication patterns (Moody 2004; Newman 2001). Each of these
approaches the study of networks as a theoretically framed “orientation” for thinking
about the world that infers a method rather than as a methodological analysis strategy
alone.
To date, innovations in SNA’s theoretical framework and its empirical
findings have developed independent from the numerous substantive sub-disciplines
that exist within modern sociology. This has happened largely because SNA is
viewed by those from outside its practice largely as a methodological alternative to
more traditional research strategies.2 Further, much of current work that includes
network variables and theories or relies upon network methodologies among their
analytic strategies, still rarely fully capture the implications of the shift in focus which
underpins these initial structural analyses. Instead much of this work substitutes
network concepts into models as better ways of conceptualizing and measuring
properties of individuals. As I suggest here,3 SNA offers more than a new set of
variables to be introduced into individual models, an alternate set of theories for
social scientists to examine, or a new methodological approach to be substituted for
regression analyses (or other techniques).
Examples of existing studies that approach “networks” as properties or
variables of individuals, come from studies of individuals’ social capital (Coleman
1988), friendships (Adamczyk and Felson 2006; Marsden 1987; Zeggelink 1995) or
discussion partners (Behrman, Kohler, and Watkins 2002; Gerland 2006; Kohler,
26 Behrman, and Watkins 2000; 2001; Marsden 1987). Similarly, sociologists study
individual-based questions by employing networks as a methodology to capture
influences on those individuals (e.g., Haynie 2001) or the context within which they
are studied (Adimora and Schoenbach 2005). Finally, network theories have been
applied to a number of individual level-studies, perhaps the most famous of which is
Granovetter’s study showing the importance of weakly-connected relationships in
successful job-seeking (Granovetter 1973). In line with Coleman’s (1986) approach
however, each of these takes an approach that places the individual at the center of
focus – as the starting point for study.
A Structural Social Science
Bruce Mayhew (1980; 1981), argues that a structuralist framework is the only
truly sociological enterprise. He goes so far as to suggest that structuralists are not
concerned with “the individualist concern with ‘human behavior’… [s]tructuralists
have nothing at all to contribute to the non-problems individualists have posed for
themselves” (Mayhew 1980: 357). In this dissertation, I leave more room for
exchange between individualist and structuralist scholars than Mayhew, but do agree
with his assessment that the two approaches ask fundamentally different questions,
“make radically different assumptions, conceive of social phenomena in radically
different terminologies, and generally approach any topic from opposite directions”
(Mayhew 1980: 345).
A Network Paradigm
Wellman and Berkowitz (Berkowitz 1982; Wellman 1988; Wellman and
Berkowitz 1988) have argued for the treatment of SNA as a paradigmatic shift that
27 moves beyond the existing dominant individualistic framework in much of empirical
and even theoretical sociology. The defining characteristics of a scientific paradigm
according to Thomas Kuhn (1970) are cumulation of knowledge and a fundamental
reconstruction of prior understanding, both of which are found in SNA. Linton
Freeman, in his book accounting for the historical development of SNA as a field,
argues for the first of these (Freeman 2004: 6). Scholars who have adopted the SNA
framework have addressed the second of these questions (see summary above), but
those outside of this community have not readily accepted the relevance of this point.
Many of the advances of SNA to date have not been seen by the rest of the
sociological community in their full implications. Only by applying an SNA
perspective to new sets of questions will this limitation be overcome, and the full
potential of its contributions to the nature of social science as a whole be fully
appreciated. While Mayhew claims that relational, or as frequently termed in SNA
literature – structural, approaches to social science are the only truly social approach;
I temper his suggestion slightly. SNA offers a fundamentally different way of
examining the social world that moves research away from its present individualistic
assumptions. Perhaps through better incorporation of these ideas into mainstream
sociological efforts, we can better understand their impacts on commonly held
assumptions and, to-date-unquestioned empirical findings interpreted as social facts.
In his Introduction to Structural Analysis, Berkowitz (1982) summarizes the
main thrust of the structural approach as shifting the researcher’s focus from points
and individuals to the spaces and relations between them. This seemingly benign shift
has deceptively grand implications, which are not captured by the individualistic
28 approaches to social science which continue to dominate present sociological work
(Freeman 2004: 1). Returning to the Coleman’s problem, this suggests an alteration
to, and ultimately an inversion of his theoretical “boat”. First, Berkowitz’s notion of
taking as primary not the individual, but the relation, necessitates altering Coleman’s
“boat” by re-labeling the levels of analysis from micro and the macro to foci on node
and meso (or relational) levels. With the micro-macro distinction, there is a
fundamental implication of scope in the difference between the two levels of analysis.
No such distinction in the level(s) of aggregation is necessary in the altered version.
Instead of the size or aggregation focus of the macro-micro distinction, the alternate
version differentiates along lines of location. The nodes (whether individuals,
organizations, or other entities) and the description of their properties occupy one
“level”, while the spaces between them and the properties of the relationships that fill
those spaces occupy the other.
Whereas in Coleman’s version, the methodological individualism found in
much of social science is accepted and incorporated into the model, the second
alteration to his scheme instead places the focus of empirical research at the meso
level. This is an inversion of the order of explanation found in Coleman’s model. In
other words, a researcher’s interest is not focused on the size, scope or traits of the
nodes, but on the relationships between them. The pattern of relationships between
individuals, organizations, nations, etc. can all be identical, even when the nodes are
very different. With a focus on the relationships, and organization of relationships
between nodes, and not on the nodes themselves, properties of those particular nodes
(including their scope) are no longer the fundamental organizing principle behind
29 research strategies. Figure 2.2 presents this shift visually – as in Coleman’s version,
the level to be directly analyzed is on the bottom.
Relations as Starting Points, not Variables
The question then to address here is – “Upon what do social scientists then
affix their focus?”. Traditional social science defines actors and their various
characteristics, then groups and levels of groupings that are said to be relevant for
describing similarities, differences or even processes that arise (or exist) between
them. The key to the SNA approach is its focus on relationships and relationship
patterns instead of a focus on individuals, organizations, nation-states, or whatever
other “points” are commonly at the center of social-scientific research – frequently
described as methodological individualism. It is a fundamental inversion of the way
that social science has typically progressed. Perhaps the most important factor this
introduces to social science research is the recognition that "the elements of those
subsystems upon which structuralists base their definitions of groups are not
independent" (Berkowitz 1982: 14) as is fundamentally assumed in more traditional
methodological strategies. SNA is fundamentally interested in explaining how social
phenomena arise from node inter-dependencies.
[Figure 2.2 about here]
Figure 2.2 helps to explain this shift. A structural approach to studying social
phenomena takes the first meso-level relationship labeled in the bottom-left of Figure
2.2 as its focal beginning point. This focus on the meso-level moves away from the
individual (or node-level) focus common to most social science, and inherent in
Coleman’s model. Contrary to a purely structural approach to social science (Black
30 2000a; 2000b; Tucker 2002), SNA does not remove individuals from focus
altogether, but by placing the focus on meso-level relations, addresses individual and
organizational relationships in the same manner. While the Coleman model presumes
a link between individual beliefs and behaviors; such behavioral motivations are not
the focus of SNA. This is especially salient since researchers frequently dismiss its
feasibility for macro level “actors.” An SNA perspective of social structure therefore
focuses on questions of relational position and aggregation, rather than on presumed
boundaries based on formal organizations, or shared individual characteristics. While
individualistic approaches focus on the aggregation of individual characteristics into
groups that are presumably relevant, SNA is concerned with how relations aggregate
into networks, which can represent links between individuals, groups, or other macro-
level nodes.
An SNA researcher’s focus is therefore particularly suited to observe links
between (meso-level) relational structures, as illustrated in the next step in Figure 2.2.
(the transition labeled with the “*”). These can focus on how on relational structure
changes over time, or how one structure influences another. Recent work has begun
to investigate the importance of transitions in network phenomena. For instance,
while measures of social network properties frequently focus on static properties,
such as cohesion, centrality, degree, etc. (see Scott 2000; Wasserman and Faust
1994); scholars have recently begun exploring the ways that these properties can be
dynamically evaluated (Moody 2007). The implications for topics such as the
diffusion of ideas or diseases are direct, and are similarly important for other topics.
The transitional relationship presented in Figure 2.2 (*) can therefore represent the
31 impact of one type of meso-level relationship on another – e.g., the impact of marital
or trade relationships on ruling authority (Padgett and Ansell 1993) – or on these
dynamical shifts in network properties (Moody 2007).
As Mayhew contends (1980; 1981), SNA scholars therefore fundamentally
address different research foci than scholars who are engaged in research targeting
more traditional “actors” (whether individual or organizational nodes). However, this
does not leave the approaches unable to build on one another as Mayhew, and others
(Black 2000a; 2000b; Tucker 2002) would suggest. Instead, the links in Figure 2.1
between macro-structure and individual behaviors (1), or individual outcomes (2), are
still observable, and can be provided important explanations in the revised Figure 2.2,
but are now no longer questions across scope (i.e., macro-micro), rather explicitly
focus on nodes and the relational structure within which they are embedded. Whereas
Coleman’s boat implicitly assumes the importance of Link 2 as a connection in the A-
I-A chain, such a link is not required (though not removed) from the revised version
of the figure. Therefore in situations where behaviors are constrained in such a way
that choice is limited, or that individual behaviors have little control over the
outcomes they are intended to produce, a focus on meso-level dynamics – whether
relating two different types of relationships, or a single relationship at different points
in time – provides a mechanism by which to explain such a detachment. Perhaps most
importantly, this approach removes the necessity of establishing the link between a
node’s beliefs and it’s behaviors, but rather focuses on how readily a node’s
behaviors can potentially contribute to node-level outcomes (the focal question
presented in the agency-structure question as introduced with Table 2.1). Again,
32 inverting our focus, and addressing first the spaces between nodes (and the
relationships that fill them), reverses the nature of the problem proposed by Coleman.
It does not remove the importance of establishing the link between node-level
behaviors and meso-level relationships, but instead simply shifts our focus to the
fundamentally social aspect of the question, leaving how that then affects individual
properties and outcomes as a secondary consideration.
Finally, Figure 2.2 therefore provides some clarification on the initial
discussion from Table 2.1. The contribution of a focus on meso-level relationships
that are not necessarily bound to any particular individual-level properties (focusing
on * in Figure 2.2) can allow explanations of phenomena that fall in Table 2.1’s cells
2 and 4 in ways that simply are not possible if we focus only on individual traits, or
the aggregation of individuals. By focusing on structural characteristics that are, in
fact, relationally structural, and not simply structural proxies built on aggregated
individuals, we can develop mechanisms of explanation that are not otherwise
possible. Below, I elaborate how this directly changes both the questions addressed in
this dissertation, and therefore the answers when researchers attempt to explain the
link between religion and HIV.
Link to Empirical Work
The aims of the chapters that follow in this dissertation are therefore to apply
this relational perspective to improve our understanding of the relationship between
religion and the present HIV/AIDS epidemic in sub-Saharan Africa. By drawing on
my involvement in several large-scale projects in rural Malawi, I both reconceptualize
33 the way religion has been incorporated and how HIV-risk has been operationalized in
previous studies interested in this particular relationship. In particular, I address the
construction and dissemination of HIV-prevention related messages within religious
organizations and investigate how readily religious affiliation shapes HIV-risk
networks. Each of these questions moves away from treating messages, religious
organizations, their doctrine, and HIV risk as node-level properties, which has been
the exclusive approach of existing research on these topics.
Instead, SNA offers a fundamentally altered way of thinking about the nature
of social relationships that should substantially shift the approach to many questions
of social scientific interest. For example, in recent years researchers have shown a
vast interest in the link between religious affiliation and participation and health
outcomes. The SNA perspective can offer substantial insights into reformulating
researchers’ conceptualizations both of religious organizations and of health related
risks. In particular, some researchers have recently moved away from incorporating
religion as a merely individual level concept, even in individual level studies, opting
instead for a “moral communities” hypothesis of religious influence (e.g., Regnerus
2003; Welch, Tittle, and Petee 1991). An SNA approach to the study of religion
would offer a structural approach to investigate those communities that is not
available under the research strategies currently employed which must rely on proxies
or aggregation to approximate community effects. The importance of an SNA
approach to studying health related risks is similarly substantial. A voluminous
34 literature has recently developed in this area, particularly as it relates to the
transmission, spread and prevention of infectious diseases (e.g., Morris 2004). This
dissertation however is the first work to apply this perspective to each of these areas
simultaneously, and contributes some of the first extensions of the SNA paradigm
into sub-field analyses not primarily interested in the networks themselves.
SNA, Religion and HIV in Rural Malawi
I focus each of the three empirical chapters (4-6) on investigating the
relationship between religion and HIV prevention strategies – both their content and
their capacity for success – via the relational focus presented above. By addressing
the meso-structural context of the construction of religious HIV-prevention messages,
and the structural properties of sexual partnering, I shift the focus away from “what”
religious organizations say and an individual’s “risk behaviors,” to address the spaces
where those messages are constructed and the alters with whom those behaviors take
place. This shift produces a substantial alteration to our understanding of
organization-level responses within religious denominations and how readily those
organizations contribute to their participant’s risk contexts.
In Chapter 4, I detail the relational patterns of local congregational leaders to
first investigate not what they say about HIV, but where they develop those messages.
By first setting these leaders’ relational contexts as the focus, I am able to evaluate
what we should expect these leaders to be doing in response to the epidemic. Existing
work largely treats denominations as having high internal-consistency in their
35 responses, and therefore addresses what a denomination says. If the relational
structure within these organizations is incorporated into prevention strategies at all, it
is to assume hierarchical centralization, suggesting top-level leaders’ offices as the
launching point for the development and dissemination of any prevention strategies.
To date, little is known about whether this assumption matches reality, and any
breaks from this model could readily undercut the presumed internal consistency for
those studies that do not consider relational structure at all. In essence the shift in the
question here moves the research focus from “1” in Figure 2.1 to studying the initial
meso-relations (lower left) in Figure 2.2. This chapter primarily focuses on describing
the social and physical distances between relationships that congregation leaders draw
on in developing their response to the HIV epidemic.
The next chapter then turns to evaluating the similarities and discrepancies
that exist within denominations regarding how the present HIV epidemic is viewed
and what religious organizations are attempting to do in response. While many of the
descriptions in this chapter display substantial variations between local congregation
leaders and their corresponding national representatives, the differences that arise are
largely directly connected to the structure of the leaders’ relationships described in
chapter 4. Local congregation leaders aren’t drawing on national leaders’ opinions or
edicts, because they aren’t among the salient members of congregation leaders’
networks. Further, the distance that exists between national leaders and the lives of
individuals being impacted by the disease creates a response among national leaders
that is itself largely detached from the nature of the epidemic. This chapter moves the
36 focus from describing the content of messages as a property of individuals – a focus
on the “individual” beliefs represented in Figure 2.1, to a comparison between
discourse across varying levels within a religious organizations – focusing on the “*”
in Figure 2.2. Each of the chapters described above suggests that if religious
organizations are to continue to be engaged by donors and strategists to improve
potential successes in the fight against HIV in SSA, future efforts should engage not
“religious organizations” or denominations as wholes, but should address their
relational patterns in a way that more readily matches the local-clustering observed.
In the final empirical chapter here, I move away from measuring individuals’
HIV-risk solely as a property of their own behaviors. Existing literature demonstrates
a substantial relationship both between religious participation and HIV-infection as
well as between religious participation and presumed “risky” behaviors. Little is
known however, about the context within which those behaviors take place. I
simulate a series of networks based on the differences we do observe in number of
partners by religious affiliation, to demonstrate that where those relationships fall in
the larger sexual networks can prove more important than the number of such
partners. This chapter demonstrates one way that Link 3 in Figure 2.2 can suppress
any potential association represented by Link 2 in the same figure, and similarly can
produce an inability to produce any meaningfully observable association represented
by the third link in Figure 2.1. This shift is particularly relevant in that Coleman’s
model, and any research strategy employing traditional individual-focused techniques
would not have a mechanism by which to account for what I demonstrate in the
37 chapter.
Each of these chapters focuses first on relational patterns (meso-level) rather
than on individuals or organizations (node-level). In doing so, I demonstrate the ways
that much of the existing literature interested in the link between religion and HIV in
SSA limits the possibilities of explanation in the approaches they take. Further, this
approach moves beyond the potential dead-end of negative findings in research that
takes the individual-focus. For example, finding a weak association between
individuals’ “risky” behaviors and actual risk of contracting HIV, leaves any
prevention strategies targeting changes in individual behaviors virtually powerless to
evince meaningful change. The strategy in this study helps us not only to move the
focus of how we study what is already taking place for prevention efforts, but begins
to provide a frame of reference for what prevention efforts could be more effective
than those that presently key on individual factors alone.
38 NOTES
1 I should note at this point, that many sociologists do not study individuals only, and many of the insights from organizational studies will be incorporated into the later portions of this dissertation. However, even the ways that organizations are studied frequently imposes organizational-level “individuality” into the modeling techniques applied, where organizations are thought of as independent “actors”. For one substantial exception, see organizational ecology (Baum et al. 1996; Hannan and Freeman 1977).
2 As described elsewhere (Freeman 2004), this is perhaps best evidenced by its inclusion in the Mathematical Sociology section of the ASA.
3 The paradigmatic treatment of SNA suggested here is consistent with the work of several others. See especially Berkowitz (1982), Freeman (2004), Wellman (1988) and Wellman and Berkowitz (1988).
39
Actions’ Control Over Outcomes High Low Actors’ Control High 1 - Free-choice 2- Constrained Choice over Actions Low 3- Embeddedness 4 - Structural determinism
Table 2.1 Structure Versus Agency in Actions and Outcomes
40
Macro level Structure Structure
1 3
Individual 2 Behavior Micro level Values Orientations
Source: Coleman (1986: 1322)
Figure 2.1 James Coleman’s “Boat” – Macro-Micro-Macro Relations and Methodological Individualism
41
“Individual” 2 Node level Behaviors “Individual” Outcomes
1 3
Meso level Meso * Meso Relations Relations
Figure 2.2 Network Rhombus – Meso(-Node) – Meso(-Node) Relations and Methodological Structuralism
42
CHAPTER 3
DATA: THE MALAWI RELIGION (MRP) AND
MALAWI DIFFUSION AND IDEATIONAL CHANGE (MDICP) PROJECTS1
Malawi – The Context
Malawi is a religiously diverse country, its AIDS epidemic is typical of the
rest of the region, and religious organizations, primarily congregations, are a central
component of rural life where the majority of Malawians live. These factors make
rural Malawi an ideal setting for examining the role of organizational networks in
responding to the AIDS crisis more closely. As is characteristic of the pandemic
across most of SSA, the HIV epidemic in Malawi is a generalized one – that is, the
spread of the disease occurs primarily through heterosexual transmission, the male to
female infection ratio approximates 1:1, and perinatal transmission is common (Green
2003b). Malawi’s epidemic is estimated to be eighth highest in the world, with
approximately 14.1% of the population presently infected, and recent estimates
suggest prevalence has stagnated since 2000 (NAC 2004; NSO 2005; UNAIDS
2006b). There is, however, wide variation across testing sites – from 2.9 percent to
43 35.5 percent – suggesting that some areas have been more successful in avoiding
infection than others (NAC 2004).
The vast majority of Malawians are either Christian or Muslim. Figures for
Malawi from the World Christian Encyclopedia (Barrett, Kurian, and Johnson 2001)
suggest that 77 percent of the population is Christian, 15 percent Muslim, and most of
the remainder practice traditional African religions (eight percent). Malawi differs
only slightly from AIDS-belt countries in eastern and southern Africa in its
proportion of Christians (e.g., 82 percent in Zambia, 83 percent in South Africa) but
has a higher proportion of Muslims than most. The major Christian denominations as
a percent of the total Christian population are Roman Catholics (25 percent), mission
Protestants (20 percent), and African Independent Churches or AICs (17 percent);
groups like evangelicals and Pentecostals are rapidly growing in Malawi, particularly
in urban areas, and together account for about 32 percent of the country’s Christians
(Jenkins 2002b). The religious composition of Malawi’s rural areas differs somewhat
from the national figures, with Muslims comprising a majority in the South and
Mission Protestants being dominant in the North.
The Malawi Diffusion and Ideational Change Project (MDICP)
The MDICP is an ongoing longitudinal household survey collected in four
waves (1998, 2001, 2004 and 2006) that examines how individuals’ ideation,
behavior and HIV-risk are shaped through informal discussion networks. These data
focus on three distinct rural districts of Malawi: Balaka in the south, Mchinji in the
central region, and Rumphi in the north (in red in Figure 1.1). The MDICP was
initially designed to examine two key empirical questions: the roles of social
44 interactions in (1) the acceptance (or rejection) of modern contraceptive methods and
of smaller ideal family size; and (2) the diffusion of knowledge of AIDS symptoms
and transmission mechanisms and the evaluation of acceptable strategies of protection
against HIV. While the MDICP sample was intended to represent the populations in
the three sampled regions and not necessarily all of Malawi, it does closely resemble
one nationally representative sample (the Malawi Demographic and Health Survey,
MDHS – NSO 2004) on several key factors such as age, education, and select
indicators of socio-economic status (Watkins 2004; Watkins et al. 2003; Anglewicz et
al. 2006). The sample represents 119 villages and includes roughly 1500 ever-married
women and 1000 of their spouses in each of the interview years (1998, 2001 and
2004). Researchers have shown favorable evaluation of these data's reliability,
attrition and representativeness (e.g., Watkins and Warriner 2003; Bignami, Reiners
and Weinreb 2003; Watkins et al. 2003; Anglewicz et al. 2006).
Since the project’s initial conception, the MDICP has expanded in several
ways: In 2004 (1) a sample of adolescents (age 15-24, married and unmarried) was
added to the base sample of ever-married women and their husbands; (2) Biomarkers
for HIV and other sexually transmitted infections were collected from all respondents
who consented, and the results were provided to those who requested them; (3) GPS
coordinates were collected for all sampled households; (4) An expanded religion
module was added to the survey; (5) Wave four (2006) also collected data on the role
of social interactions in altering the consequences of the AIDS epidemic. The
individual level analyses in this dissertation (chapter 6) rely on MDICP Wave III data
(2004), since Wave IV data (2006) is not yet available.
45 Analyses of the MDICP data have provided analyses of numerous HIV-related
issues in Malawi, such as risk perception (Behrman, Kohler, and Watkins 2003;
Helleringer and Kohler 2005), related behaviors (Kohler 2000), infection rates and
mortality (Doctor and Weinreb 2003) and patient care (Chimwaza and Watkins
2004). While the sample has changed in several ways through attrition across the
waves of this study, analyses demonstrate that these changes in the composition of the
sample have not changed many of the observed relationships between series of
important predictor variables and the outcomes at the heart of the project, and the
variables of interest included in this dissertation (Anglewicz et al. 2006; Watkins et
al. 2006; Bignami et al. 2003).
[Insert Table 3.1 about here]
Religion in the MDICP
Previous work using these data has also shown high levels of religious
participation among MDICP respondents (Trinitapoli and Regnerus 2006; Watkins
and Chimbwete 2004). Table 3.1 presents a brief summary of religious affiliation and
participation patterns from MDICP respondents in 2004. Data collected in Wave II of
MDICP (2001) facilitated the important task of studying the relationship between
religion and HIV risk at the individual level, and found that risk behaviors, perceived
individual risk and STD/HIV related attitudes vary by both religious affiliation and
participation (Trinitapoli and Regnerus 2006).
During Wave III (2004) of MDICP data collection, researchers conducted two
pilot studies to expand these initial efforts to studies conducted at the organizational
leveling addition to individual-level studies of the influence of religious organizations
46 among MDICP respondents. Researchers piloted a survey of 60 religious leaders in
the Mchinji sample area, and collected ethnographic doctrinal summaries known as
‘sermon reports’ of 116 religious services – 68 in the Balaka sample area, and 48 in
Rumphi (Trinitapoli 2006b).
The Malawi Religion Project (MRP)
The MRP was subsequently planned as a large-scale cross sectional, mixed-
methods data collection project. The principal aim of the MRP was to collect data on
religious organizations in order to examine how these organizations and their “moral
communities” influence responses to the epidemic in a sub-Saharan African country
with a major HIV/AIDS epidemic. The data collection during summer 2005 included
four primary target populations: leaders of local congregations, local congregation
members, national level denomination leaders and leaders of non-Governmental
organizations (NGOs) active in the three sample areas.
The sample for this study is based on a strategy known as hypernetwork
sampling (Chaves et al. 1999; McPherson 1982; Spaeth et al. 1996), which holds that
a random sample of organizations can be derived by finding organizational
affiliations among a random sample of individuals, then sampling from the named
organizations. As such, the MRP sample consists of all congregations named by wave
III MDICP (2004) respondents. To locate the leaders of these congregations, the MRP
employed interview scouts – locals who have intimate knowledge of the research
area, who have completed the equivalent of high school education, and who have
previously worked for the MDICP.
Congregation Level Data
47 [Insert Figure 3.1 about here]
Defining the sample of congregations was a rather complex process, as
congregations in rural Malawi are frequently hard to identify. Virtually none have a
sign bearing the congregation’s name, and many do not meet in their own building at
all. It is common, for example, for congregations share a building with other
congregations or to not have a building at all (e.g., in one of the sites 3 of the
congregations met under a tree). Often times a single congregation is known by
several different names (including, but not limited to, the name of the village, the
name of the current leader, or the name of the founding leader or mission). As such
the research team refined the congregation list in a multi-stage approach (Figure 3.1).
Of the 3386 respondents in MDICP III, who were asked to name the religious
congregations in which they regularly participate, 3243 provided valid data on this
question. To create the sample of congregations, the research team identified all
different spellings and similar names within this initial list of congregations,2
reducing the list to 251 potential unique congregations. This list was then discussed in
the field daily by the research team, interview supervisors and interview scouts to
further clarify additional multiple namings or difficult to identify congregations.
The resulting congregational leader data comes from surveys and in-depth
interviews with the leaders of 187 congregations, which represents a response rate of
93.5%.3 The interview and survey were conducted in a single setting, with the bulk
of the time devoted to the unstructured interviews. These survey was intended to
focus primarily on six components: 1) Networks, 2) Organizational Structures, 3)
Informal Structures, 4) Social Services, 5) Doctrinal Issues, and 6) Denominational
48 Mobility, each with a specific focus on the relationship to HIV/AIDS. The semi-
structured interviews consisted of four thematic sections: the leader’s personal
religious history, a description of the congregational leadership and its history,
recounting any major problems faced by the congregations, and (if it had not already
been addressed in the other three qualitative sections or the survey) HIV/AIDS related
topics specifically.4
Additional data from congregation members was gathered through semi-
structured interviews from a stratified random sample of previous female MDICP
respondents (N=110). These interviews focused on five primary themes: personal
religious history, congregational norms and discipline, HIV/AIDS, religion and
HIV/AIDS, and family planning, with an added interest in the respondent's outlook
for the future in light of these topics.
Congregational Leaders’ Network Data
One of the primary goals of the MDICP is to examine diffusion of beliefs,
behaviors and knowledge of HIV/AIDS transmission and symptomology; being able
to examine the potential and actual routes of diffusion is an essential aspect of this
task. One of the MRP’s primary aims is to explore the potential effects of a
congregation’s relationships with other congregations, larger denominational
organizations and the few other formal organizations that do exist in these rural
settings. Previous research would lead us to expect significant relationships between
these larger networks and congregational responses to the present HIV epidemic, but
the specifics of the relationships are difficult to hypothesize. Existing evidence is
largely anecdotal (Parry 2003), limited to studies of single communities (Garner
49 2000; Gregson et al. 1995), or based on very small samples (e.g., Lagarde et al. 2000;
Takyi 2003). On one hand, largely poor congregations may gain capital through
support networks that provide opportunities for education and intervention that would
not otherwise be possible. Alternately these networks may embed a congregation
within a strong hierarchy with particular doctrinal systems that discourage protective
behaviors such as condom use. A further possibility is that the remote rural setting
imposes geographical constraints on these networks, limiting the influence of external
contacts (whether from denomination, NGO, or other sources). The reality is that the
existing literature and available data have been simply unable to explore the precise
nature of these organizational relationships; the design of the MRP aims to fill that
void.
Both the coded qualitative interview data and the survey responses to network
questions collected from religious leaders in MRP interviews focus on network
contact information for several relationship types. The initial aim of the network
component was to gather information about the connections between the sample
congregations and (a) other congregations (whether within or outside their own
denomination), (b) their denominations, and (c) other community organizations (e.g.,
NGOs). The network component of the survey addressed each of these potential
relationships through an open-ended listing of contacts for the religious leader’s
personal friendship networks. This section focused on gathering information on the
various people with whom the religious leader may develop their attitudes on any
number of topics, including but not limited to issues of sexual behavior and HIV. The
same open-ended format was also used to gauge congregational co-participation by
50 gathering information about formal cooperation that exists between organizations,
such as joint services, revivals or choir festivals. While these sorts of relationships
may not directly show potential avenues of information flow, they can provide
additional evidence for a congregation's level of embededness or seclusion within
their community. We asked congregation leaders about the individuals with whom
they discuss doctrinal related issues in order to get a picture of the sources they draw
upon for developing sermons and building their doctrinal repertoires in a more
general sense. Also key for this portion of the network data are accounts of
congregation leaders' educational histories and participation in training seminars.
Finally, we asked directly about the contacts with whom they discuss topics relating
to HIV/AIDS. Additionally, several questions examined the frequency of some of
these contacts in terms of formal congregational practices.
The first three waves of MDICP data include many indicators of respondents'
beliefs about HIV, their own risk behaviors, and their levels of perceived and actual
risk. Individual level analyses indicate the importance of religious affiliation and
involvement for many of these factors (Trinitapoli and Regnerus 2005), and an
overview of the messages available in the religious organizations in two of the sample
regions addresses the content of religious organizations' HIV-related messages
(Trinitapoli 2006b).To date, the means by which such organization's behavior
prescriptions are constructed has not been systematically examined. By combining
these congregational network data with the existing individual level data and the
national level denomination leader data, we can examine the match (or potential mis-
match) between the positions at each of these embedded levels. These unique multi-
51 level data will allow researchers to follow diffusion processes, specifying the
directions of the flow of information and other organizational influences.
Survey Data
A primary component of the MRP survey includes a four-section module
designed to collect organizational network information (adams and Trinitapoli 2007).5
First interviewers asked congregation leaders about the frequency of their contact
with leaders of their denomination, other congregations, other denominations, NGOs
and government officials. Next they asked the respondents to provide information
about any other organizations with which their congregation co-participated in
services or programs. Finally, respondents were asked to name those individuals with
whom they addressed "issues of religious belief or church doctrine" and issues related
to HIV/AIDS and list their organizational affiliations. These sections each included a
series of probes asking respondents to specify these connections separately for leaders
of other congregations, denominations, NGOs, and other community leaders.
All MRP interviews were conducted in the local language by trained
interviewers who were hired in each of the sample locations. The national language of
Malawi is Chichewa, however many of the interviews in Balaka were conducted in
Yao, and in Rumphi all interviews were conducted in the local language of Tumbuka.
Interviewer training was conducted in each site; the training emphasized interviewing
techniques to prepare the interviewers to conduct free-flowing, unstructured
conversations covering the four primary themes of the project (listed above).
Network Data from Qualitative Interviews with Leaders
52 The network data from the qualitative interviews is, therefore, obtained by
reading each interview transcript and coding all described relationships that arise
throughout the course of the interview. The entire transcript is used in the coding of
the network data derived from the qualitative interviews, however, the congregational
structure sections of the interview are the main source of relevant information.
One primary component of the unstructured qualitative interviews asks the
respondent to provide an account of their personal religious history. Within this
section of the interview, respondents frequently describe their educational
experiences and give accounts of various training seminars they have attended. In
some of these interviews respondents mention other individuals with whom they were
in school and whether they are still in contact with those individuals today. In some
instances those relationships describe significant contributions to the interpersonal
contexts within which the respondent constructs their doctrine or HIV-realted beliefs
and attitudes and would therefore be coded as ties accordingly. Similarly, their
description of participation in training seminars (whether religiously themed or HIV
specific) may demonstrate particularly salient relationships that presently contribute
to the space within which they develop the positions reflected to their congregations,
and were therefore coded along with other similar relationships as relevant. It is
important to note that because interviewers were trained to conduct the interviews as
naturally as possible, the interviewer could skip these informal network sections if
they felt that the respondent had sufficiently answered the question at another point in
the interview (e.g., in the formal survey).
53 The second and third thematic sections of the unstructured interviews focus on
congregational structure and prevailing problems faced by these conversations.
Similar to the methods discussed above within personal religious histories, these
frequently described interpersonal or inter-organizational network relationships which
were coded accordingly. As an example, in respondents' descriptions of present
congregational problems they frequently also addressed means of conflict resolution.
In some instances these resolutions point to the involvement of individuals from
outside the congregation that contribute doctrine-related content, and were therefore,
as appropriate, coded among the respondent's doctrinal network contacts. One of the
primary goals of this coding of qualitative transcripts for network information was to
allow the respondent's salient relationships to drive network nominations more than
the research team's preconceived notions of relevant ties. This necessitated an
iterative coding process that adapted to each additional reading of each transcript.
Daily debriefing meetings between the interviewers, interview supervisors,
and project coordinators were crucial to the success of this project. The immediate
and thorough evaluation of each transcript as soon as it was completed allowed the
interviewers to conduct callback interviews as necessary when portions of the
interview were either not covered, or were lacking in sufficient detail.
Organizational Data from Individual Congregation Members
The interviews with congregation members focused on their own religious
histories and corresponding attitudes. Occasionally, however, these interviews also
provided information about organizational level relationships. A particularly relevant
example comes from an individual member who mentions events they have
54 participated in with their congregation that also involved other congregations (e.g.,
choir festivals) when describing the extent of their participation in religious activities.
An unplanned but nonetheless valuable source of network data, the individual level
MRP interviews were also coded for organizational network data in the same manner
as that described above for the unstructured portions of the congregation leader
interviews. The focus of this coding remained at the level of organizational ties -- not
that of individual networks.6 While the organization level ties reported in these
interviews were less frequent than those from the congregation leader interview, they
reveal some additional ties that are relevant to the investigation of information flows
within religious contexts. They also provide additional evidence to support the
network ties identified through other sources.
Denomination Level Data
The final component of MRP data that I use in this dissertation is a series of
interviews with national denominational leaders that represent the congregations in
the samples at the individual (MDICP) and congregation leader (MRP) levels. During
the summers of 2005-2006, we conducted 48 interviews with 45 different leaders,
representing 44 denominations or organizations.7 For these interviews, researchers
approached the highest ranking official for each denomination represented in the
MRP sample. Where possible, the team interviewed the president (or equivalent; e.g.,
General Secretary, or director) of each denomination, and each of the main inter-
denomination and interfaith organizations represented in Malawi. Several included
denominations (e.g., CCAP, Church of Christ and New Apostles) do not have
national-level coordinating organizations, so we interviewed regional leaders (e.g.,
55 Synod General Secretary; N=7). In other cases where the president was not available
(N=16), we interviewed Vice Presidents or department heads (e.g., HIV/AIDS
coordinator). We successfully recruited at least one national level leader for all but
one of the denominations represented in the MDICP sample.8 These interviews were
conducted in English by a mix of Malawian and American researchers, with each
lasting between 45 minutes and two hours. For a comparison of the national leaders’
sample and the MRP sample, see Table 3.2.
[Insert Table 3.2 about here]
These national leader interviews were composed of five main sections, which
focused on: (1) the history of the religious organization in Malawi and an overview of
its organizational structure; (2) doctrinal related issues, particularly those which are
distinct to the particular denomination, compared to others in Malawi; (3) the
intended and actual collaborations of the denomination with other organizations,
including congregations within the denomination, other denominations, NGOs,
government and other international organizations; (4) a summary of the individual
leader’s personal religious history; and (5) a discussion of some of the primary
problems facing the denomination and how the organization attempts to resolve such
problems , with a particular focus on how issues of HIV/AIDS are addressed.9
The empirical chapters that follow draw primarily on three components of the
data described above: (1) MDICP-III, individual level surveys; (2) MRP congregation
leader surveys and interviews; and (3) denomination leader interviews.
56 NOTES
1 Large portions of this chapter come directly from three papers that are presently under review (adams & Trinitapoli 2007; Anglewicz et al. 2006; Watkins et al. 2006).
2 The majority of the “duplicate” congregation names were simply issues of recording or data entry which produced different spellings. However, for some cases, it was also possible to identify that, for example, “X CCAP” and “Y CCAP” were the same congregation through identification processes, such as matched leader name(s), location, etc.
3 We actually conducted a total of 194 interviews, which included interviews with two leaders from each of seven congregations. In congregation-level analyses, only one of these respondents is included, which is where the N of 187 comes from.
4 The training document used to guide interviewers through training for the qualitative component of the congregation leader interviews, is included as Appendix A.
5 The full network section of the MRP survey is included as Appendix B.
6 Adding this to the MRP would have been redundant, since the MDICP survey contains thorough individual-level network data.
7 Five individuals were interviewed both in 2005 and 2006, and representatives for 6 organizations were interviewed in both years. One individual was interviewed in each year, representing a different organization in each year. With the exception of this individual, all of the other “repeat” interviews were with representatives of interfaith and interdenominational organizations.
8 The lone refusal in our attempts to obtain interviews came from representatives of the Jehovah’s Witnesses in Malawi. This is not especially surprising given the history of Jehovah’s Witnesses in Malawi. Under the presidency of Kamuzu Banda they were formally banned from Malawi, and only recently have regained legal recognition (Jubber 1977). It should also be noted that the only Church of Christ interview conducted was with a national-youth coordinator. One of the foundational beliefs of the Church of Christ is in the authority of the local congregation, which results in few “coordinating” or oversite organizations within the Church of Christ (Jenkins 2002).
9 The training guide used to prep interviewers for the national leader interviews, is included as Appendix C.
57 New RC MP Pent AIC Muslim MP None Total N Individual Characteristics Respondents N 449 508 217 450 630 246 24 2524 % 17.79 20.13 8.60 17.83 24.96 9.75 0.95 100
Gender Ratio a 0.56 0.65 0.66 0.57 0.59 0.52 0.25 0.59 2524
Participation > Weekly 14.00 15.50 19.62 12.20 16.52c 10.34 0 14.16 1730d Weekly 54.30 42.86 47.37 47.32 13.26e 54.31 0 48.61 2-3 x a month 21.87 33.76 23.44 31.74 6.09f 27.59 0 28.38 Monthly 4.42 4.46 5.74 4.15 11.74g 3.45 0 4.39 < Monthly 5.41 3.61 3.83 4.63 19.78h 4.31 100 4.45 32.61i
Attended last week 63.37 62.30 76.64 62.84 54.31 68.98 0 62.04 2495
Born again / Made 19.82 30.22 40.93 20.68 29.31 29.88 5.56 27.17 2477 Tauba
Congregational Characteristics Congregations N 21 40 32 38 22 34 187 % 11.23 21.39 17.11 20.32 11.76 18.18
Regular adult 37.65 34.44 21.61 31.55 76.36 36.32 37.39 184 attendees j (44.15) (52.56) (16.35) (31.71) (101.48) (44.07) (52.81) Last week 124.14 72.49 35.03 80.09 81.6 54.18 71.11 174 attendance j (164.27) (72.58) (25.53) (114.81) (67.94) (53.13) (92.03)
Talk about AIDS 66.67 85 50 78.95 72.73 73.53 72.19 187 weekly / almost weekly
a Calculated as the proportion of MDICP respondents who are female. b Muslims were asked a different question – “How frequently did you do daily prayers during the last week?” (N=460). Responses are coded descending as: c - 5x per day, e a few times a day, f - once a day, g - few times a week, h – once, i – never. d Only sums respondents from Christian denominations. j Means and (standard deviations). All other numbers presented in the table are column percents, except where noted.
Table 3.1 Summary of Religious Affiliation and Participation MDICP-3
58
Congregation Leaders National Leaders Balaka Mchinji Rumphi Total Presidenta Other Total Roman Catholic 7 10 5 22 2 Quadiriya Muslim 13 1 2 16 2 Sukutu Muslim 2 0 0 2 1 CCAP 6 8 7 21 5 Baptist 9 2 3 14 4 Anglican 2 2 2 6 1 Pentecostal 11 6 12 29 6 SDA 1 0 3 3 3 Jehovah's Witnesses 2 1 1 4 0 Church of Christ 5 9 12 26 1 New Apostolic 0 0 6 6 1 Indegenous 2 19 17 38 0 Christian Other Christian 1 0 0 1 6 Other Muslim 4 0 0 4 6 Other 1
Inter-faith / Inter- 6 3 9 denominational Organizations Total 65 58 71 194 25 23 48
a To maintain anonymity of respondents, I provide only summarized information on the number of presidents, national and regional leaders interviewed.
Table 3.2 Comparison of Coverage – National and Congregation Leader Interviews
59
MDICP Respondents
N = 3386 Pre-fieldwork
Named Congregations No Congregation Reported Sample Preparation N=3243, 95.8% N=143, 4.2%
Unique Congregation Names Duplicate Names N=251, 7.7% N=2992, 92.3%
62
Unique Congregations Duplicate Congregations Fieldwork N=200, 79.7% N=51, 20.3% Sample Cleaning
Interview Complete Incomplete N=13, 6.5% (Response Rate) N=187, 93.5%
Not Found Refusal Disbanded Balaka Mchinji Rumphi N=11, 84.6% N=0 N=2, 15.4% N=65 N=54 N=68
Figure 3.1 Malawi Religion Project Sample
60
CHAPTER 4
EXAMINING POLICY LOCATION (NOT LOCUTION):
CONGREGATION LEADERS’ HIV-RELATED DISCUSSION NETWORKS
While sponsored programs frequently funnel resources and prevention
strategies into umbrella organizations – with the assumption that those then filter to
the local level – we presently know little about how well those intentions match the
reality. In this chapter I refocus the question from what organizations say to where
they construct (and speak) those messages. Following this shift, I develop an
empirical picture of religious organizational structure in Malawi, rather than
following prevention models’ assumptions. I present descriptive analyses of the
networks within which local religious leaders develop their HIV-related messages. I
find that congregational leaders’ networks are largely locally clustered, with little
distance between the space where these messages are constructed and where they are
disseminated. This chapter provides initial evidence, which is elaborated further in
the next, suggesting that religious organizational structure in SSA follows a more
congregational organizational form, consistent with patterns observed in recent
61 literature elsewhere.
Introduction
With the development of the U.S. President’s Emergency Plan for AIDS
Relief (PEPFAR 2007), the U.S. Department of State built “the largest commitment
ever by any nation for an international health initiative dedicated to a single disease—
a five-year, $15 billion, multifaceted approach to combating the disease around the
world” (PEPFAR 2007). This comes in addition to voluminous other resources
already devoted to the problem (USAID 2003). There has been considerable public
debate about the approach of this plan; particularly it’s reliance upon faith-based
organizations as a substantial arm in the distribution of funds and the implementation
of prevention and care strategies. A recent report from the World Health Organization
(ARHAP 2006) calls for increased cooperation with faith-based organizations (FBOs)
in the ongoing fight against AIDS (PEPFAR 2007). Others have been strongly critical
of such a response (Caldwell et al. 1999). In many of the areas where AIDS
prevention and care are the most pressing, rural sub-Saharan Africa in particular,
FBOs are frequently the only existing organizations (Green 2003). Combining this
interest in culturally appropriate strategies with the prevalence of religious
organizations frequently leads PEPFAR, as well as numerous other international
agencies engaged in developing HIV-related prevention and intervention, to develop
strategies that incorporate formal partnerships with various FBOs.
[Table 4.1 about here]
62 In fact, in many countries these governmental-FBO partnerships are
formalized to establish FBOs as the primary umbrella organizations through which
most finances and strategies for HIV intervention efforts flow. Though Malawi is not
one of the 15 focus countries of the PEPFAR initiative, the structure of prevention
efforts there does reflect the umbrella structure common to these efforts. Since 2003
in Malawi five umbrella organizations have coordinated the strategizing of local HIV-
intervention efforts. Each of the 27 districts in Malawi is assigned to one of these
umbrella organizations, and they are primarily responsible for the local efforts
targeting HIV-prevention and care strategies (NAC 2006). Table 4.1 displays how the
districts are filtered into these umbrella organizations in Malawi. Two of these
organizations are explicitly religious in nature and are responsible for 14 of the 27
districts, which are home to over 60 percent of the Malawian population.
Religious Organizational Structure
When researchers and policy makers address existing and potential links
between religion and HIV interventions, they frequently implicitly build upon a
hierarchical conceptualization of religious organizations. Sociologists with peripheral
interest in religion are commonly among those making hierarchical assumptions,
frequently building on notions dating to the early foundations of the discipline
(Weber 1978). This approach has been deemphasized by scholars who focus
explicitly on religious organizations, demonstrating that the empirical reality
frequently does not match this perspective (Chaves 1993).
Recent research in U.S. religion has substantially moved away from such a
63 hierocratic model of religious organization, increasingly noting that congregations are
the fundamental unit of analysis, at least for American Protestant congregations, as
members “belong” to congregations and not to denominations as a whole (Chaves
1993; Warner 1993). Chaves, in moving away from a hierocratic conceptualization of
religious denominations suggests denominations as dual structures – religious
authority structures and religious agency structures. I will discuss the potential
application of this distinction to the Malawian case below. Despite this focus on
congregations in the U.S. sociology of religion (Ammerman 1997; Chaves 2004), a
corresponding shift has not taken place in policy structure and scholarship that
focuses on SSA. In particular, researchers interested in the link between religion and
HIV devote their attention to differences between religious denominations
(Agadjanian 2005; Agha, Hutchinson, and Kusanthan 2006; Takyi 2003), or
discussions of religion in general terms (Gray 2004; Green 2003; Parry 2003).
However, whether this is the way religious organizations are structured in SSA, and
how readily this structure contributes to the funneling of HIV-intervention efforts,
should be an empirical question to verify and not something that is presumed. Recent
evidence demonstrates the targeted responses religious leaders have adopted in
response to the epidemic, particularly highlighting dialogue about the problem,
improving existing prevention efforts, and participation in care initiatives for PLWA
(Trinitapoli 2006).
In this chapter, I evoke a fundamental shift in how researchers examine
organizational policies; in effect reframing what has traditionally been a question of
describing what organizations’ HIV-related policies are, to one of where policy is. I
64 draw on the structural framework that underpins much of SNA to describe the
contexts within which congregational leaders develop their HIV-prevention
strategies. I describe the contexts within which local religious leaders develop their
HIV-related messages as ego-networks, and I present several of the characteristics of
these networks. I especially highlight how those networks differ from the implied
hierarchical organizational networks that underpin the assumptions in many of the
existing prevention models. In this chapter I pay attention to the close social-
proximity of these message-construction networks to the networks within which these
messages are disseminated.
Background
Current research frequently addresses organizational policies based in a
fundamentally essentialist framework treating policies as positions which
organizations or individual leaders hold. This takes place both explicitly – when
research focuses on the formal policies and organizational forms implemented to
meet those policies; and implicitly – when outcomes are modeled to differ by
organizational memberships, inferring internal consistency within a given
organization. Even when researchers compare informal to formal organizational
structures and policies, researchers still frequently defer to conceptualizing policies
that necessarily “are.”
This is particularly evident in cases of religious organizations, where
discussions wrapped in theological verbiage are frequently interpreted as, if not
intended to be, concrete absolutes describing the way things are within religious
organizations. For example, this is manifest in literature concerning the potential link
65 between religious organizations and HIV - in the debate about whether religious
organizations are helpful allies or harmful foes in the ongoing fight against the HIV
pandemic presently facing SSA. These literatures focus on how religious
organizational policies either promote or inhibit behavior change or other outcomes
relevant to halting and reversing the present epidemic trends (Caldwell 1999; Green
2003; Parry 2003}.
Recent work (e.g., Trinitapoli 2006b) demonstrates some of the considerable
variation that exists within religious organizations, particularly with reference to HIV-
related teachings and strategies. These variations suggest that most work built on
conceptualizations of religion in general terms and based on singular frames, or even
studies that describe denominational differences (Caldwell 1999; Green 2003; Parry
2003) – most treating them as internally consistent - potentially masks many of the
relevant ways in which religious organizations are actually engaging the epidemic.
While these criticisms may best apply to external evaluations of religious
organizational messages, similar problems can arise from representative interviews of
leaders within such organizations. Previous research presents ample evidence for
individuals’ proclivities for placing themselves closer to the center of relational
spaces than objective comparisons support (Kumbasar, Romney, and Batchelder
1994).1
Interventions in health outcomes are fundamentally local processes, but most
of the existing literature does not address local policies. This approach, which focuses
on formal policies, leaves a potentially large gap between the policies examined and
the location where they are implemented or their intended individual level outcomes.
66 While this is a common approach to investigating responses to the HIV epidemic
(USAID 2003), it has been criticized as a poor predictor of policy effectiveness
(Gauri 2006). In this paper, I therefore turn the attention to intraorganizational
networks to examine the local implementation of interventions, focusing separately
on their construction and implementation.
Influence as a Structural Property
While the existing literature examining the link between religion and HIV-
efforts devotes little attention to religious organizational structure, the implied
hierarchy of prevention models is not the only place to look to build expectations
about the relational structures in these organizations. In particular, literature that
builds on the dyadic nature of influence provides a framework that is useful for
guiding the analyses below. In principle the approach draws on the importance of
reducing role conflicts, a central theme of Robert K. Merton’s work (1968).
Robert Merton’s attention to role conflicts leads directly to the perspective of
Noah Friedkin, which provides two key foci for the analyses here. Stated formally,
Friedkin suggests that “Actor j’s influence on i depends on i's knowledge of j’s
opinions…[and] j’s influence on i depends on the salience or value of j’s opinion for
i” (Friedkin 1998: 68).2 These two factors, which are pertinent for the present
analyses, draw directly on Merton’s characteristics that serve to reduce role conflict –
(a) importance of those influencing the actor and (b) insulation from observability,
respectively.
The hierocratic conceptualization of religious organizations stated in terms
67 consistent with Friedkin’s perspective would require that denomination leaders are
both seen and salient for local religious leaders. Evidence suggesting a lack of
influence from national to local religious leaders, within Freidkin’s framework would
suggest that one (or both) of these are lacking in this particular context. First, in this
chapter the network properties I describe serve as measures representing
congregational leaders’ awareness or knowledge (or lack thereof) of national
denominational leaders HIV-related messages.
Some additional familiar ideas from SNA literature would also suggest a
move away from focusing exclusively on denomination leaders. One positional
estimate of intraorganizational influence is centrality.3 In the religious organizational
context centrality estimates only advantage denominational leaders in situations of
pure hierarchy, assuming relationships between denominational authorities and
congregational leaders even exist. Figure 4.1 presents a hypothetical religious
organizational structure, for one potential denomination. The central (red) node is a
denominational authority, and the five yellow nodes represent congregational leaders,
with the blue nodes representing congregational members, with the number of
members approximating the average size of MRP congregations.
[Insert Figure 4.1 about here]
This figure illustrates some of the properties of networks that are well-known
in network literature. While the denominational leader appears at the middle of all
four graphs in layout, and appears to suggest they are also relationally central, the
addition of any shortcuts in the graph quickly generates reductions in the red node’s
centrality. In Panel A, the case of pure hierarchy, the denominational leader is the
68 most central, using both closeness and betweenness centrality measures. In Panel B,
however, where each congregational leader is connected to one other, the red node
has the highest closeness centrality, but the yellow nodes are higher on betweenness
estimates. Previous literature establishes that while higher closeness centrality
improves dissemination efficiency, higher betweenness centrality produces greater
likelihood of effecting the transmission of information (Freeman 1979; Friedkin
1991). Adding additional ties between the congregation leaders, or removing any of
the local-national links only further exacerbates the congregational leaders’ positional
advantages (Panel C & D).
Adding additional denominations and congregations would change the scope,
but not the nature of the changes illustrated in Figure 4.1. While the network
properties I present later in this chapter are ego-based and cannot calculate centrality
estimates, any substantial lack of congregational-denominational relationships would
not only distance individual leaders from their own denominational authorities, but
also make the overall picture more likely to fall somewhere between Panels B and D,
or even the high congregational-leader connectivity of Panel C. Either way it would
not resemble Panel A, which is the only one suggesting a positional advantage for
denominational leaders.
For local congregation leaders, the social distance between denomination
leaders and congregation leaders, particularly in rural Malawi, is great. This distance
may be such that in the determination of what local congregation leaders contribute to
HIV-prevention efforts, they may not identify national leaders, even from their same
denominations, as among their salient referents. As such in this chapter I investigate
69 the composition of the networks within which local congregation leaders develop
their HIV-related prevention strategies. This suggests that proper referents, drawn not
from the formal organizations within which particular leaders are embedded, but from
their actual day-to-day interactions, would provide higher levels of agreement than
other within-denomination comparisons.
Data and Methods
For the analyses in this chapter, I draw primarily from the social networks
component of the religious leaders' survey from the Malawi Religion Project (MRP).4
The MRP includes interviews with 194 congregational leaders in three rural districts
in Malawi - Rumphi in the northern region, Mchinji in the central region, and Balaka
in the south.5 These interviews combined open-ended discussion topics and a formal
survey component. For a more detailed description of the MRP data, see Chapter 2,
adams and Triniapoli (2007) and Triniapoli (2005).
The bulk of the analyses in this chapter provide descriptive properties of (1)
religious organizational networks and (2) doctrinal and (3) HIV-related discussion
ego-networks of religious leaders. I describe the properties of these networks to (a)
examine networks within which religious leaders discuss topics that potentially
directly contribute to the differences in content within those organizations, and (b)
demonstrate the general relational landscape within which these individuals are
embedded.
In these descriptive analyses, I draw on most of the gathered relational
information from religious leaders in the MRP, focusing separately on four sections
of the religious leader's interviews. I first describe the general relational patterns the
70 leader describes between him/herself or their organization and leaders of several other
leaders or organizations. Then I describe the co-participation of religious
organizations in shared service, fellowships or similar gatherings. Finally, I
investigate the discussion partners with whom the religious leader discusses topics
concerning religious doctrine and issues related to HIV/AIDS.
Additionally, I supplement these descriptive analyses with three other analytic
approaches. First, I include excerpts from qualitative interview transcripts that help
elaborate the patterns demonstrated in the quantitative responses. Second, I
demonstrate some of the substantial variation in these patterns that exist across
varying denominational affiliations. Third, I discuss, though do not present in detail,
several regression analyses predicting several of the described network relations, to
examine whether any of the differences by religious tradition - included in the
previous sections - are potentially better explained by factors other than religious
tradition.
Organization of the Discussion
Scholarship on religious organizations frequently uses broad classificatory
groupings of congregations or denominations based on various taxonomies of
tradition (Smith 1990) or historical organizational development (Steensland 2000). In
MRP interviews, each congregation was classified into one of 12 denominational
categories – Roman Catholic, Quadriya Muslim, Sukuti Muslim, Church of Central
Africa Presbyterian (CCAP), Baptist, Anglican, Pentecostal, Seventh Day Adventist,
Jehovah’s Witness, Indigenous Christian, Indigenous Non-Christian, and Other. All
responses of “Other” were then specified and later coded into the appropriate
71 category. An additional grouping – Church of Christ – was generated after data-
collection based on the number of “Other” responses who indicated this particular
affiliation. For use in analyses elsewhere (e.g., Trinitapoli 2007), these denominations
are further reduced using a strategy similar in motivation to the historical basis that
underpins schema in Steensland et al. (2000). In the analyses here, I use these six
categories - Catholic, Muslim, Pentecostal, AIC, traditional mission Protestant (i.e.,
Presbyterian, Anglican, Baptist), and new Mission Protestant (Seventh Day
Adventist, Church of Christ, Jehovah’s Witness) – to organize the discussion.6 I do,
however, highlight a few of the denominational differences that are particularly
strong contributors to the overall patterns described.
Findings
In the sections that follow, while I highlight many of the specific frequencies
and differences across religious leaders, I hope to draw your attention, not to the
particulars of each variable or the specific differences observed, but to the general
patterns across Tables 4.2-4.5.7 Regarding the importance, and openness of existing
relationships among religious leaders in his community, one Baptist pastor explains:
Our relationship [with other pastors in this area] is so fine and perfect. We don’t have problems. We meet every month to discuss some issues of our churches. And we encourage each other on areas that we see our friends need encouraging.
Organizational Interconnections
In Table 4.2, I present by - religious tradition - the frequency that religious
leaders report a series of connections between themselves and the leaders of various
other organizations (Panel A) and their congregation and other organizations (Panel B
72 and C). Virtually all religious leaders are (at least minimally) linked into their
denomination, with over 95% of leaders having met with congregational and
denominational leaders from their own denomination within the past year (numbers
not shown, though in essence sum rows 1 and 2 of Table 4.2). They are also
remarkably similarly tied into the religious leaders of other denominations, with
roughly three-fourths of all leaders reporting similar meetings with congregation and
denominational leaders of other denominations.
[Insert Table 4.2 about here]
The differences observed between meetings with these leaders and
missionaries, NGO representatives and governmental officials begin to demonstrate
the local-"outsider" discrepancy that will be further elaborated in what follows in later
tables. While denominational officials (whether from the leader's own or another
denomination) potentially could represent leaders outside the local context,
qualitative data, and the specified leaders nominated as conversational partners below
suggest that these are likely regional leaders, or denominational authorities who
happen to be locally stationed – not denominational authorities from Lilongwe
(Malawi’s capital), or beyond – venturing to the remote villages that are represented
in the MRP. Barely more than a third of religious leaders report having contact with
leaders who are necessarily located outside of their own communities (i.e.,
73 missionaries, NGO officials or governmental representatives).
Most of the differences observed in Panel A are not particularly surprising –
e.g., congregations in decentralized denominations (such as the Church of Christ)
have lower levels of contact with denominational leaders; Pentecostal congregations
are more likely to have contact with missionaries; and New Mission Protestant groups
report higher levels of missionary support, due in large part to these organizations
having developed a presence in Malawi only relatively recently (within the past 50
years). There are two significant differences in this first panel worth special attention
however. First is that while Muslims are comparatively more isolated on virtually
every other measure presented in Table 4.2, a comparatively higher proportion of
Muslim leaders have met with clergy from other denominations. Mosques, and
Muslims, in our sample are predominantly located in Balaka, where villages are much
larger than the other two research locations. Because of these village-level
differences, Muslim leaders are simply more likely to be in direct contact with leaders
of other religious organizations. If compared only to other religious leaders within
Balaka district, this Muslim advantage disappears. Second, Roman Catholics are
substantially more connected to the leaders of NGOs. This is due in large part to the
combined facts that the Roman Catholic and CCAP8 denominations (co-)sponsor
several of the dominant NGOs in Malawi and that they represent two of the three
74 largest single denominations in Malawi.9
Of particular importance in Table 4.2 is that roughly one fourth of the
congregations report never having been visited by outside denominational authorities
or missionaries (labeled "religiously isolated"). While the congregational leaders are
engaged with religious leaders from their denominational hierarchies (though the
degree of that engagement will be questioned below), this translates relatively
infrequently to their denominations initiating contact with the congregation itself.
Again, on this measure, Muslim leaders report their ROs to be substantially more
isolated than other religious leaders. In describing this disconnect between local
religious organizations and outside influence, one CCAP elder responds to a question
of whether his congregation has ever received support from people outside the
church:
Outside our church? Ah! Ah! [laughs] No! [Never before?] No. Never.
Congregational Co-Participation
Table 4.3 summarizes leaders’ reports of their congregations’ involvement
with other congregations in co-sponsored programs and shared services, fellowships
or revivals. Roughly two-fifths of all religious leaders report that their congregation
has participated in services or programs with other congregations in the previous
year. Muslims and New Mission Protestants are less likely, while Mission Protestants
75 are more likely to report involvement in such shared participation.10 Church of Christ
and Jehovah's Witness congregations account for most of the New Mission Protestant
"disadvantage" on this particular variable, which may be explained, in part, by these
organizations embattled histories within Malawi, which is especially well-
documented for Jehovah's Witnesses (Jubber 1977). As one Church of Christ pastor
explained why his congregation is not involved in such "fellowships":
It is different for us to get together because of differences in our teaching. It would be like we are condemning them…We must follow our way of preaching. If we came together, we would be indulging in fabricating rules without consulting the Bible. It would be like letting our members temporarily remove their beliefs as easily as you would a chain around your neck.
While there are substantial variations also in the numbers of congregations with
whom a congregation is involved in such co-participation, a comparison of the last
line of Panel 1 (showing significant differences when all congregations are included)
and the last line of Panel 2 (showing no significant differences when limited to those
who have at least one such tie) suggests that this is largely a matter of whether
congregations do so at all, and not reflective of the degree of such involvement.11
[Insert Table 4.3 about here]
In the bottom Panel of Table 4.3, I present a comparison of the reporting
congregation’s denomination to the denomination of those congregations with whom
they co-participate. There are few congregations who only co-participate with others
of their same denomination. In the rural context of this study there are no villages that
76 can support multiple congregations from the same denomination. Any co-
participation between two congregations of the same denomination thus necessarily
reflects somewhat intentional efforts by one of the congregations to move beyond
their local context. Therefore, these same-denomination co-participations could
potentially reflect embededness within stronger hierarchical denominations, or better
coordination within a single denomination. However, this possibility is largely
undermined by the fact that the two groups who are substantially more likely to report
this within-group limitation, Pentecostals and New Mission Protestants,12 are the two
religious traditions whose denominations are most likely to lack formal
denominational hierarchy (e.g. Church of Christ, which is explicitly congregationally
organized). This is therefore again probably better explained by the somewhat
embattled positions of these newer denominations within the religious landscape, than
it is of intentional exclusivity.
Conversely, I also calculate a measure representing those congregations who
only co-participate with congregations from denominations different from their own.
Roughly half of all congregations who report any co-participation reflect this pattern.
In contrast to the above mention that same-denomination coparticipation reflects
moving outside of a congregation's local context for such partnering, this measure is
likely a proxy for congregations whose co-participation(s) is(are) focused locally.
Roman Catholic and Mission Protestant congregations (the latter difference is again
77 driven almost entirely by Baptist congregations) are more likely to report locally-
focused co-participation. As one Roman Catholic priest described this practice:
Whenever we are involved in anything – whether a funeral or any activity – if we are really caring about others, we must inform others what is happening with us. And if those others have really strong love they come.
Doctrinal Discussion Partners
While U.S. congregational leaders generally have some form of religious
training, secondary education is rare, with approximately 29 percent of MRP
respondents having attended school past primary school (8 years); post-secondary
education is rarer still, with less than six percent of MRP respondents (N=10) having
attended any college or seminary education. Approximately two-thirds do report
having some additional religious training, but the vast majority of those described are
short-term (e.g., less than a week) bible studies or AIDS training seminars. As such,
religious leaders’ formal and informal relationships likely contribute substantially to
the form that their religious organizations take, and the content discussed in its
services. Table 4.4 presents whether MRP respondents discuss doctrinal issues with
leaders of a series of organizational types (Panel A), and with how many of each type
of leader they have such discussions (Panel B).
[Insert Table 4.4 about here]
While the vast majority (86%) of religious leaders consult some other
individuals about doctrinal related issues. Two-thirds of MRP respondents consult
other religious leaders about such topics (not shown, combines rows 1 and 2).
78 Among the various religious traditions, only Muslims are substantially different on
this measure with only about half reporting discussing doctrine with anyone, and 43
percent consulting other religious leaders. Slightly less than half of these leaders
report covering such topics with leaders from their denomination, while nearly two-
thirds rely on other congregational leaders. When investigating how many of each
type of conversational partners religious leaders report, other congregational leaders
and local villagers (i.e., not leaders of any religious organization) account for the
largest number of such partners, 2.3 and 2.6 respectively among those who have at
least one such partner. MRP respondents are substantially less likely to have doctrinal
related conversations with denominational leaders (p<0.01), and when they do so,
have them with substantially fewer such leaders (p<0.001). 13 Similar discussions
with NGO leaders are even less common still.
Panel C of Table 4.4 presents data for whether, and how many other religious
leaders MRP respondents report having regular interactions apart from doctrinal
discussions or formal (e.g., planning co-participation) relationships. This provides
additional evidence for the importance of local religious leaders as support to one
another, above and beyond their potential contributions to explicitly religious topics.
AIDS Discussion Partners
Finally, Table 4.5 presents the frequency and number of conversation partners
for conversations particularly pertaining to HIV/AIDS. The vast majority (72%) of
MRP congregational leaders are also engaging leaders of other organizations about
HIV-related topics (60% if limited only to other religious leaders). Again, Muslim
leaders are substantially less likely to engage in such discussions, across all alter-
79 types, and have fewer such partners overall, though in this case, the numeric
difference in this case is a difference in kind, not degree.
[Insert Table 4.5 about here]
As is the case with doctrinal discussion partners, for discussing topics related
to HIV/AIDS MRP congregational leaders are more likely to turn to other
congregational leaders than they are to denominational leaders (p<0.05), and when
they do so, incorporate more congregational leaders in such discussions than they do
denominational leaders (p<0.01). Again, NGO leaders are the alters to whom they are
least likely to turn. Unlike the doctrinal discussions described above however, other
local villagers (not leaders of any of these organizations) are the most likely
nominated discussion partners (p<0.05, compared to congregation leaders - the next
highest), and substantially more of them are nominated than any other alter types
included in this table (p<0.01, compared to congregation leaders - the next highest).
Estimating Robustness of Denominational Differences
The differences of primary interest in this chapter are those between ties to
congregational and denominational leaders, however, the various differences between
denominations in their likelihood to report particular tie types, observed in the
previous four tables, are also informative. To examine whether those differences
remain when controlling for other factors, I also calculated a series of logistic
regressions predicting the likelihood of a leader reporting having any of each type of
tie. In these models, while the focus is on religious tradition, I also control for the
district and the congregation leader’s comparative evaluation of the HIV problem in
their congregation compared to the rest of their village. I also estimated these models
80 controlling for the size of the congregation, tenure of the leader, and whether the
leader would identify their congregation as supporting “born again” beliefs.14 Because
of the limited number of cases in these models, I only present those that had a
significant direct effect or change the effects for the religious tradition variables.
[Insert Table 4.6 about here]
Table 4.6 presents the results for denominational conversation partners.
Largely, the same patterns demonstrated in the descriptive tables above remain when
controlling for other factors as well. Additionally, beyond what can be seen in the
table, Muslims report fewer congregational leader ties and “other villager” ties than
Pentecostal or new mission Protestant leaders, and fewer denominational leader ties
than AIC or mission Protestant leaders. I also estimated these models as a series of
OLS regressions predicting the number of such ties. As with the models presented
here, the descriptive patterns presented in Table 4.4 are not substantially changed.
[Insert Table 4.7 about here]
Table 4.7 presents the same information for AIDS-discussion partners. This
table presents a slightly different result than Table 4.6 in that the only significant
difference that remains is for Muslim leaders, who are substantially less likely than
Roman Catholic leaders to discuss HIV-related topics with denominational
authorities. In comparisons not shown here, they are also less likely to report such
denominational ties than all of the other religious categories. They also report fewer
congregation leader ties than new mission Protestants and Pentecostals, and fewer
“other villager” ties than mission protestant leaders. All of the relationships presented
in the descriptive tables above remain, suggesting that other factors do not appear to
81 alter the observed differences, and in fact, very few other factors are themselves
significant predictors of such ties. Those leaders who evaluate their congregation’s
problem to be more like that of their surrounding community are more likely to name
other congregation leaders as important doctrinal discussion partners. Leaders in
Rumphi are substantially less likely to nominate virtually every type of tie, which is
not surprising given the smaller villages in Rumphi. What is surprising is that they are
no less likely to nominate denominational leaders among their important doctrinal
discussion partners, largely reflective of their unimportance across all settings.
Finally, religious leaders from Mchinji have substantially greater likelihood of
interactions with NGO leaders; though this is likely simply a matter of proximity,
given that Mchinji is much closer to Lilongwe – Malawi’s capital, and the
headquarters of most NGOs.
Discussion and Conclusion
Overall, the pattern presented here demonstrates that MRP congregation
leaders are likely to focus their relationships - and those of their religious
organizations - locally. Particularly relevant for this dissertation is that this local
clustering of networks is especially strong when we focus on the relationships that
respondents report contribute to their development of HIV-related discourse. Susan
Watkins (2004) (among others, see also Gerland 2005; Poulin 2006), demonstrates
that rural Malawians’ AIDS-related and family planning beliefs and behaviors are
substantially influenced by discussions that take place alongside the activities of daily
life (e.g., collecting water from the local borehole). Rural Malawian religious leaders
appear to adopt a remarkably similar approach in the relationships that shape the
82 governance of their religious organizations - especially related to how they develop
HIV-targeted information. The quantitative comparisons presented here demonstrate
that these particular relationships are frequently limited to local partners. The
qualitative accounts suggest that these type of conversations are as likely (if not
substantially more so) to take place with friends alongside daily activities - like
walking the weekly trip to the trading center - as they are to occur in training
seminars or through formal relational or organizational channels.
A revealing pattern across these tables is that Roman Catholics are the least
likely to have interactions with other (Table 4.2, Panel 1), or even their own
denominational authorities (Panel 2), the most likely to focus their co-participation
locally (Table 4.3, Panel 2), and report the fewest denominational authority doctrinal
discussion partners (Table 4.4). Catholicism is one of the most hierarchically-
organized of the religious organizations represented in the MRP sample; yet religious
hierarchy appears to have the least impact on these congregations. Religious leaders'
HIV-related messages are largely constructed in the local context. These observations
suggest it would be a mistake to continue to assume that congregations in rural
Malawi are deeply embedded in strong hierarchical religious organizations. In the
next chapter, I elaborate how this relational structure contributes to the HIV-relevant
messages of congregational leaders. I demonstrate that this relational distance
between congregational leaders and their denominational authorities lends itself to
local religious leaders’ tactics for engaging the epidemic being more reflective of
their immediate context, and producing more suitable responses.
83 While the vast majority of MRP congregational leaders report having some
contact with the national (or regional) leaders of their denomination, they are much
more embedded in locally-clustered relationships. Denominational presence is
limited, at best, in the local congregation. Congregation leaders are much more likely
to turn to other people in their community - whether religious leaders of other
traditions, members of their own congregation, or simply other villagers who they
know intimately - than they are to outside leaders. This local preference exists for
specifically religious conversations, but is exacerbated even further for conversations
about HIV/AIDS.
84
A - Pure Hierarchy B - Hierarchy with Local Short-cuts
C - Leader Clique D - Local Clustering
Figure 4.1 Potential Denominational Leadership Relationship Structures
85
Umbrella Organization Districts Region Population (1998)a Action Aid Malawi Chitipa North 126,799 Chiradzulu South 236,050 Mulanje South 428,322 Mwanza South 138,015 Phalombe South 231,990 Thyolo South 458,976 Zomba South 546,661 (Total = 2,166,813 ) Canadian Physicians for Aid and Karonga North 194,572 Relief (CPAR) Likoma North 8,074 Nkhata Bay North 164,761 Rumphi North 128,360 (Total = 495,767 ) Plan International Malawi Kasungu Central 480,659 Mzimba North 610,994 (Total = 1,091,653) Save the Children – USA Balaka South 253,098 Blantyre South 809,397 Lilongwe Central 1,346,360 Mangochi South 610,239 Nkhotakhota Central 229,460 Salima Central 248,214 (Total =3,496,768 ) World Vision International Chikwawa South 356.682 Dedza Central 486,682 Dowa Central 411,387 Machinga South 369,614 Mchinji Central 324,941 Nsanje South 194,924 Ntcheu Central 370,757 Ntchisi Central 167,880 (Total =2,682,867 )
NOTES – The bold-italicized districts in are those in the MDICP & MRP samples. a Source –(NSO 2006).
Table 4.1. HIV-Umbrella Organizations by District – Malawi, 2005
86 RC MP Pent AIC Muslim New MPTotal In the past year have you met with… Clergy from another denomination 71 83 77 72 91 f 65 e 76 Authorities from another denomination 62 b 88 a,f 81 74 68 67 b 75 Missionaries 14 b,c,f 38 a 50 a,d 21 c 36 41 a 34* Representatives from an NGO 50 c,d,e 54 c,d,e 22 a,b 24 a,b 18 a,b 33 34** Government leaders 29 51 38 42 32 39 40
Has your congregation ever been visited by… Denominational authorities 57 b,d,e 83 a,e,f 78 e,f 83 a,e,f 23 a,b,c,d 42 b,c,d 65** Government officials 19 23 23 34 f 13 14 d 22 Missionaries 29 36 53 e 34 18 c 41 37
89 Religiously Isolated g 33 e 15 e 19 e 16 e 68 a,b,c,d,f 32 e 27**
Has your congregation ever been directly helped by… NGO programs 5 18 17 13 14 9 13 Mission Work 24 12 f 27 15 f 9f 51 b,d,e 24** Correlation - Mission visit & help 0.38 0.45* 0.31 0.45* 0.26 0.60** 0.43* Correlation - NGO visit & help 0.24 0.03 0.22 0.15 -0.20 0.00 0.07
N 21 40 32 38 22 34 187 NOTES: Numbers presented are column percentages, except where noted otherwise. * p<0.05, ** p<0.01 (χ2 Distribution, across the specified row) Letters denote significant differences (p<0.05) from: a Roman Catholic, b Mission Protestant, c Pentecostal, d African Independent, e Muslim, f New Mission Protestant g Religiously Isolated congregations are those that have never been visited by any “outside” religious leaders (denominational authorities or missionaries)
Table 4.2 Organizational Interconnections by Religious Tradition
87 RC MP Pent AIC Muslim New MP Total
All Congregations Co-participation in 60 e,f 67 c,d,e,f 41 f 41 b,f 18 a,b 18 a,b,c,d 41** programs or services with other congregations # Other Congregations g 2.7 c,f 2.4 c,f 1.3 a,b 1.9 f 1.6 0.6 a,b,d 1.7* (2.4) (1.9) (1.6) (2.9) (1.7) (1.4) (2.2) N=187
Congregations that Co-Participate with others Alter Congregations’ Denominational Affiliation All same as ‘ego’ 8 c,f 19 c 46 a,b 27 25 50 a 26* Mixed 8 12 31 33 50 0 20 All different from ‘ego’ 83 c,d,e 69 c 23 a,b 40 a 25 a 50 54 # Other Congregations g 3.8 2.7 2.5 3.9 2.8 3.0 3.1 (1.9) (1.8) (1.5) (3.0) (1.3) (1.7) (2.1) N=76 * p<0.05, ** p<0.01 (χ2 Distribution, across the specified row)
NOTES: Numbers presented are column percentages, except where noted otherwise. (may not sum to 100, due to rounding). Letters denote significant differences (p<0.05) from: a Roman Catholic, b Mission Protestant, c Pentecostal, d African Independent, e Muslim, f New Mission Protestant g Numbers presented are means and (standard deviations).
Table 4.3 Congregational Co-Participation with Other Religious Organizations
88 RC MP Pent AIC Muslim New MP Total Any 86 e 82 c,d,e 97 b,e 97 b,e 52 a,b,c,d,f 91 e 86 ** Congregational Leaders 38 c,f 60 75 a,f 63 43 c,f 74 a,e 61 * Denominational Leaders 29 53 e 47 53 e 25 b,d 50 45 NGO Officials 14 25 9 13 10 15 15 Other Villagers 29 c,d,f 50 f 69 a,e 66 a,e 30 c,d,f 76 a,b,e 57 ** Other Congregational Leaders N 0.6 b,c,e 1.8 a 1.5 a 1.2 e 1.0 1.9 a,d 1.4 (0.9) (2.0) (1.3) (1.3) (1.7) (1.6) (1.6) N (>0) g 1.5 b,f 3.0 a,c,d 2.0 b 1.9 b 2.4 2.6 a 2.3 (0.8) (1.7) (1.0) (1.2) (1.7) (1.3) (1.4) Denominational Leaders N 0.4 d,f 1.2 e 1.0 e 1.1 a,e 0.3 b,c,d,f 1.1 a,e 1.0 (0.8) (1.7) (1.4) (1.4) (0.4) (1.4) (1.4) N (>0) 1.5 2.3 2.2 2.1 e 1.0 d,f 2.2 e 2.1 (0.8) (1.8) (1.4) (1.2) (0.0) (1.1) (1.3) Officials from Other Non-Governmental Organizations N 0.2 0.6 0.2 0.3 0.2 0.2 0.3 (0.7) (1.3) (0.6) (0.9) (0.6) (0.5) (0.9) N (>0) 1.7 2.4 1.7 2.2 2.0 1.4 2.0 (1.2) (1.7) (1.2) (1.3) (0.0) (0.5) (1.3) Other Individuals (Local Villagers) N 0.6 c,d,f 1.2 f 1.9 a,e 1.6 a,e 0.8 c,d,f 2.4 a,b,e 1.5 (1.0) (1.6) (1.7) (1.6) (1.3) (2.1) (1.7) N (>0) 2.0 2.5 2.7 2.4 2.5 3.1 2.6 (0.6) (1.4) (1.3) (1.4) (1.0) (1.8) (1.5) Religious Leader Friends Any 76 e 70 e 81 e 84 e 38 a,b,c,d,f 71 e 72 ** N 1.9 e 1.6 e 1.9 e 2.3 e 0.6 a,b,c,d,f 1.7 e 1.7 (1.5) (1.4) (1.5) (1.6) (0.9) (1.4) (1.5) N (>0) 2.5 2.3 2.4 2.7 e 1.6 d 2.4 2.4 (1.2) (1.1) (1.3) (1.4) (0.8) (1.1) (1.2) * p<0.05, ** p<0.01 (χ2 Distribution, across the specified row) NOTES: Numbers in Panel A (and Panel C, row 1) are column percentages (may not sum to 100, due to rounding). Numbers in Panel B (and Panel C, rows 2 and 3) are means and (standard deviations). Letters denote significant differences (p<0.05) from: a Roman Catholic, b Mission Protestant, c Pentecostal, d African Independent, e Muslim, f New Mission Protestant g N (>0) denotes calculations using only those who have at least one such discussion partner.
Table 4.4 Leaders’ Doctrinal Conversation Partners Outside the Congregation
89
RC MP Pent AIC Muslim New MP Total Any 71 70 81 e 69 45 c,f 85 e 72* Congregational Leaders 48 55 e 59 e 56 e 23 b,c,d,f 67 e 53* Denominational Leaders 52 e 55 e 34 50 e 13 a,b,d,f 53 e 44* NGO Officials 33 e 33 e 19 22 5 a,b 22 22 Other Villagers 57 e 65 e 63 e 61 e 27 a,b,c,d,f 78 e 61* Other Congregational Leaders N 1.1 e 1.5 e 1.0 e,f 1.1 e 0.4 a,b,c,d,f 1.7 c,e 1.2 (1.4) (1.6) (1.1) (1.3) (0.9) (1.4) (1.4) N (>0) g 2.4 2.7 c 1.7 b,f 2.0 1.8 2.5 c 2.2 (1.0) (1.3) (0.8) (1.1) (1.1) (1.0) (1.1) Denominational Leaders N 1.0 e 1.1 c,e 0.6 b,d,e 1.2 c,e 0.1 a,b,c,d,f 1.2 e 0.9 (1.2) (1.2) (0.9) (1.4) (0.4) (1.5) (1.2) N (>0) 2.0 e 2.0 1.6 2.3 1.0 a 2.2 2.0 (0.8) (1.0) (0.8) (1.1) (0.0) (1.4) (1.1) Officials from Other Non-Governmental Organizations N 0.7 e 0.5 e 0.3 0.4 0.0 a,b 0.3 0.4 (1.0) (0.8) (0.6) (1.0) (0.2) (0.7) (0.8) N (>0) 2.0 1.6 1.3 2.0 1.0 1.4 1.7 (0.6) (0.7) (0.8) (1.1) (0.0) (0.8) (0.8) Other Individuals (Local Villagers) N 1.6 1.5 2.1 e 1.6 0.8 c,f 2.3 e 1.7 (1.9) (1.4) (2.2) (1.5) (1.5) (1.9) (1.8) N (>0) 2.8 2.3 c 3.3 b 2.5 2.8 2.9 2.7 (1.8) (1.0) (1.8) (1.1) (1.7) (1.6) (1.5) * p<0.05, ** p<0.01 (χ2 Distribution, across the specified row) NOTES: Numbers in Panel A are column percentages (may not sum to 100, due to rounding). Numbers in Panel B are means and (standard deviations). Letters denote significant differences (p<0.05) from: a Roman Catholic, b Mission Protestant, c Pentecostal, d African Independent, e Muslim, f New Mission Protestant g N (>0) denotes calculations using only those who have at least one such discussion partner.
Table 4.5 Leaders’ AIDS Conversation Partners Outside the Congregation
90
Congregational Denominational NGO Other Villagers Muslim a 0.97 0.59 1.46 0.95 (0.69) (0.47) (1.53) (0.72) AIC a 3.45* 3.21 1.01 5.52** (2.10) (1.97) (0.84) (3.46) Pentecostal a 6.72** 2.28 1.02 6.79** (4.47) (1.44) (0.94) (4.46) Mission Protestant a 3.00 3.05 2.7 2.87 (1.78) (1.84) (2.06) (1.73) New MP a 5.70** 2.43 1.26 9.17** (3.68) (1.51) (1.04) (6.08) Balaka a 1.01 1.00 0.26* 0.73 (0.48) (0.43) (0.15) (0.34) Rumphi a 0.38* 0.56 0.32* 0.42* (0.16) (0.23) (0.17) (0.18) AIDS problem worse 1.59 1.98 0.47 1.41 in village b (0.56) (0.69) (0.21) (0.50) AIDS problem same 5.95* 3.23 0.94 2.32 in congregation (5.17) (2.14) (0.81) (1.68) and village b Intercept 0.54 0.30* 0.43 0.44 -0.3 -0.17 -0.3 -0.25
-2 Log-likelihood -107.25 -114.56 -70.42 -108.14 χ2 23.22 16.54 14.09 26.67 Pseudo R2 0.10 0.07 0.09 0.11 N 179 178 178 178 ** p<0.01, * p<.05 NOTE: Numbers presented are odds-ratios and (standard errors). a Reference category is Roman Catholic. b Reference category is that the leader evaluates the HIV problem in their congregation to be worse than in the surrounding village.
Table 4.6 Logistic Regression Predicting Doctrinal Conversation Partners
91
Congregational Denominational NGO Other Villagers Muslim a 0.43 0.14* 0.19 0.42 (0.31) (0.11) (0.22) (0.29) AIC a 1.27 0.83 0.65 1.47 (0.74) (0.48) (0.42) (0.86) Pentecostal a 1.64 0.43 0.56 1.41 (0.98) (0.26) (0.39) (0.85) Mission Protestant a 1.39 1.05 1.32 1.66 (0.79) (0.59) (0.81) (0.96) New MP a 2.00 0.88 0.67 2.83 (1.20) (0.52) (0.44) (1.80) Balaka a 0.53 1.10 0.39 0.73 (0.23) (0.47) (0.20) (0.32) Rumphi a 0.63 0.97 0.71 1.09 (0.26) (0.38) (0.31) (0.46) AIDS problem worse 1.01 1.09 1.12 0.67 in village b (0.35) (0.38) (0.45) (0.24) AIDS problem same 0.73 1.02 1.30 0.51 in congregation (0.47) (0.65) (0.98) (0.34) and village b Intercept 1.40 1.14 0.57 1.72 (0.74) (0.60) (0.33) (0.93)
-2 Log-likelihood -115.94 -114.48 -89.75 -109.3 χ2 12.21 13.57 11.58 15.27 Pseudo R2 0.05 0.06 0.06 0.07 N 177 176 176 176 * p<0.05 NOTE: Numbers presented are odds-ratios and (standard errors). a Reference category is Roman Catholic. b Reference category is that the leader evaluates the HIV problem in their congregation to be worse than in the surrounding village.
Table 4.7 Logistic Regression Predicting AIDS Conversation Partners
92 NOTES
1 While the cited study focuses on estimates of relational structure and maps of relational space per se, the content of those relationships and projections of beliefs, particularly within groups of which a respondent is a member, are ready extensions of such findings (Horowitz 1983).
2 Friedkin continues these statements to point out the reasoning behind the respectively being that “invisible opinions cannot be directly influential” and “irrelevant or valueless opinions cannot directly influence i” (Friedkin 1998: 68).
3 Other important measures of intraorganizational influence are hierarchical authority and resource control (Astley 1984). I will focus on the former of these in the next chapter, and contend that the latter is largely beyond the scope of this project.
4 For the complete networks section of the MRP survey, see Appendix B.
5 That these leaders are located in rural settings is particularly relevant to several of the points elaborated later in this chapter. For example, in the United States, governmental or NGO leaders could as readily be found in virtually any contexts. In the rural settings at the center of this study, local religious organizations are generally the only organizations present in these settings (Green 2003), and representatives from other organizational types are regularly cast as outside influences, because of the very fact that they are not proximally located (Watkins and 2000).
6 This grouping of the discussion into reduced categories both serves to preserve anonymity of the responses provided by individual respondents, and to simplify the presentation of a vast amount of information.
7 It is for this reason that I include most of the observed information, rather than selecting particular variables, and building models in step-wise fashion. I will highlight the cases where the specific differences do represent a potentially salient pattern.
8 Church of Central Africa Presbyterian (CCAP) is the single largest of the Mission Protestant denominations.
9 The third large single denomination is Islam. The comparatively higher involvement in meetings with NGO leaders, for Roman Catholic and CCAP leaders in our sample, does not however translate into any comparative advantages in receiving aide from these organizations (see Panel 3).
10 The New Mission Protestant disadvantage on this variable is driven largely by Jehovah’s Witnesses and Church of Christ congregations (who each report no such partnerships), while the Mission Protestant advantage is driven largely Baptists (who virtually universally, 92%, report such partnerships). The Baptist advantage on this variable may derive in part from the importance of the Providence Industrial Mission (a Baptist sponsored organization) in Malawi’s gaining independence, and related celebrations of those events (on top of other “religiously themed” co-participation). The founder of PIM in Malawi, John Chilembwe, was an early revolutionary pressing for Malawian independence,
93
which is now honored in part with the inclusion of his picture on the front of all denominations of Malawian kwacha banknotes.
11 As can be seen in Appendix B, the number of network ties that respondents could nominate was capped at seven, however interviewers were instructed to record additional responses if they were provided. There is substantial debate within existing network literature about the ways such caps can bias estimates of network size (Erickson 1983; Erickson 1981; Granovetter 1977; Marsden 1990; Marsden 2003; Wasserman 1994). In MRP data, however, only 1 respondent exhausted the 7 slots available on the survey, suggesting that these networks are artificially truncated by the instrument, or inflated to fill the available options.
12 Further, because of the vast number of Pentecostal denominations, the “same- denomination” ties for Pentecostal congregations were actually more liberally coded than other categories, including all Pentecostals as same-denominational alters, rather than selecting only on the specific denomination of the respondent’s congregation. This should not skew the information presented in any particular direction since Pentecostal congregations have only very recently started finding their way into the rural settings of the MRP sample. Therefore, like other denominations, there are also no villages with multiple Pentecostal congregations, even with this more liberal coding of denominational matches.
13 The difference between whether leaders engage congregational and denominational leaders is not significant, though in he same direction, for Roman Catholic or Mission Protestant congregation leaders. Similarly, the difference in number of doctrinal conversation partners is also insignificant when looking only at Roman Catholic or AIC congregation leaders.
14 Being “born again” has been the focus of several previous articles on religion and HIV in Malawi, suggesting a unique identity among individuals who self-identify or congregations that promote such beliefs (Watkins 2004). In these models, it only had an impact on the number of congregational leaders reported as connections apart from or doctrinal discussion partners – positively predicting such ties. This variable suggests that “born- again” leaders therefore have more informal interaction with other religious leaders that is not necessarily specifically religious in nature, but also not specifically tied to HIV.
94
CHAPTER 5
WHAT’S LILONGWE GOT TO DO WITH IT?
COMPARING RELIGIOUS HIV DISCOURSE BETWEEN
DENOMINATIONAL AND CONGREGATIONAL LEADERS
In this chapter, I investigate how readily the HIV-prevention strategies
described by local congregation leaders in the sample villages of the Malawi
Diffusion and Ideational Change Project match the strategies described by national
level leaders of the particular religious organizations within which they are
embedded. I draw on qualitative interviews (2005) and survey data (2005-06) from
congregation leaders in the Malawi Religion Project, and qualitative interviews
(2005-06) with the national leaders of the denominations represented in that
sample. Building on the previous chapter's finding of locally clustered networks, I
examine how readily this translates into (in-)consistent HIV-related messages
across varying levels of religious leadership. I pay particular attention to the
differences that exist between formal and informal recommendations.
95 Background
The existing scholarly criticism of religious organizations’ involvement in
HIV-intervention efforts in SSA often treats religious organizations as monolithic
structures.1 Teachings of religious traditions are frequently treated as though they
contain high, if not complete internal consistency within any given denomination,
and sometimes even within a religious perspective as a whole (e.g., Christianity or
Islam). The edicts of a particular organization are commonly presumed to funnel
directly from the top position in the organization to each of the leaders at other
levels in the organization – if not to the individual adherents themselves. Perhaps
the most obvious example of the attention devoted religious organizations in
popular criticisms and the scholarly literature focuses on how (un-)willing they are
to prescribe condom use, though empirical support for this claim is limited.
The ABC’s of HIV prevention is a widespread effort suggesting that the
best way for an individual to protect themselves from HIV is to Abstain from sex
before marriage, Be faithful to one partner in marriage, and use a Condom if A&B
fail (Singh 2003). The Catholic church’s (and several other denominations)
opposition to proscribing condom use is well described, though not similarly well
investigated. Increasingly, some organizations have recommended condom usage
in particular instances, such as within marriages for the protection of a non-
infected partner from their infected spouse, or as a means of family planning in
general, which also has implications for STD prevention. What is unclear however
96 is how well these (and other) proscriptions translate from the top levels of the
organization to the local level congregations, and subsequently to members.2
Caldwell and colleagues express a common perspective, which is also
frequently universally attributed to religious leaders, claiming that “[HIV] is a
punishment, in the words of the evangelical preachers, for fornication and adultery,
and more generally for disobeying the instructions of the holy books” (Caldwell,
Orubuloye, and Caldwell 1999). This view ascribing a “punishment” perspective
of HIV among religious leaders is frequently assumed to limit religious leaders’
contributions to HIV-prevention and care efforts. Recent evidence demonstrates
that this assumption is not accurate, with religious leaders frequently engaging
HIV-intervention strategies (Trinitapoli 2006b; Parry 2003). Each of these popular
criticisms infers a universal view of the way HIV is addressed within single
denominations (and sometimes even within religions as wholes). In this chapter, I
examine as an empirical question the consistency within denominations, something
to date that has largely been absent from the dialogue.
While financial donations are not the focus of this chapter, those models
are similarly presently based on a presumed hierocratic organizational structure.
The extension of findings in the level of agreement in information across religious
organizations, can potentially also contribute some suggestions to the effectiveness
of the existing financial models as well. These extensions, which obviously will
have to be more tempered than will discussion of the topic directly addressed here
will be considered further in the discussion following the analyses.
97 Central Questions for the Chapter
The main aim of this paper is therefore to describe the similarities and
differences that exist in several levels of HIV-related discourse between national
and local religious leaders. With this investigation, I then evaluate whether the
hierocratic organizational form - presumed in intervention strategies, the
congregational form - common in recent U.S. sociology of religion, or potentially
some other form, more readily matches the observed differences. While the
investigation of Malawian religious organizations’ treatment of AIDS-intervention
strategies here will provide several useful insights into the organizational structure
of denominations there, it is important however to keep in mind that this is one
relatively narrow aspect of the application of that organizational structure.
Therefore, in the discussion concluding this chapter, I will also consider how
readily the insights gained in these two chapters can be extended to religious
organizational structure more generally in Malawi. Additionally, I will also
hypothesize about how readily the hierocratic or congregational models may
extend to religious organizations in the rest of SSA, given that the existing
literature on the topic focuses almost exclusively on the United States.
Expectations
To date, SNA remains a largely positivistic pursuit, interested in explaining
the underlying social structure supporting particular outcomes. As such, data
collection and methodological toolkits used to gather and analyze social network
data are built to detect and analyze ties that somehow objectively exist. This is
98 appropriate for the vast majority of the topics studied with SNA - kinship patterns
can be genealogically established (White 1963); the diffusion of particular
innovations start in and spread from particular locations (Coleman and Katz 1966);
coauthorship entails multiple authors claiming ownership of a particular idea
(Moody 2004); trade involves the exchange of particular goods between two
parties (Keister 2001); information gathering requires a source and receiver
(Borgatti 2003); contracting an STD requires fluid exchange between infected and
uninfected individuals (Morris 1993).This list could go on.
Social influence has recently become a pertinent topic of interest in SNA
(Friedkin 1998), and in many approaches has adopted the very same techniques
available in other social network research. The approach gathers information on
ties that purportedly objectively exist and infers influence from those exchanges or
reported relationships. Influence however, like power or friendship, is not
something that one person has and passes to another, in the same was as applies to
body fluids or information. Similarly a conversation is not merely the coordinated
interlacing of two people alternately speaking and listening. I contend that many of
the religious leader's suggested HIV-prevention strategies mentioned in this
chapter are best conceived of as part of an ongoing conversation about HIV. Recall
the Baptist Pastor's account of HIV quoted in the introduction:
“AIDS is becoming like a common song that people are talking over whenever they meet with one another.” (Baptist Pastor, Balaka).
99 Conceptualizing, and therefore analyzing influence, friendship, or a conversation
should incorporate the ways that these sorts of relationships are different from
those that are the focus of other types of network ties (Mische and White 1998).
Turning to the HIV discourse presented in this chapter, HIV conversations
take place with high regularity in rural Malawi, given its high prevalence (Watkins
2004). These conversations are co-generated events that take place between two
(or more) individuals and produce content that belongs to neither of its participants
alone. As demonstrated in the previous chapter, local congregation leaders'
involvement in that conversation takes place largely at the local level. The vast
majority of the conversations that they take place in are therefore not with
denominational leaders, or other outside authorities, but rather with other religious
leaders, and villagers from their surrounding community.
Therefore, given the denominational leaders low hierarchical authority,
lack of resource control, and peripheral position in their religious networks (see
Chapter 4), congregation leaders likely devalue the salience of those
denominational leaders’ recommendations. Rural Malawian congregation leaders
construct and disseminate their HIV-related messages in a locally clustered
environment, which largely excludes national religious leaders' voices from the
conversation. This is not because those voices are devalued, but because they
simply aren't present. Therefore, their voiced opinions about HIV are much more
likely to reflect the ongoing conversation in their community with those they see
on a regular basis. Given this perspective, we should expect the alters with whom
these opinions should agree would be the other individuals engaged in that
100 conversation - other members of their local villages, their neighbors and family
members - not denominational authorities.
This Chapter’s Approach
In this paper, I therefore compare three types of HIV-related messages
between local congregation leaders and the national leaders of their corresponding
denominations to examine how readily the presumed hierocratic structure of
religious organizations fits the empirical reality. First I examine the leaders’
evaluation of the scope of HIV as a problem, in their community, Malawi and SSA
as a whole and within their particular religious organization. Next, I address their
respective evaluation of the problem, particularly as it relates to their religious
organization. In this section I especially draw on their diagnoses of the source of
the problem in their religious organization, and the discussion surrounding their
responses to questions evaluating whether, for those infected, AIDS is a deserved
consequence of their behaviors - addressing the statement quoted above from
Caldwell and colleagues. Third, I describe the leaders prescribed prevention
strategies for people in their congregation, focusing in part on their willingness to
incorporate condom usage into those strategies. For each of these three topics, I
examine the similarities and discrepancies in the way that these three topics cluster
within and across denominations.
For these comparisons, I take an approach that builds on concordance
studies - an important stream of literature in SNA research. In these studies,
researchers examine differing reports of a single social relationship as an estimate
101 of reliability of the existence of the named relationship. Previous studies find a
wide range of concordance estimates (Bell, Montoya, and Atkinson 2000; Bernard
and Killworth 1977; Killworth and Bernard 1976; Marsden 1990), particularly
noting that constraining the time window about which respondents are reporting
(adams and Moody 2007; Brewer et al. 2006), and focusing on more intimate
relationships (e.g., sexual partners as opposed to social acquaintances) produce
higher levels of consistency across multiple reports of the same relationship
(adams and Moody 2007; Bell et al. 2000). While most of these studies focus on
(dis-)agreement in reporting of the existence (or absence) of a particular
relationship, this approach can also be readily adapted to make comparisons of the
content of relationships, when both parties agree the relationship exists (e.g.,
shared identification within a single denomination).
Estimating Religious Organizational Structure
By leaving the empirical validation of denominational structure, and
variations within them, outside of the scope of existing research, there are
numerous assumptions left unexamined empirically. In this chapter, by
investigating the level of congruence in HIV-related messages within and across
denominations between national and local leaders as described above, I address
several of the potential pitfalls in leaving these concerns out of previous research.
First, as described above is how readily denominations actually display the implied
levels of internal consistency. Secondly, and perhaps more relevant to this chapter,
I examine whether denominations are the relevant organizational form for religious
102 organizations in rural Malawi. Finally, while previous research frequently resorts
to religious explanations for inter-denominational differences described, with my
analyses here, I also demonstrate the importance of organizational structure in the
differences observed - both in combination with, and beyond the doctrinal
differences between denominations.
These concordance estimates, based on the predictions from the previous
chapter - of a lack of salience and importance of hierarchical authority - predict
that there should be low levels of agreement between congregational and
denominational religious leaders' HIV-discourse. High levels of agreement
between religious leaders across levels of leadership within denominations would
suggest that these relationships demonstrate the salience assumed in present
hierocratic religious models underpinning most of the existing intervention
strategies. If, however substantial differences are present between leaders within
single denominations, the inferred corresponding social distance between the
levels of leadership would provide additional support for a more congregationally-
based model of religious organizations in this context.
Data & Analytic Methods
In this chapter I draw on the MRP interviews with local congregational and
national denominational religious leaders to describe their (1) descriptions of the
scope of AIDS as a problem, (2) evaluation of the source of HIV-infections, and
(3) suggestions for HIV-prevention strategies. I focus mainly on the qualitative
interviews with each of these leaders, which were conducted in 2005-2006. Each
103 of the congregational leader (N=173) and national leader (N=48) in-depth
interviews was read and coded using ATLAS.ti software (Muhr 2004) searching
for a series of topics including, but not limited to, direct mentions of HIV/AIDS,
illness in general,3 discussions surrounding family planning decisions and sexual
practices – particularly their support for / opposition to the ABCs. Software such
as ATLAS.ti provides two distinct advantages that were helpful in this coding
process. First, it allows ready application of a range of codes to large quantities of
text, excerpts from which can then be organized, selected, and retrieved for
analysis (Barry 1998). Second, it eases the necessarily iterative nature of the
coding process as additional themes emerge with subsequent readings of the
interviews.
For each of the analytic sections below, I also present introductory
descriptions of congregation leaders’ responses to related MRP survey questions.
These descriptive statistics will be drawn from the full sample of religious
congregational leaders (N=194), while only those who were interviewed in 2005
(N=173) and responded to the qualitative component of the interview are included
in the primary analyses in this chapter.4 As in the previous chapter, I describe these
differences, largely based only on the (six category) collapsed religious tradition
variable used in this project, both to simplify the discussion, and to protect the
anonymity of MRP respondents. The specific analytic comparisons that underpin
these summarized comparisons, though, are based on the more specific
104 denominational identifications.
Previous work in this region demonstrates that religious leaders do actively
discuss the problem of HIV both formally and informally, have developed
strategies for prevention, and have engaged congregations’ responses to results of
HIV infections (such as care of people living with AIDS – PLWA – funerals, or
AIDS-orphan care) (Trinitapoli 2006b). The quantitative descriptions from the
survey questions will help to illuminate how readily congregation leaders differ on
some of the concepts addressed. I will then explore these in more detail in the
descriptions of their similarities and differences as compared to their
corresponding national leaders.
Scope of the HIV/AIDS Problem
Within Malawian religious organizations, leaders at all levels acknowledge
that the HIV epidemic in Malawi is a substantial problem facing the population of
the country. There is however some variation among leaders regarding how they
rank it among the most pressing problems facing the country, with national leaders
commonly elevating it higher on the list than local congregation leaders. Further,
there is also variation in the “face” these leaders put to the problem, particularly in
how distanced they estimate the problem to be from themselves. While there is
variation on each of these components both within and across levels of religious
leadership, national leaders are slightly more likely to elevate the overall
importance of the problem but downplay the scope of the problem within their
particular religious organization than are local-rural congregation leaders. National
105 leaders commonly describe the epidemic as one of the most pressing issues facing
the country as a whole, however distancing themselves from the problem more so
than do local religious leaders.
[Insert Table 5.1 about here]
To summarize some of the views expressed by congregational religious
leaders in the MRP sample, Table 5.1 presents MRP leaders’ responses to several
questions describing their perspective of the scope of the HIV epidemic in general,
in their community and in their congregation. In leaders’ evaluations of the scope
of HIV as a problem compared to others faced by the congregation, there is
substantial variation demonstrated in Table 5.1 by religious tradition (p < 0.01),
however these differences are differentially associated across the three MRP
research locations (p < 0.001), and the denominational differences become
insignificant if examined only within any single location.5 Together these facts
suggest that local religious leaders are less likely to describe the HIV epidemic as
the most pressing problem their denomination faces than are national leaders,
though this gap narrows with corresponding increases in prevalence that we
observe across the MDICP/MRP research locations (i.e., in Balaka where
prevalence is roughly twice what is found in Rumphi (Obare 2005), leaders are
more likely to express HIV as among their most pressing problems).
For national and local leaders, poverty is the problem most commonly
mentioned as the most pressing issue facing their religious organization. Where
(and whether) HIV subsequently falls in their rankings of these issues varies
106 slightly across leadership levels. With large portions of the interviews for both
leadership levels focused on issues surrounding HIV, perhaps most telling in this
respect is the unprompted mentions of HIV (before HIV was raised by the
interviewer). National leaders virtually unanimously mentioned HIV, unprompted,
as one of the most severe problems facing their religious organization.
The major challenge is poverty. That’s the biggest problem. Members cannot offer as we want them to offer…Once they see someone who looks like he has means they gravitate to them and say “We want this, we want that.” So it is a big big problem. That is the major problem. Then HIV and AIDS is
the second. (New Mission Protestant Denomination, National Leader)
While some congregation leaders focused exclusively on poverty and hunger,
others did bring up HIV.
The problems here exist because when the parish was founded, people were used to donations. Those who provided these things are no longer here. They have left the congregation in the hands of its members. People are being taught to be self-reliant. Because of this, we don’t have money, so we don’t have fuel in our cars to visit people. So, the problem is that we don’t have money here at our parish. (Roman Catholic, Congregational Leader)
Problems that we receive from church members and others are emergencies – like hunger and poverty. Also disease - some diseases are incurable.6 (Baptist, Congregational Leader)
Often times these mentions are coupled, with the resultant problems being
compounded, with HIV producing increased needs, and a lack of resources
leaving those needs unmet.
The problems which we are facing in our church of CCAP are shortage of food. So, when we heard that other churches
107 are donating food to their members, as we have heard the Pentecostal churches do. So when our members see these things they come to us and complain saying, “Why can’t our church do the same?” We try to explain to them saying donation comes out of free… Because of this deadly disease, orphans are increasing fast so in our church we have more children who are orphans whose parents died of AIDS. Now these children need help and assistance. If we meet these children, we do try to give them the little we have because we do not have much. (CCAP, Congregational Leader)
For both national and local leaders, once prompted about whether HIV is a
problem facing their congregation, virtually every leader agrees that it is. One of
the more telling differences between local and national level leaders in relation to
the scope of HIV is in the distancing of their particular religious organization from
the source of HIV. National leaders are more likely to frame the HIV problem as
one that their organization deals with because it is a general problem facing the
country as a whole, which they must therefore deal with. They are less likely to
discuss it directly as a problem found within their religious organization, such as
through acknowledging that their members are actually contracting the disease.
This distancing is much less common among local congregational leaders (see
question 3 in Table 5.1). This comes through more clearly in the next section,
which focuses on leaders’ evaluations of the sources of HIV as a problem.
Evaluating the Source of the HIV Epidemic
There is a strong assumption within some scholarly literature suggesting
that religious leaders judgment of PLWA is a significant source of stigma, rather
than representing these leaders as contributing to alleviating such problems.
Unfortunately, for the most part these are assumptions based largely on hearsay or
108 select observations (Caldwell 1999; Orubuloye 1993). In part, this perception is
built on the potential condemnation of those infected with HIV as getting what
they deserve, or having been judged by God for disobedience (usually through
sexual promiscuity). However, empirical research is limited on what religious
leaders actually say on this topic, where it comes from, and how readily religious
participants absorb it.
In this section, I demonstrate that national leaders are actually less likely to
hold such stigmatizing views than are local congregation leaders. While they do
recognize the role of sexual behaviors in the proliferation of the disease, they are
reticent to assign blame to individuals who have contracted HIV.
I have heard a lot of sermons to the effect that God has sent AIDS to punish sinners, and we need to repent so God will forgive us. Now, I believe that AIDS is progressing because of sin, but if we say that those that are suffering from AIDS is because of sin, then we will be making a big mistake. Plus it is not only sexual immorality that is sinful; killing is sinful, abortion is sinful. There are so many other things. I don't think that God can pick just one and say this particular sin I want to punish. (Mission Protestant, National Leader)
Some of the distinctions that national religious leaders make appear to arise from
their distancing the problem from their own religious organizations. This takes
place in two stages – first, national leaders are more likely to distance the source of
infection from its probable source than are local congregation leaders. They
frequently evoke explanations of HIV infections based on needles, blood
transfusions, etc. which are thought to contribute very little to the present HIV
epidemic in Malawi.7
109 I don't condemn those who have the disease, because some of them, you know, they got it in some weird ways. They are not loose. Some of them are victims of the disease. (New Mission Protestant, National Leader)
This is a disease, like any other disease…for example, HIV and AIDS, it's not only through sexual intercourse…but you know even blood transfusions, even taking blood, or using the same needles. I remember there was a time; there was an accident when we were driving from Mangochi, and we found people bleeding. We didn't even have gloves, but we said handle them carefully. Because we might have a cut, and if it is, we might get an infection. And you can't say that it was because of sin. But you know some have contracted this disease who were giving first aid or helping. (Mission Protestant, National Leader)
Second, for those individuals who they do acknowledge contracted HIV through
sexual activity, they are more likely to suggest that they contracted it within a
marital relationship or that these individuals joined their church after infection.
National leaders are particularly reticent to attribute the spread of HIV to sexual
promiscuity among members within their own denominations.
You can contract AIDS in different ways. For example maybe through adultery - then you know for sure that you have sinned. But sometimes you can get it from your husband or wife, so the problem is with not you but you still have the disease. That is why we still have to differentiate that. There the sin can be for the spouse who has brought AIDS to his partner or wife. (Mission Protestant, National Leader)
Congregation leaders are more likely than national leaders to attribute
sexual sources of HIV-contraction, and “blame” the individuals who are positive,
something that even national leaders recognize.
110 I would say that in our villages, most people would say, this is a punishment. The stand of the church is that it is a collective sin, because the people who are getting it are our people." (Pentecostal, National Leader)
He continued, explaining that caring for individuals with HIV is something that
should not vary according to how individuals got the disease in the first place.
These leaders may be aware of the popular criticisms painting religious leaders as
contributing to HIV-related stigma, or their awareness of (or at least their
perception of) local religious leaders' framing of the HIV-sin connection. National
leaders' distancing on this particular issue may partially be in response to these
issues. In these interviews they frequently implicitly attempted to de-stigmatize
any individuals within their denomination who have contracted HIV, as is the case
in the latter half of the quote above, even in the situation of strong denominational
positions that may suggest otherwise. Perhaps the greatest irony of this particular
"protective" distancing, on the part of national religious leaders in Malawi, is that
within MRP congregation leaders, despite relatively high levels of "sin" and
"blame" attributions for those who contract HIV, these leaders, and their
congregations are still highly engaged in care for PWLA, suggesting that this
particular issue is of little concern, at least in the MDICP/MRP research locations
(Trinitapoli 2007).
Congregation leaders do not display nearly as much distancing of the
problem, instead focusing almost exclusively on the sexual sources of contracting
HIV (whether implicitly - as in the first quote below, or explicitly - as in the
111 second). They are also much more likely to use phrases attributing HIV-infection
to sinful behavior or God's judgment.
Sometimes when we are reading the Bible we say the reward of sin is death. And if you are walking with a weapon you will die with the same weapon. (Roman Catholic, Congregational Leader)
I believe AIDS and sin are the same because those who do not comit adultery cannot get aids so it is like they are the same. (New Mission Protestant, Congregational Leader)
So that referring to the Bible we are just assuming that it is the epidemic that God promised which God said that it will be incurable. (Roman Catholic, Congregational Leader)
What I know is that AIDS is a combination of diseases that is gained by practicing unprotected sex. (Islamic Sheikh)
Virtually all (86%) congregation leaders view those who have contracted
AIDS through sexual promiscuity (commonly referred to as being movious8 in
Malawi) as haven "gotten what they deserve” (see question 1, Table 5.2). As
mentioned above however, this perspective - commonly perceived to be
stigmatizing - has little effect on their subsequent care for PWLA. The leaders'
views on this topic also appear to have little effect on what congregation members
themselves believe, as they respond to this question with substantially lower levels
of agreement (Trinitapoli 2007). Further, congregation members substantially
underestimate the levels with which their religious leaders espouse such views
(Trinitapoli 2007), suggesting that it is not something that leaders frequently
preach, or if they do, it is not something that their congregation members readily
absorb. 9
112 [Insert Table 5.2 about here]
While national leaders are more likely to evaluate promiscuity as a problem
“out there” contributing to the prevalence of HIV as a whole, but less focused on
the issue within their particular denomination, local congregation leaders do not
similarly distance themselves, or their congregants from the problem.
Congregational leaders report high levels of sexual promiscuity among the people
in their villages, and in their religious congregations. In fact, while Table 5.2
displays several significant differences by religious denomination in the reporting
of congregational and community levels of marital infidelity and adolescent
promiscuity (Pentecostals and New Mission Protestants are slightly more likely to
favorably evaluate their members than other congregation leaders from other
traditions, p<0.05), virtually all of those differences are rendered insignificant
when examined within particular districts. Further, of these leaders’ evaluation of
these problems for community versus their congregation members, only
Pentecostals estimate their congregation members to be slightly less promiscuous
than other leaders. However, if we compare a single leader’s responses to the same
question regarding their congregation members to the community around them,
none of the differences are significant.10 In other words, all local congregation
leaders are likely to estimate that their congregation members are like the other
members of their communities regarding levels or pre- and extra-marital sexual
partnerships. Conversely, national leaders were more likely to estimate sexual
promiscuity as a problem among the general population, not as likely to be
113 affecting their particular denominational affiliates, as evidenced in the quotes
above.
Proposed Religious Organizational Prevention Strategies
The final comparison I examine between religious leaders at the national
and congregational levels is what they proscribe as the course of action individuals
should take to protect themselves from contracting HIV. Since the dominant
paradigm for approaching prevention keys in on individual behaviors, as expressed
in the “ABC”s of HIV prevention, I focus especially on the importance placed
each on abstinence, faithfulness and condom use by each of these leaders. I
demonstrate below that national religious leaders are more likely in their
behavioral recommendations to espouse absolutist positions than are their
congregation leader counterparts. Focusing in particular on the willingness of
leaders to suggest that individuals use condoms as a means of protecting
themselves, I find considerable variation across denominations, which roughly
corresponds to expected patterns, but variation within denominations that is much
higher than previously expected. Further, I find that local congregation leaders,
even in the “strictest” denominations are more willing to employ contingency
recommendations (particularly condom use) than are national leaders. Table 5.3
presents a summary of the frequency with which religious leaders teach particular
HIV-prevention strategies, focusing especially on the ABCs.
[Insert Table 5.3 about here]
Religious leaders all focus on varying components of the ABC's of HIV-
114 prevention, though to varying degrees. Abstinence before marriage and monogamy
within marriage are the cornerstones of prevention strategies for both national and
local religious leaders. National leaders, however, are much more likely to espouse
"absolute" positions regarding what should and should not be taught within their
religious organization. The majority of the attention in discussions with national
leaders, not surprisingly focused on their position related to the prescription of
condom usage. Most of the national leaders were reluctant to include condoms
among the repertoire of prevention measures supported by their denomination.
We don't want Christians to divorce each other…We are trying to very much encourage marriages to continue. Because our work is to make people marry. When people come together and we wed them, it's no use of wedding them then letting them go and undo the marriage. Now, with regards to condoms, Christians should not use, it is not wise to use the condoms. Because if we are dealing with Christians, what we tell them is to be honest with their wives; to be honest with their husbands; not to be running around with everyone. It's not our teaching to our member Christians that you can use condoms. (Mission Protestant, National Leader)
Those who do discuss the use of condoms do so with a clear qualifier that it is
meant for family planning methods, not for use by un-married individuals.
Whenever I go out, I always mention this pandemic. Not to scare them, but we want to get rid of it. We always teach the young ones to wait until when they get married. I am not one of those advocators of condoms. Condoms, from my own teaching, should be used [only] for family planning." (Mission Protestant, National Leader)
We are telling them that the best way is abstinence. We do not encourage our young people to use condoms. We are telling them, please abstain. But for those, for the sake of
115 family planning, and everything, yes we tell them to go ahead. For them, they can use the condom. (Pentecostal, National Leader)
While this "A and B, but not C," refrain, was common among
national leaders across most of the denominations, they did mention several
specific exceptions to the rule. The most common exception mentioned
addressed the situation in which one spouse is known to be HIV-positive.
…without consecrating the option of divorce, [they] should be using the condoms, you see? So, rather than to talk about divorce, we would rather encourage the use of condoms. (Mission Protestant, National Leader)
This exception, when mentioned however, was explicitly tied only to cases of
known discordant HIV-status. It is important to point out, with respect to this
requirement, according to the 2004 Malawi Demographic and Health Survey, only
5.6% of rural women and 12.3% of rural men have been tested for HIV and have
received their test results (NSO and Macro 2005).
Most relevant to the question addressed in this chapter however, is how
well those proscriptions at the national level translate to recommendations among
local congregational leaders. One particularly illuminating example of the
discrepancy on this particular subject comes from national and local leaders within
the Roman Catholic Church.
That is non-negotiable as far as the church is concerned…the only thing that the church accepts, according to the teaching is natural planning. Any artificial planning is not accepted; it is against the Catholic teaching
116 and doctrine. Because…God created people to procreate. Yes we can control, but we have to use the natural way, not an artificial way of controlling.
…I think as far as the messages are concerned, I don’t think there are any differences from other organizations; we have the standardized message…with the exception of one – the issue of condoms. We don’t talk about it, but if someone talks about it, we have to come in and say, this we don’t accept. That one is for us, I wouldn’t say non- negotiable…We don’t make a general ruling, but the condom is the crucial issue that the Catholics don’t deal with. We have got our doctrine and we have to stick to that. (Roman Catholic, National Leader)
Local leaders are not so absolute in their views, and are much more
likely than national leaders to discuss cases where condom use is
acceptable.
We have problems like people who do not manage to abstain. Previously I was keeping condoms, capsules, creams and they used to come to collect such things for their personal safety. As for condoms? We have bales and bales of them. (Roman Catholic, Congregational Leader)
I do not want to paint too-rosy of a picture regarding condom promotion though; as
indicated in Table 5.3, less than a third of local religious leaders overall
recommend people use condoms as a method of protecting themselves from HIV.
Our denomination does hate condom use because that is against our faith. Our point is that the person has to have self-control, but if he/she wants to get a sexual partner, one has to choose a man or a woman who hasn’t a bad record. They should choose someone who he/she knows he/she is going to stay with for a long time. They shouldn’t choose someone, who after you marry, you can hear other people saying you have married a person who has AIDS. They want a person who has self-control starting from the beginning so that they can be living together and not going around with everybody. All the days I do tell them that this is going to
117 put then into Gehenna. (Islamic Sheikh)
However, they are much more likely to qualify the “A&B” portions of the popular
prevention message to adapt for cases where individuals may not perceive of
abstinence and fidelity as feasible options.
One additional recommendation that came up - more among congregational
leaders than national leaders - is using HIV-testing as a means of self-protection.
In particular, it is increasingly an informal recommendation that when a young
man and woman want to get married, they should first go for testing.
When these children who belong to our church are on holiday they do go to retreat where they are advised that when they want to get married they have to go for blood tests first; but this is not said in the church. (Mission Protestant, Congregational Leader)
For both local-congregational and national-denominational religious
leaders, virtually all of the proscribed prevention efforts focus on one of the
components of ABC - more "A" and "B", than "C", while few also mention the
potential role of HIV testing and knowing one's (or one's spouse's) HIV status, as
ways they suggest people within their religious organization can protect
themselves from contracting HIV. In the discussion below, I will focus on how, for
congregation leaders, this fits well within the other ways that they discuss their
experience and views of the epidemic, while national leaders' positions paint a
picture that is a little more obscured.
Summary of Findings
Religious leaders at the local and national levels in Malawi regularly
118 address the issues relating to HIV/AIDS with people in their religious
organization. While popular accounts, and some scholarly attention suggests
otherwise, leaders are not shying away from this topic. In fact, most congregation
leaders (72%, MRP) describe sexual morality and issues directly concerning
HIV/AIDS as among the topics they broach weekly in their congregation's
services.11 National leaders consistently rank HIV/AIDS as among the most
pressing issues their denominations face. Continuing to suggest that religious
leaders avoid addressing HIV would require ignoring the empirical evidence.
What these leaders say about HIV, however, varies considerably across
levels of religious leadership. Rural congregational leaders are entrenched in local
village contexts, and evaluate the problem as one facing the people they know and
interact with regularly. Their referent community is their village, and those close
by; it is not tied exclusively to their congregation or denomination. As a result,
their descriptions of the how their congregation faces the epidemic are tightly knit
with their descriptions of how their community as a whole does so. Despite
professing more judgmental views about PLWA than their congregants or national
leaders, they still respond with compassion to them, and encourage others to do
likewise. These leaders craft careful scripts that negotiate the particulars of the
problem as they see it, and enact every means available to promote change - even
promoting condom usage more readily than popular accounts, or national leader
perspectives would lead us to expect.
National denomination leaders also devote substantial attention to the HIV-
119 epidemic - unlike congregational leaders - almost universally including it among
their denomination's most pressing problems. However, their discussion of its
manifestations, and subsequent proscriptions for prevention efforts reflect their
distance from the vast majority of their constituency, and perhaps better reflects a
problem they see as one "out there" facing the country as a whole, rather than one
to which they are intricately linked. They are more likely to evoke iatrogenic and
non-sexual explanations of HIV transmission than evidence supports. They do this
at least in part in an attempt to de-stigmatize PLWA. However, their subsequent
focus on prevention strategies are much more likely to the reflect religious (and
international) "party line" - i.e., "A" and "B", but not "C" - in absolutist tendencies
that do not reflect these peculiarities.
The differences between local congregational leaders and denominational
authorities HIV-related discourse subtly reflect their relative positions in the
national landscape. An especially reflective example of this distance between
religious leaders comes from one common national leader response to questions
about their views on HIV/AIDS, and particularly condom use. A small majority of
these leaders mentioned position papers or policy documents that their
denomination adopted as a whole.12 Not one of the local congregational leaders we
interviewed mentioned their use of similar documents when asked similar
questions. Either these documents are not finding their way to the local leaders, or
if they do, the leaders deem them irrelevant for their situation. Both of these
potentials support the image threaded through discussion in this chapter - that, at
120 least related to the topic of HIV responses - rural Malawian congregations are
rarely impacted by any form of religious hierarchy.
Conclusions
Denominational authorities, who are virtually all located in either Lilongwe
(Malawi’s capital) or Blantyre (the other major city – located in Malawi’s southern
region),13 recognize the importance of the AIDS problem in Malawi, but are
distanced from the particularities of the problem. The way they discuss the
epidemic and potential interventions is likewise – distanced, both from the
problem itself, and in the connections between their evaluation of the problem and
their proposed responses to it. Congregational leaders’ discussion of HIV is not
substantially influenced by their national denominational leaders, is more directly
linked to the scope and source of the problem they observe, and more readily
connects to the prevention suggestions they make. The largest difference between
the two levels however, is that there is much greater room for difference in the
informal recommendations that congregation leaders make than is the case for
national leaders. Congregation leaders are simply more likely to see the people that
the epidemic is affecting more often, and respond in ways that reflect this
difference.
Studying Religious Organizations
Combined, these two chapters demonstrate that for the case of HIV-related
discourse and response among religious leaders in rural Malawi, strong
organizational hierarchies are not important. Congregational forms of religious
121 participation have recently become the primary focus of many scholars in the
sociology of religion. This takes place for two primary reasons that are particularly
relevant for thinking about how the local clustering found in the rural Malawian
setting – could translate into other studies of religion in SSA, and in developing
contexts more generally. First congregational studies benefit from the ability to
combine local context into the way religious organizations are studied
(Ammerman 1997). In a similar way that Presbyterian churches in Boston are
substantially different from those in Chicago; so too is the experience of members
of an AIC congregation in Balaka substantially different from those from an AIC
church in Rumphi. Second is that congregational analyses allow more fine-tuned
studies of organizational form and religious content in the ways that they differ
across religious congregations, even within the same denomination (Chaves 2004).
While suburban mega-churches orient their religious programming substantially
different from centuries old inner-city congregations, or small rural parishes in the
United States; our studies should allow for similar differences between
congregations at the trading centers located near major crossroads or between two
large villages, and those located in more remote settings.
Both for scholars interested in studying religious communities in SSA, and
those who wish to include religious participation in models of other outcomes,
studying religious organizations as congregations in these settings would more
readily reflect factors likely related to the outcomes they are interested in studying.
122 It misrepresents these religious organizations, and is likely a disservice to any
studies including measures representing them, to continue to paint with broad
strokes descriptions of the importance of religion – or even the differences across
varying religious denominations – in SSA.
HIV Policy
These two chapters set out to describe the religious organizational setting
that contributes to HIV-prevention strategies in rural Malawi in part to examine
whether the empirical reality matches the assumptions underpinning most present
policy strategies. Largely, it appears that present models are built on a presumed
organizational hierarchy, which for religious organizations does not exist. The
question then remains whether present policy models are (1) still effective despite
this mismatch, or are (2) ineffective - whether (a) due to this mismatch or (b) due
to other reasons that have nothing to do with religious organizational structure.
Answering this question lies largely beyond the scope of this project, but I would
like to present a preliminary answer drawing on some of the existing knowledge
from other work.
Reponses to the HIV epidemic are having an impact on HIV, and the
contributions of religious communities are among the strongest predictors of
having extramarital sexual partners, and even HIV status (Trinitapoli 2007).
Unfortunately, at this point it is, however, unclear whether this reflects
comparative successes - with particular ROs especially well equipped to respond to
the epidemic - or if it simply reflects a “reshuffling of the deck - with some ROs
123 attracting people who are HIV-positive to their congregation faster than others.
Future analyses using the soon-to-be-available longitudinal data on religious
affiliation and participation coupled with multiple measures of sero-status and
other behavioral markers can perhaps better illuminate this distinction. So the
verdict on whether policies, which funnel large sums of their resources through
religious organizations, are having effect, is still murky – but appears to suggest
that there is a little of both, perhaps leaning more towards beneficial effects.
In terms of what generates the ineffectiveness of those programs that do
fail, I suggest two primary avenues for future research to tease this out. The first
draws again from the evidence above that religious leaders are having a substantial
impact on the epidemic. Those religious leaders who are actively engaged in
combating the epidemic – whether through preaching, policing or whatever
avenues – appear to be having a substantial impact. Further, they appear to have
most of the information they need to provide accurate behavioral
recommendations. However, many of these leaders are still remiss to formally
apply all of the tactics at their disposal. For example, while more religious leaders
are promoting condom use than we would have expected prior to fielding the
MRP, many still are not. And perhaps more problematic – those who are,
frequently only do so informally. The mixed messages that come through formal
and informal channels may contribute to the well-known lack of association
between knowledge of AIDS risk and subsequent behavioral adaptations. While
recent research demonstrates that religious leaders are having a positive impact on
124 their congregation members’ risk of contracting HIV, they may simultaneously be
driving others of their congregation away from these protective effects.
Another important angle for future research to examine is to move beyond
simply focusing on the impact of religious organizations on behavior change.
There are other important factors that contribute to the spread of the epidemic,
such as poverty, and familial patterns – which religious organizations in this
setting also likely can contribute to, and they potentially can have a more direct
impact on the fight against HIV. Further, it is also feasible that without changing
some of these larger supra-individual contributors to the progress of the epidemic,
we will not see substantial changes in the present epidemic even if individuals
largely adhere to behavior prescriptions. In particular the next chapter
demonstrates that the behavioral changes at the heart of virtually all present
intervention models – the ABCs are, afterall, an exclusively individual, behavioral
approach – can, in particular ways, generate little-to-no actual change in
individual’s risk of contracting HIV.
125
NOTES
1 It is important to note however that this practice is not limited to scholarship linking religion and HIV interventions. For example, evaluations of governmental responses to HIV frequently focus on formal policies and the existence of organizations or task- forces to address the epidemic (USAID 2003). Unfortunately, little has been done to evaluate how readily such policies generate changes at other levels of organization (national vs. local governments) or how readily observed changes are directly linked to such efforts (Martin 2003). For example, while there have been wide-circulated reports of a decline in HIV prevalence in Uganda (Green et al. 2002; Kirungi et al. 2006), there is little consensus about the source of observed changes (Brody 2004; Green et al. 2002; Hallett 2006), and even debate about whether those shifts are in fact genuine or merely methodological artifacts (Parkhurst 2002). Similar limitations exist in other areas as well, not just religious and governmental responses to the epidemic. As such, the critiques found in the argument here should not be interpreted as unique to religious organizations, only unique in the particular substantiations I describe for this particular setting.
2 For one way this resource flow (and the corresponding “skimming” that takes place at each stage of the process) has been conceptualized, see Appendix D.
3 Previous research demonstrates that while pastors do frequently talk about HIV specifically by name, they also commonly employ euphemisms (such as “this illness”) for the disease (Pfeiffer 2004; Trinitapoli 2006b).
4 Limiting the quantitative responses to only those congregation leaders also represented in the qualitative interviews does not substantially change any of the differences presented, except where noted below.
5 The non-significant difference could be the result of a small-N problem, and substantial differences may remain by denomination. There are not enough congregations of each denomination within each of the research locations to approximate standard levels of significance for each denomination separately.
6 As mentioned above (see Note 3), this “incurable disease” is likely a thinly-veiled reference to HIV, which most listeners would recognize as such, and not as a reference to other diseases.
7 While there is some recent debate over this contention, with several scholars suggesting that non-sexual sources of HIV infection in SSA should be more directly considered (Gisselquist et al. 2004, 2002; Brewer et al. 2007), it is still widely accepted that the vast majority of HIV transmission in SSA occurs through heterosexual sexual contact (UNAIDS 2006; Green 2003).
126
8 “Promiscuous” is not a perfect translation of the word movious, though it captures most of the relevant connotations for the present discussion.
9 Congregation members largely project their own levels of agreement with such statements onto others, under-estimating the degree to which their religious leaders espouse such views (Trinitapoli 2007). Congregation leaders also seemingly project their own views onto those around them, conversely over-estimating the degree with which those in their congregations and villages agree with this statement.
10 The only difference that approaches significance is that Pentecostals are marginally more likely to estimate their congregation members to be less promiscuous than their surrounding community (p < 0.10). This near-effect only applies to estimates of adults’ extra-marital partners.
11 Evidence presented elsewhere suggests that while these leaders may moderately over- report the frequency with which they address HIV-related topics, they do not over- exaggerate whether they engage these issues (Trinitapoli 2006b).
12 One leader, in fact, would not elaborate further; instead stating that policies directly follow what is written in their denomination’s document (which he co-authored).
13 Two of the leaders interviewed as national representatives were actually located in Mzuzu (a much smaller city in the northern region). Both of these, however were regional leaders representing organizations without a central national coordinating office.
127
Mission Catholic Prot. Pentecostal AIC Muslim New MP Overall How worried do you think people in your congregation are about AIDS? None 0 5 22 15 0 18 11 Little 10 5 9 20 5 24 13 Lot 90 90 69 65 95 58 76
Compared with other problems your congregation faces, how big of a problem is AIDS currently? Not a problem at all 0 7 12 7 9 26 11 Somewhat of a prob. 29 24 42 29 0 40 29 Big problem 48 41 24 39 41 20 35 Single biggest prob 24 27 21 24 50 14 25
Is the AIDS problem in your congregation more or less of a problem than in your village? Less of a problem 10 3 6 10 5 9 7 No difference 48 55 73 63 60 60 60 More of a problem 43 43 21 28 35 31 33
Do you think the AIDS problem in your congregation will…? Get better 62 77 72 78 65 74 73 Stay the same 5 3 6 3 5 13 6 Get worse 33 21 22 19 30 13 22
Do you think the AIDS problem in your village will…? Get better 50 45 41 68 60 50 52 Stay the same 0 3 13 3 5 3 4 Get worse 50 53 47 29 35 47 43
Proportion of congregation deaths in past 12 months attributed to AIDS – Mean & (Std. Dev) Adults 0.42 0.26 0.41 0.36 0.15 0.38 0.34 (0.24) (0.30) (0.44) (0.39) (0.21) (0.07) (0.37)
Children 0.08 0.02 0.02 0.01 0.04 0.07 0.04 (0.37) (0.06) (0.07) (0.05) (0.10) (0.22) (0.14)
N 22 41 33 41 22 35 194 NOTE – All numbers are column-percentages, except where otherwise noted.
Table 5.1 MRP Congregation Leaders’ Description of the Scope of the HIV Problem
128
Miss. Catholic Prot. Pentecostal AIC Muslim New MP Overall
Those who are movious and got AIDS through sex have gotten what they deserve Strongly disagree 14 8 21 13 5 9 12 Disagree 0 10 24 15 14 6 12 Agree 41 26 21 18 23 29 25 Strongly agree 45 56 33 55 59 57 51
People in your congregation feel that those who are movious and got AIDS through sex have gotten what they deserve Strongly disagree 5 10 23 8 10 6 10 Disagree 19 10 23 23 14 15 17 Agree 38 35 26 33 33 38 34 Strongly agree 38 45 29 38 43 41 39
People in your village feel that those who are movious and got AIDS through sex have gotten what they deserve Strongly disagree 14 18 24 10 5 9 14 Disagree 18 5 21 18 23 15 16 Agree 50 38 18 28 27 27 31 Strongly agree 18 40 36 45 45 48 40
Marital infedelity is rampant among people in your congregation Strongly disagree 18 34 24 23 14 34 26 Disagree 18 24 58 41 43 37 37 Agree 36 24 0 21 19 26 20 Strongly agree 27 18 18 15 24 3 16
Marital infedelity is rampant among people in your village Strongly disagree 5 28 9 8 14 21 15 Disagree 18 10 22 32 43 21 23 Agree 50 28 31 26 24 24 30 Strongly agree 27 33 38 34 19 35 32
Promiscuity is rampant among adolescents in your congregation Strongly disagree 24 16 34 15 20 30 23 Disagree 19 26 44 50 25 52 38 Agree 43 39 9 20 15 12 23 Strongly agree 14 18 13 15 40 6 16
Promiscuity is rampant among adolescents in your village Strongly disagree 18 8 19 18 5 23 16 Disagree 14 5 16 18 25 20 16 Agree 32 26 19 26 10 34 25 Strongly agree 36 61 47 37 60 23 43 NOTE – All numbers are column-percentages, except where otherwise noted.
Table 5.2 MRP Congregation Leaders’ Evaluation of the Source of HIV as a Problem 129 Catholic Miss. Prot. Pentecostal AIC Muslim New MP Overall
Abstinence Sex before marriage is unacceptable if the 95 84 94 85 91 91 89 couple loves each other a
Do you ever privately advise members of your 100 98 97 90 91 97 95 congregation to be faithful or stop promiscuous behavior
Be Faithful Sex outside marriage is unacceptable for a 100 100 97 100 95 100 99 married woman a 132
Sex outside marriage is unacceptable for a 100 100 94 100 95 100 98 married man a
Condom Use Do you ever privately advise members of your 23 15 27 35 59 11 26 congregation to use condoms
Other Do you ever privately advise members of your 29 34 21 38 59 20 32 congregation to leave spouse suspected HIV+
Do you ever privately advise members of your 77 83 61 65 59 51 66 congregation to get tested for HIV a These questions are reverse coded from the actual survey question. All numbers presented are percent of yes responses.
Table 5.3 MRP Congregation Leaders’ Suggested HIV-Prevention Strategies
130
CHAPTER 6
DAMNED IF YOU DO, DAMNED IF YOU DON’T:
RELIGIOUS AFFILIATION AND HIV RISK NETWORK STRUCTURE1
Introduction
Few people would question that sub-Saharan African is the epicenter of the
present HIV/AIDS pandemic. Prevalence rates for countries in this region are
consistently higher than any other region of the world, with as much as 40 percent of
the population of some countries being presently infected (UNAIDS 2006). In the
case of Malawi, a small landlocked country in southeastern Africa, present estimates
suggest that roughly 14 percent of the population is infected (NAC 2004). In most
countries in SSA, the epidemic is a generalized one, in that it is passed primarily
through heterosexual intercourse, has a male-to-female ratio of approximately one-to-
one and frequently results from perinatal transmission (Green 2003). As a result,
intervention efforts frequently focus on behavioral modifications, particularly the
ABCs (abstinence, being sexually faithful, and condom use).
Religious participation in the region is virtually ubiquitous (Bedaiko 1995),
and a growing body of recent research in SSA demonstrates empirical associations
131 between religion, mortality and health (Gregson et al. 1995; Lagarde et al. 2000;
Pfeiffer 2002). The literature addressing the link between religious participation and
health outcomes, including specifically HIV in SSA, focuses almost exclusively on
the relationship between religious affiliation (or participation) and subsequent
individual behaviors, attitudes and risk. As one recent example, in an attempt to move
away from conceptualizing religious participation as a solely individual phenomenon,
the "moral communities" thesis has shown that religion can be thought of more as a
group phenomenon than an individual one (Stark 1996), and correspondingly can
reduce particular risk behaviors (Ellison, Burr, and McCall 1997; Pescosolido 1990;
Pescosolido and Georgianna 1989; Regnerus 2003). Even these studies, which argue
for conceptualizing religion as a group level phenomenon, then turn to apply it to the
study of risk as an individual level outcome. For example, this thesis has been applied
to the study of HIV in SSA (Trinitapoli and Regnerus 2006), by estimating
individuals' risk behaviors and individual's perceived HIV risk as a factor of
individual, community and organization level religious affiliation and participation.
This is an important step forward in understanding the link between religion and
HIV-related outcomes. However, this approach fails to incorporate the structural
properties of risk, something recent network epidemiologic work (Morris 2004) has
proven is vital to understanding HIV-risk.
I present the first work that incorporates this recent reconceptualization of
religion as a group level phenomenon into the network epidemiologic notion of risk
as a structural property. While reductions in risk behaviors are a vital component to
reversing the HIV epidemic, such behavior modification is only effective if
132 universally adopted. In this paper, using a series of simulations based on observed
sexual partnership data, I demonstrate how those who do not adopt such changes and
engage in “risky” behaviors can affect risk both for themselves and those who have
adopted behavior modifications such as marital fidelity. This focus will move beyond
the current literature’s focus on how religious participation is helpful or harmful in
regard to the present HIV epidemic, presenting one mechanism by which both can
happen simultaneously. I draw on data which establishes the link between religion
and HIV-risk behaviors to build simulation models of risk network structure. I
illustrate network properties that demonstrate how observed reductions in risk
behaviors, which have been shown in recent research to vary by religious
participation, can actually generate little-to-no effect on group level risk, and in
certain circumstances can even correspond to increases in group level risk.
Background
The literature linking religion and health outcomes has a long history, largely
focusing on the beneficial impact of religious participation on risk-related behaviors,
and subsequently on particular health related outcomes (Chatters 2000; Ellison 1991;
Hummer et al. 1999). These changes in health outcomes result from changes in
particular behavioral differences between religious groups, such as decreased
likelihood of smoking (e.g., Strawbridge et al. 1997) or alcohol consumption (e.g.,
Ford and Kadushin 2002). Others explore the development of social support and its
outcomes that coincide with religious participation (Idler 1995; Krause et al. 1998).
Of more direct importance to studying HIV-risk, still others have observed that
religious affiliation is associated with declines in extramarital sexual partnerships
133 (Hill et al. 2004) and delayed sexual onset (Rostosky, Regnerus and Wright 2003),
but also coincides with lower likelihood of condom use in sexual encounters (Agha et
al. 2006; Bearman and Bruckner 2001).
Recent scholars point out that the link between religion and health related
outcomes is not necessarily unidirectional, and that continuing to frame them this way
potentially limits researchers’ ability to understand this important relationship. Idler
(1995) raised this possibility, particularly suggesting that if health outcomes are
studied over time, different associations may be observed according to religious
participation at different points in time. To date, however, no studies have shown that
these pathways can occur simultaneously. Instead research focuses on different
behaviors that alternately improve or decrease health related outcomes. By remaining
with an individual focus, such "either/or" mentality is likely to remain. Researchers
who specifically study the link between religion and HIV follow the trend of
conceptualizing the link as unidirectional. If instead researchers in this area begin to
think of HIV-risk in terms that move beyond "risky" behaviors, and instead focus on
the structural properties of risk, we can arrive at a mechanism, which simultaneously
explains detrimental group level changes that coincide with helpful individual
behavioral adaptations.
A number of recent studies investigate the relationship between religion and
HIV. These studies evaluate the role of religious organizations in the development of
HIV prevention and intervention messages (Green 2003; Hearn 2002; Leibowitz
2002; Parry 2003; Pfeiffer 2004). To date, Trinitapoli (2006b) is the only researcher
to model individual HIV infection by religious affiliation and participation. Previous
134 studies focus on the link in less direct fashion. In an attempt to capture the broad-
scale influence, Gray (2004) showed that, among the few indicators he modeled,
proportion Muslim was the only one negatively associated with HIV prevalence for
38 countries in SSA. The vast majority of studies however explore the link between
religious participation and particular "risky" behaviors. For example, several studies
show that affiliation with particular religious traditions delays sexual onset (Agha, et
al. 2006). Garner (2000) for example shows that among four churches in Kwa-Zulu
Natal, South Africa, the Pentecostal church has lower levels of extra- and pre-marital
sexual partnerships than other congregations. Hill and colleagues (2004) find a
similar effect in Brazil for evangelicals, a category which largely overlaps with
Pentecostalism in their sample. Others question whether there is a relevant connection
between religious participation and declines in risky behaviors (Lagarde et al. 2000).
These studies all conceptualize HIV-risk as an individual level property. They
recognize that HIV is contracted through pathogens that only pass through blood and
other bodily fluids. Only particular types of contact can therefore put a person at risk
of HIV infection, and those are the sorts of behaviors these models attempt to explain,
inferring the relationship then to HIV risk. The assumption is that if researchers can
understand the causes of these particular sets of behaviors, then models of
intervention and prevention can be better conceptualized. The implied model
therefore suggests that infection risk is a function of individual risk behaviors, which
can be stated in probabilistic terms as:
Pi(inf) = Ci x SC + f(i, r, c, n) + ei , (6.1)
135 where Pi(inf) is the probability that an individual (i) will get infected as a function of:
(a) Ci, the number of risky behaviors in which i engages; (b) SC, the susceptibility of
infection in a single exposure through contact by a particular risk behavior; and f(i, r,
c, n) represents a combination of any number of other factors, which could vary at the
individual (i), religious organization (r), community (c), or even national (n) levels.
Such models typically attempt to estimate Ci, as "risky" behaviors, for example as the
number of extramarital or extra-relational sexual partnerships of an individual (i).
Other times, they model a proxy for Ci x SC, for example, as the adoption of condom
use within such behaviors, which reduces SC (Gray et al. 2001). Researchers estimate
these models with any number of parameters among their independent variables (the
f(…) portion of Eq. 6.1); and have recently explained a considerable amount of
variation in risky behaviors that is associated with religious participation (Agha,
Hutchinson, and Kusanthan 2006; Trinitapoli and Regnerus 2006), and have even
begun to link these to HIV infection rates (Trintapoli 2006b). Such models however
tell only part of the story, and in effect are based more on the epidemiology of non-
infectious diseases, like cancer, than on infectious diseases, like HIV.
This application of a "cancer model" of the relationship between risk
behaviors and actual risk to the case of HIV and other infectious diseases has been
discussed elsewhere (e.g. Morris 2004), but as yet has not been incorporated into the
literature connecting religion and HIV. What these models fail to incorporate is that
Pi(inf) is a function not only of an individual’s “risky” behaviors, but also of a number
of properties of the alters (j) with whom individuals (i) engage in such behaviors. If
for example an individual has 100 extra-marital partners, none of whom are infected,
136 the risk of infection for that individual, SC, is zero, even though their number of
partners (Ci) is high, something equation (1) is not equipped to incorporate. A better
representation of the probability of infection for an individual would be:
Pi(inf) = • { Cij x SCij | Pj(inf) } + f(i, j, r, c, n) + ei + ej , (6.1a)
where the differences from Eq. 1 are the inclusion of the sum (•) for all of i's alters (j)
of: (a) Cij, the number of an individual's contacts (Ci) with each of i's alters (j); (b) the
inclusion of infection rates which are partner specific (SCij ) and not uniform across
all "risky" behaviors; and (c) the dependence of these infection rates on the given
probability that each alter is infected (| Pj(inf)). For the present paper, I only
incorporate the differences in Eq. 6.1 and 6.1a that occur in the (a) and (c) terms,
although the independent variables included in later models could also incorporate
properties of each alter f(j) as well as properties of each of the other factors separately
2 for i and j (e.g., f(ri, rj)).
Within the network literature, there is considerable debate about the role of
degree (the number of risk-behavior partners) in network structural properties that
drive infection risk. One argument focuses almost exclusively on the role of the
number and distribution of partners to explain risk-network properties in a given
population. Recent popularity of explanations citing a "scale-free" distribution of
partners have championed this focus (Barabasi and Albert 1999; Dezso and Barabasi
2002; Dezso and Barabasi 2001; Liljeros et al. 2003; Liljeros et al. 2001; Newman
2002),3 suggesting that not only are network "hubs" (those with high numbers of
partners) at highest individual risk, but that they also serve an important role in
connecting the entire network, and correspondingly are central to group-level risk
137 estimates. However, recent work by others has demonstrated that high network
connectivity can arise in situations without such hubs, where no individuals have high
numbers of partners (Handcock, Jones, and Morris 2003; Jones and Handcock 2003;
Moody et al. 2003). This latter group of findings suggests that if researchers want to
understand the role of risk-network structure in infection probabilities, they must
include more than the number of partners for individuals in the network.
Based on this literature, there are numerous factors beyond Ci, which could
constrain Cij. This paper focuses on the way religious affiliation contributes to those
constraints, and incorporates a range of estimates of those effects in the simulated
networks described below. First, while previous work has established a link between
religious affiliation and sexual partnering, no studies to date have shown how those
individual-level “risk” behaviors equate to changes in structural properties of risk.
This first focus of the paper will therefore examine how changes in Cij (number of
“risky” behaviors) change estimates of population level infection risk. Second, by
incorporating the known prevalence rates of HIV infection for the population into
these network simulations, I also estimate the effects such changes can have on the
probability that those behaviors take place with an infected partner (Pj(inf)), which is a
difficult property to incorporate into individual models (for one approach, see
Trinitapoli 2006b).
It is well known that epidemiological models based on individual behaviors
alone misestimate STI risk (Aral 2002), and that sexual network structure can explain
differences that are not observed in individual level models. For example, Laumann
(Laumann and Youm 1999; Youm and Laumann 2002) finds that blacks’ sexual
138 partnerships more frequently serve as network bridges than do whites’ partners. They
show that even beyond individual level risk behaviors, the populations where blacks
draw risk partners put them at much greater risk of STI infection than is the case for
whites. Similarly, (Moody and adams 2006) show that in one high risk population
different risky behaviors differentially connect the network. They show that sexual
ties form bridges across otherwise disconnected portions of the population, while
needle sharing partnerships (which have higher transmission rates per contact)
provide largely redundant connectivity, and therefore do not contribute substantially
to the observed networks overall risk potential. To date no research examines how
religious affiliation can influence the partnering patterns of sexual partnerships,
which may similarly drive structural components of STI/HIV risk, particularly in
SSA, where intravenous drug use contributes little to the present HIV epidemic. The
premise of this paper is therefore to combine the observed differences that arise in
number of sexual partnerships by religious affiliation into models of risk network
structure. While these changes in individual behaviors have been the focus of
previous research investigating the link between religion and HIV-risk, little is known
about how they contribute to network properties and therefore the structural
properties of risk. In the simulations that follow, I demonstrate how the declines in
risky behaviors associated with some religious affiliations actually produce little to no
change in estimates of risk when conceptualized in network terms.
Simulation Parameters
The following sections describe the parameters used in estimating network
connectivity for each of three simulated populations. For each simulated population,
139 gender, religious affiliation and HIV status are assigned to match their representation
in combination (i.e., HIV prevalence matches by gender and religious affiliation) in
sample level observations, and held constant through variation of each of the other
simulation parameters. The number of sexual partners for each simulated respondent
is then assigned according to a series of reports on sexual partnership data from the
Malawi Diffusion and Ideational Change Project, Wave-3 (described below). For
each type of sexual partnership reporting, I assign sexual partners in the simulated
populations to approximate the observed distribution of partnerships first randomly,
then constrained by gender, then constrained by gender and religious affiliation. For
each generated distribution, I then simulate 100 networks with the observed
properties. I generated all of the random population characteristic assignments and
network partner matches using the R programming environment (R 2006).
For each produced network, I calculate the size of the component individuals
are members of, whether they are members of components of at least size three, and a
measure of graph reachability. 4 A component of a graph is a subset of the graph
where at least one path connects all nodes, and is the largest possible range for
epidemic spread. Some have suggested components as a minimal measure of potential
epidemic spread (Moody et al. 2003; Moody and White 2003). A component of at
least size three is the minimal requirement for representing a situation in which either
an individual or their partner has at least one other partner. Because component size is
not comparable across graphs, I calculate the proportion of the entire graph reachable
from an individual on at least one path. Finally, for each individual in each graph, I
calculate whether or not the component within which they are embedded includes an
140 HIV-infected individual. I then summarize each of these individual estimates of
structural-risk properties across religious affiliation to approximate how readily
observed changes in individual “risk” behaviors translate into changes in structural
estimates of risk. I calculated each of these network characteristics using the SPAN
packages (Moody 1999) for the SAS program (SAS 2004), and statistical
comparisons using available SAS routines (SAS 2004). These simulations show that
while religious affiliation has been reported to alter number of sexual partnerships,
these changes do not produce corresponding changes in structural components of risk,
unless adopted as abstinence.
Data from the Malawi Diffusion and Ideational Change Project
The networks generated in the simulations here are based on characteristics
observed in data from the Malawi Diffusion and Ideational Change Project – Wave 3
(MDICP-3) (2004). The MDICP is a longitudinal household survey, which in wave 3
surveyed 1542 women and 1081 of their husbands in 119 villages in three rural
districts of Malawi. While the MDICP sample is intended to represent the populations
of the three sampled regions and not necessarily all of Malawi, it does closely
resemble one nationally representative sample (the Malawi Demographic and Health
Survey) (NSO 2005) on several key factors such as age, education, and select
indicators of socio-economic status (Watkins 2004; Watkins et al. 2003). Researchers
have shown favorable evaluation of the reliability, attrition and representativeness of
these data elsewhere (Bignami-Van Assche, Reniers, and Weinreb 2003; Watkins
and Warriner 2003; Watkins et al. 2003). The MDICP data also closely follow
estimates of religious affiliation from other samples where 77 percent of Malawians
141 being Christian, 15 percent Muslim and the remainder practicing traditional African
religions. Presently, the major Christian denominations as a percent of the total
Christian population are Roman Catholics (25 percent), mission Protestants (20
percent), and African Independent Churches or AICs (17 percent); while groups like
evangelicals and Pentecostals are rapidly growing in Malawi, particularly in urban
areas, and together account for about 32 percent of the country’s Christians (Jenkins
2002). While the total MDICP sample closely resembles national rates of religious
affiliation, the northern sample district is a majority Mission Protestant, while the
southern sample district is predominantly Muslim. The effects of these differences
will be discussed in further detail below.
[Insert Table 6.1 about here]
In this paper, the simulations I generate are designed to represent the
properties separately for each of the MDICP's three research locations with regard to
population size, gender composition, marital composition, religious affiliation and
HIV prevalence (constrained by each of these other parameters). To introduce some
of the differences that exist in the sample districts, Table 6.1 displays the HIV
prevalence by religious affiliation figures for MDICP3 summarized for the entire
MDICP3 sample, and sorted by sample district. This table demonstrates a few
properties of religious affiliation and HIV prevalence that motivate the current paper.
For MDICP3, respondents were asked to which religious group they belonged, and
were provided with 14-groupings from which to choose.5 For analytic purposes, those
14 categories are collapsed to seven: none, Roman Catholic, Mission Protestant
142 (includes CCAP, Baptist and Anglican), Pentecostal, Muslim and other (includes
Jehovah's Witness, Seventh Day Adventist, "Just Christian" and other).
Sample-wide there are few significant differences in the HIV prevalence by
religious affiliation (with the exception that members of AIC churches have
consistently higher infection rates than other groups). The differences across sites
however, are substantial (with Balaka having the highest infection rates, and Rumphi
the lowest among the three sites). Further, there are substantial differences by
religious affiliation within district, with a moderately protective effect of being a
member of the district's dominant religious group (e.g., Muslims in the south and
Roman Catholics or Mission Protestants in the center, though the pattern does not
hold for AICs in the northern region), while being a member of minority groups is
moderately associated with increased risk. While it is difficult to speculate the precise
properties that give rise to such discrepancies, present research suggests that the
differences in intervention strategies and behavioral adaptations across
denominational classifications do not appear substantial enough to explain such
variance (see chapter 5; also Trinitapoli 2006b). This paper therefore investigates the
presumed pathway in existing literature that suggests number of “risk” behaviors as a
significant contributor to this difference. I postulate that these changes observed by
religious affiliation in risk behaviors will not coincide with the expected
corresponding changes in structural properties of risk. This expectation is consistent
with knowledge from existing network epidemiology literature, and stands in contrast
to the assumptions in the present religion and public health literatures, the latter of
which under-gird present intervention models. If found to be accurate, this would
143 force some rethinking of the way researchers and public health officials currently
think about intervention strategies.
Individual / Behavioral Constraints
Fixed Parameters
For the simulations in this paper, I draw on several sources of known
information regarding sexual partnerships and religious participation from MDICP
data. These parameters are based on findings that are described more fully in work
elsewhere (e.g., Trinitapoli 2006b; Trinitapoli and Regnerus 2006; Watkins et al.
2003). I simulate three populations, which approximate the total population for all
persons in the sampled villages included in MDICP-3 separately for each of the three
sample districts (Balaka in the south, Mchinji in the center and Rumphi in the north).
Each individual within a site is then assigned a gender, marital status,6 and collapsed
religious affiliation – each according to proportional assignments that approximate
the distribution of each of these parameters for MDICP-3. HIV status is then assigned
to match the distribution across of the other parameters within each site.
Variable Sexual Partner Parameter
I then vary the number of sexual partners for each individual in the simulated
population according to a series of estimates of sexual partnership data also drawn
from MDICP-III data. The first series of estimates incorporate all sexual partners an
individual reports having in the past 12 months, as reported in the sexual network
partnership section of MDICP-III. Each run of the simulated networks first assigns
144 the sexual partnership information randomly; subsequent runs constrain the
assignment of partners according to observed information by gender then additionally
by religious affiliation. Because the wording of the question about extramarital
partnerships limited the respondent to list no more than three sexual partners
(including spouses), and only those in the past 12 months, I incorporate a series of
additional sexual partnership estimates that derive from other sources in MDICP-III
data. To account for the potential deflating of the number of sexual partners this
constraint introduces, I also add an additional partner to individuals based on
respondents who reports having an additional sexual partner at any point during their
current marriage, for those respondents who did not report this additional partner as
someone with whom they had a relationship with in the past 12 months. While this
potentially expands the estimated network properties beyond present concurrency of
partnerships, I incorporate it as a means to estimate the underreporting of extramarital
partnerships observed for self-reported sexual relationships.7
The under-reporting of sexual behaviors for individuals has notoriously
plagued survey research in SSA (Nnko et al. 2004). The final estimate of number of
extra marital partners per simulated individual comes from data about people with
whom MDICP-III respondents discussed AIDS-related information. For each of these
discussion partners, respondents were asked to report the number of sexual partners
their discussion partners had in the past year.
[Insert Tables 6.2 and 6.3 about here]
Each of these reports of sexual partners are then randomly assigned as
estimates for number of sexual partners for the individuals in the simulated networks,
145 where I assign number of sexual partners to each individual as one (all included nodes
have at least one partner – those with no partners have been excluded in the
assignment of marital status) plus two random Bernoulli draws with p defined to
match the above estimates of extramarital sexual partnerships as observed in self
reports of sexual partnerships and the reported sexual partnerships for all AIDS-
discussion partners.8 I produce a series of partnership distributions to arrive at
matching average degree values, conditional on the parameters defined in Tables 6.2
and 6.3.
Mixing Model Parameters
After the number of sexual partners is defined, I run a heterosexual mixing
model, which randomly matches each simulated respondent to the assigned number of
partners.9 I produce 100 simulated networks for each of the sets of partnership
parameters, and describe the mean and standard errors of these groups of networks for
the network properties described in the results below. I also display a series of
exemplar networks here, and highlight those networks in the figures describing the
observed network properties.
Findings
The overall pattern that emerges in the simulated networks, for the majority of
the parameter sets is one of numerous small components ranging in size from 3 to
approximately 100. In all of the simulated networks, no more than 20 percent (in one
parameter set it is less than ten percent) of the population is found in monogamous
relationships despite approximately 49 percent of the population having only one
partner.
146 [Figure 6.1 about here]
Figure 6.1 shows the dominant pattern in Rumphi, where overall number of
partners is substantially lower than each of the other two sites. In this simulated
population, the vast majority (~85%) of simulated respondents are found in
components ranging from 3-77. The focus in this figure however should not be on the
overall size of the components that emerge but the distribution by religious affiliation
of individuals within those components. Remember (see Table 6.2) that in Rumphi,
Mission Protestants account for approximately 30 percent, while Pentecostals account
for about twelve percent of the population. These two groups, despite having
significantly different numbers of partners appear in Figure 6.1 to remain distributed
across the various components (and those of smaller sizes not shown in this image) in
proportions approximating their representation in the population. Pentecostals having
fewer partners on average than Mission Protestants does not protect them from being
located in such large components (see the blue and red nodes respectively).
[Figure 6.2 about here]
Figure 6.2 shows that a similar pattern emerges in the example of the
simulated population for the Mchinji population, using AIDS-discussion partners’
reports of sexual partnership data. This image shows that while more of the
population is located in these larger components than was the case in the Rumphi
example, the distribution by religious affiliation of individuals across those
components continues to reflect their overall distribution in the population, and is not
differentially associated with increases in number of partners. These two example
147 graphs demonstrate the pattern in five of the six main parameter settings where
number of partners is constrained by site, gender and religious affiliation.
[Figure 6.3 about here]
In one of the combinations of parameter settings (for the simulated Balaka
population based on AIDS-discussion partner reports of sexual partnership data), a
giant component emerges that includes about one third of the simulated population,
an example graph of which can be seen in Figure 6.3. Despite this dramatic difference
in the overall structure of the simulated networks, individuals remain distributed
within those networks in the same manner consistent with their representation in the
population as a whole, and not dependent on their number of partners. In fact, if
instead of coloring nodes by religious affiliation, I color them by number of partners,
you can begin to see the problem with using number of partners as a primary
predictor of risk. Even those individuals who have only one partner (the yellow nodes
in Figure 6.4) are found within components of varying sizes (including the giant
component), in no relation to their number of partners. There are actually more
people with only one partner found in this giant component (39%) than are found in
components of size 2, representing monogamous relationships (22%).
[Figure 6.4 about here]
Table 6.4 presents summarized network properties for the simulated networks
representing the number of self-reported sexual partners – constrained by site, gender
and religious affiliation. The first column demonstrates the differences in number of
sexual partners by religious affiliation. The second column indicates the proportion of
148 individuals who are in components of size three or greater. By religious affiliation,
the remaining columns present: the third column - the average size of the component
individuals are embedded in; the fourth column – the proportion of the full graph
reachable by an individual; and the fifth column – the proportion of individuals found
in a component that includes at least one individual who is HIV positive.
[Table 6.4 about here]
Significance tests, which are measured in comparison to values for Mission
Protestants indicate that those significant differences in number of sexual partners
that exist by religious affiliation do not regularly correspond to differences in network
position in the predicted direction. For example, in the bottom panel of Table 6.4,
Pentecostals and Others each have fewer average partners than Mission Protestants
but are not significantly different from them on any of the network measures of
estimated risk. In fact, there are almost as many significant changes in the opposite of
the predicted direction as in the expected direction. For example, in Panel 1 of Table
6.4, Pentecostals do have significantly fewer sexual partners than Mission Protestants,
and – as predicted by the assumptions in present research – measures of component
size, reachability and membership in a component with and HIV positive individual
are marginally lower than for mission protestants (p<0.10). However, in the same
panel members of AIC churches, who have marginally more sexual partners than
Mission Protestants are actually less likely to be found in components with an HIV
positive individual (p<0.05).
[Table 6.5 about here]
149 Table 6.5 presents the same set of summarized network properties as in Table
6.4, but for the simulated networks representing the number of sexual partners
reported for AIDS-discussion partners – constrained by site, gender and religious
affiliation. In this table, the overall number of partners, and measures of network
connectivity are higher on average than in the models based on self-reported sexual
partnership data. However, the general pattern from above remains, in that of
differences in number of sexual partners that vary by religious affiliation do not result
in corresponding differences in network estimates of risk. As is the case in Table 6.3,
there are almost as many estimates of network risk that work in the opposite of the
predicted direction as there are that work as present assumptions would expect.
Turning then to intervention strategies, the results presented here suggest that
existing strategies of HIV-prevention that focus almost exclusively on promoting
behavior changes (such as the predominant “ABC” model in SSA, which promotes
Abstinence, Be faithful and when those do not work, Condom use) must be rethought.
Intervention models cannot focus only on “high-degree” actors, as risk substantially
impacts individuals who have smaller numbers of partners, even those who have only
one sexual partner, and therefore are presumed by existing models to not engage in
any risky behaviors. Further, models that demonstrate the association of religious
affiliation with behavior changes are potentially limited in their actual contribution to
understanding their role in altering HIV risk. One suggestion for an intervention
strategy that moves beyond thinking of HIV risk as an individual level property is
increasing the emphasis for VCT (voluntary counseling and testing), which has
recently become increasingly accepted. Another strategy that relies on religious
150 organizations, and is consistent with existing literature is that some of the policing of
sexual behaviors and evaluation of potential partners by religious and community
leaders (see for example Watkins 2004; Trinitapoli 2006b) is to remove some of the
privatizing of HIV-risk models that present prevention strategies produce. In other
words, while present models focus on individual behavior change, inferring HIV-risk
as an individual level property, instead prevention efforts might in fact want to
encourage highly-afflicted regions to think of HIV-risk as a community level issue to
be engaged by the community rather than left exclusively to individuals.
Conclusions
As is known in network epidemiologic models, the simulations presented here
demonstrate that the average number of partners is a poor predictor of network
position, and of proximity to an HIV infected individual. This finding stands in
contrast to the assumptions of the existing religion models, which frequently rely on
the number of partners as a primary estimate of an individual’s HIV risk. The
findings here suggest that if researchers are really interested in understanding the
importance of religious affiliation or participation on HIV risk, we must investigate
how it impacts position in the network, beyond simply ego-properties. Future
research should investigate how religious affiliation impacts network structure and
not simply the number of sexual partners of an individual, to more accurately capture
its effect on HIV risk.
151 NOTES
1 This chapter was previously presented at the Annual Meetings of the Population Association of America (adams 2007).
2 Modeling pairwise infection probabilities (SCij).
3 Scale-free networks are those that, through growth and preferential attachment, have a probability, P(k) that a vertex in a network will connect to k other vertices that decays as a power-law, following P(k) ~ k-λ (Barabasi and Albert 1999).
4 A component of size three is the smallest group that reflects individuals who are, or who are tied to someone who is, involved in non-monogamous relationships. Four nodes as the smallest possible bicomponent here represents both the data’s inclusion only of reports of heterosexual partnerships, and the property of bicompnents, which requires that they have at least four nodes to include 2 node-independent paths.
5 The list of religious groups on the survey included: none, Roman Catholic, Quadriya Muslim, Sukuti Muslim, CCAP, Baptist, Anglican, Pentecostal, Seventh Day Adventist, Johovah's Witness, Indigenous Christian, Indigenous non-Christian, "Just Christian," and Other.
6 Since my interest here is not in marriage per se, but in the sexual partnerships in marriage, I assign “marriages” according to the rates of relational reporting in MDICP-3 data. As such the only people from MDICP-3 data who would be excluded from the models below are those reported as having no sexual partners in the last year. All other individuals are assigned as being “married” in the sense that they are assigned at least one sexual partner.
7 I also calculated estimates excluding this second addition to the self-reported extramarital partners, but they did not differ substantially from the models presented.
8 Unfortunately, the religious affiliation variable available for AIDS-discussion partners does not correspond exactly to the 7-category variable described above in self-reported religious affiliation for MDICP respondents (None, Roman Catholic, Mission Protestant, Pentecostal, AIC, Muslim, Other). The AIDS-discussion partner's religious identification is coded using a scheme that was adopted directly from MDICP-2 for purposes of comparison, and includes as the categories: No religion, Catholic, Protestant, Revivalist, Traditional African, Moslem, and Other). The overlap between these two variables however is considerable. I estimate this overlap by calculating a Chi-square difference test, for the classification for self reported affiliation and the reported religious affiliation for all of the respondent's best friends who are included among AIDS-discussion partners and attend the same church as the respondent. I find no significant differences between, what are in effect, two separate classifications of the same congregation. The only potential difference of note is that the AIDS-discussion partner coding includes non-Christian Indigenous churches among the Traditional African churches, and the self-reported data will include them among the Others. Unfortunately there is no way to correct this difference. Pentecostal and Revivalist appear to be functionally interchangeable, and each
152
of the other categorizations correspond across the two coding schemes. I will therefore refer only to the self-reported classification scheme, while substituting the data from the AIDS-discussion partner classifications directly into the estimates of sexual partnerships by religious affiliation.
9 Because of the strong constraints on how the number of partners is assigned (matches on site, gender, marital status and religious affiliation), there are small differences in the number of partners for men and women. In each simulated population, the gender with the smaller number of partners is used to limit the pairing of partners in the mixing assignments.
153
Balaka Mchinji Rumphi N HIV+ N HIV+ N HIV+ Catholic 110 13.89 156 7.48 90 5.62 Mission 64 11.48 155 7.69 206 6.09 Protestant Pentecostal 45 15.56 43 15 74 2.74 AIC 16 18.75 14210.29 179 8.43 Muslim 526 8.82 3 0 6 0 Other 34 8.82 102 5 87 4.76
NOTE: Numbers presented are N of the population in the particular site and percentage of that population who were HIV+ according to 2004 testing.
Table 6.1 HIV Status by Religious Affiliation MDICP-3
154 Total Women Men Balaka Mchinji Rumphi Total Balaka Mchinji Rumphi Total Balaka Mchinji Rumphi Total
Population 3648 3424 3043 10115 1856 1742 1548 5146 1792 1682 1495 4969 Estimate
Marital Status Married (N) 15-19 3.10 5.68 0.42 20-24 12.05 16.43 7.52 25+ 49.74 44.77 54.89 Single (N) 15-19 17.51 13.38 21.78 20-24 7.73 5.45 10.10 25+ 9.87 14.29 5.30 158 Religious Affiliation none 0.22 1.99 0.92 1.03 0.00 1.22 0.45 0.54 0.54 0.31 1.61 1.73 RC 13.08 27.05 13.04 17.80 13.68 25.15 12.03 16.99 12.20 29.64 14.52 18.94 Mission Protestant 7.80 23.07 30.96 19.94 7.21 24.34 36.97 21.92 8.67 21.33 22.26 17.12 Pentecostal 5.16 7.73 12.65 8.28 5.36 9.53 13.14 9.10 4.88 5.26 11.94 7.12 Muslim 67.47 0.82 1.19 24.97 68.58 0.41 0.22 16.32 65.85 1.39 2.58 24.62 AIC 2.09 22.83 27.93 16.88 1.66 23.53 26.06 25.22 2.71 21.88 30.65 17.69 Other 4.18 16.51 13.31 11.10 3.51 15.82 11.14 9.91 5.15 17.45 16.45 12.79
HIV Prevalence men 9.40 7.60 5.38 7.55 women 11.23 8.82 6.79 9.05 combined 10.50 8.32 6.23 8.45 NOTE: All numbers presented are column percentages.
Table 6.2a. Static Simulation Parameters
155
Total Women Men Balaka Mchinji Rumphi Total Balaka Mchinji Rumphi Total Balaka Mchinji Rumphi Total Self Report Overall 1.243 1.166 1.122 1.181 1.108 1.194 1.093 1.135 1.310 1.238 1.161 1.241
by 6cat relig none 1.000 1.133 1.143 1.125 N/A 1.000 1.500 1.125 1.000 1.222 1.000 1.125 RC 1.183 1.126 1.152 1.147 1.182 1.075 1.116 1.116 1.186 1.183 1.194 1.186 Mission Protestant 1.197 1.211 1.128 1.171 1.027 1.161 1.128 1.125 1.414 1.279 1.129 1.245 Pentecostal 1.000 1.205 1.080 1.095 1.000 1.172 1.019 1.057 1.000 1.267 1.176 1.154 Muslim 1.282 1.000 1.250 1.278 1.226 1.000 1.000 1.224 1.359 1.000 1.286 1.349 AIC 1.467 1.192 1.114 1.164 1.571 1.084 1.095 1.107 1.375 1.333 1.135 1.231 Other 1.143 1.120 1.121 1.157 1.158 1.090 1.068 1.092 1.125 1.155 1.170 1.157
159 For AIDS Discussion Partners Overall 1.535 1.272 1.191 1.325 1.465 1.177 1.099 1.232 1.610 1.354 1.331 1.430 by 6cat relig none 1.000 1.443 1.769 1.519 1.000 1.222 1.200 1.267 1.100 1.481 1.125 1.581 RC 1.636 1.266 1.208 1.349 1.548 1.188 1.095 1.259 1.736 1.332 1.370 1.444 Mission Protestant 1.543 1.207 1.158 1.210 1.256 1.132 1.063 1.096 1.742 1.269 1.315 1.347 Pentecostal 1.387 1.271 1.119 1.220 1.319 1.127 1.096 1.152 1.500 1.365 1.152 1.297 Muslim 1.547 1.250 1.579 1.542 1.509 1.000 1.000 1.499 1.587 1.400 1.647 1.585 AIC 1.556 1.303 1.194 1.265 1.682 1.087 1.080 1.135 1.357 1.542 1.343 1.426 Other 1.244 1.368 1.152 1.241 1.316 1.241 1.150 1.197 1.182 1.483 1.156 1.295 NOTE: All numbers are mean number of reported sexual partners.
Table 6.3. Simulation Parameters – Sexual Partnerships
156
Degree Comp>3 Comp Size Reach HIV+ N Balaka
Religious Affiliation None 1.000 0.672** 11.759** 0.001** 0.037** 10 RC 1.226 0.854 18.128 0.005 0.569 456 MP 1.280 0.853 18.304 0.005 0.589 279 Pent 1.106 * 0.837 16.999+ 0.005+ 0.555+ 170 Muslim 1.268 0.858 18.250 0.005 0.576 2495 AIC 1.494 + 0.857 18.348 0.005 0.535* 77 Other 1.118 + 0.844 17.395 0.002 0.574 161
Mchinji
Religious Affiliation None 0.988 * 0.841** 11.441* 0.003* 0.508 86 RC 1.156 0.839 10.838 0.003 0.468 918 MP 1.204 0.839 10.926 0.003 0.468 790 Pent 1.285 0.834 10.697 0.001 0.438 239 Muslim 1 0.865** 11.638* 0.003* 0.444** 32 AIC 1.221 0.838 10.946 0.003 0.464 766 Other 1.078 ** 0.847 10.991 0.003 0.454* 593
Rumphi
Religious Affiliation None 1.458 + 0.787 10.209+ 0.003+ 0.466 24 RC 1.167 0.844 11.298 0.003 0.461+ 401 MP 1.165 0.838 11.460 0.003 0.469 877 Pent 1.050 * 0.840 11.681 0.003 0.447 397 Muslim 1.444 + 0.760 9.803 0.003 0.366 36 AIC 1.113 0.841 11.420 0.003 0.463 864 Other 1.061 * 0.836 11.168 0.003 0.442 444
Table 6.4. Network Properties from Self-Reported Sexual Partnership Data
157
Degree Comp>3 Comp Size Reach HIV+ N Balaka
Religious Affiliation None 1.000 + 0.770** 816.92** 0.224** 0.581* 10 RC 1.691 0.919 1208.11 0.331 0.785 456 MP 1.573 0.915 1179.85 0.323 0.776 279 Pent 1.541 0.912 1180.67 0.324 0.778 170 Muslim 1.520 0.909 1163.88 0.319 0.770 2495 AIC 1.519 0.889+ 1107.39+ 0.304+ 0.726* 77 Other 1.211 ** 0.889* 1074.89* 0.295* 0.743* 161
Mchinji
Religious Affiliation None 1.186 0.808* 14.664 0.004 0.516 86 RC 1.291 * 0.851 15.710** 0.005** 0.499 918 MP 1.184 0.859 15.999 0.005 0.513 790 Pent 1.293 0.847 15.813 0.005 0.480+ 239 Muslim 1.281 0.826 15.269 0.004 0.448 32 AIC 1.322 ** 0.866 16.090+ 0.005+ 0.504 766 Other 1.310 * 0.851 15.677 0.005 0.485 593
Rumphi
Religious Affiliation None 1.500 0.850 28.007 0.008 0.581 24 RC 1.252 0.867 28.478 0.008 0.556 401 MP 1.195 0.867 28.519 0.008 0.559 877 Pent 1.081 * 0.870 28.501 0.008 0.545 397 Muslim 1.778 ** 0.885 32.344* 0.009* 0.540 36 AIC 1.218 0.876 28.798 0.008 0.560 864 Other 1.088 * 0.874 29.052* 0.008* 0.549 444
Table 6.5. Network Properties from AIDS-Discussion-Partners-Reported Sexual Partnership Data
158
Legend None RC MP
Pent
162 Mus AIC
Other
Pajek NOTE: Includes all components of size 20 or greater. Nodes are colored by religious affiliation.
Figure 6.1. Rumphi Simulated Population Self-Reported Sexual Partnership Data - Exemplar Network
159
163
45
Pajek
NOTE: Includes all components of size 20 or greater. Nodes are colored by religious affiliation; see Figure 1 for the key.
Figure 6.2. Mchinji Simulated Population AIDS-Discussion-Partner-Reported Sexual Partnership Data - Exemplar Network
160
164
Pajek
NOTE: Includes all components of size 20 or greater. Nodes are colored by religious affiliation; see Figure 1 for the key.
Figure 6.3. Balaka Simulated Population AIDS-Discussion-Partner-Reported Sexual Partnership Data - Exemplar Network
161
165
Legend 1 2 3
4 5 APajek NOTE: Includes all components of size 20 or greater. Nodes are colored by number of sexual partners.
Figure 6.4. Balaka Simulated Population AIDS-Discussion-Partner-Reported Sexual Partnership Data - Exemplar Network
162
CHAPTER 7
CONCLUSIONS AND SUGGESTED EXTENSIONS
Given that this dissertation is titled “Religion Networks and HIV/AIDS in
Rural Malawi,” it is not surprising that the findings presented here demonstrate that networks matter. What may be surprising however, is the way that knowledge of networks can move research beyond many of the questions in existing literature, and alter both the findings and the implications of research that is more traditionally focused on actors (whether individuals or organizations). To conclude, in this chapter
I give an overview summary of the findings presented in previous chapters, suggest some additional directions that applying a network perspective can further help research of the religion-HIV link in SSA, and finally develop a few suggestions for how the implications of what I present here could inform future policy interventions aimed at combating the HIV-epidemic.
Summary Overview
In this dissertation I examined the ways that taking a relational-structural approach changes how we study the link between religion and HIV in rural Malawi.
In particular, I applied perspectives available from SNA to two separate levels of analyses. The first section examined the existence and salience of the hierarchical 163 organization of religious denominations in rural Malawi, which present intervention models presume direct the flow of information used in prevention messages. I find that the presumed hierocratic structure is an inefficient route for disseminating prevention strategies. Local pastors and sheikhs have little contact with the national leaders of their organizations (see chapter 4), and the prevention strategies they espouse more commonly reflect the attitudes of their local community than of the formal organizations with which their organizations are affiliated (see chapter 5).
The second section of this dissertation demonstrates one mechanism by which reductions in individual-level risk behaviors can produce little change in individuals’ actual HIV-risk. An individual’s number of partners, which religious organizations do influence, does not predict to whom those individuals will be connected, or whether those partners (or their partner’s partners, and so on) are already infected (see chapter
6). At present we know little about how religious organizations influence these network-properties of risk.
Other Network Extensions
One of the primary aims of this dissertation has been to establish a link between the vast amount of work that has been done from a social network perspective and the sociology of religion and health literatures. To date, the primary audience for social network scholarship has been other social network scholars. While those scholars are largely convinced of the importance of networks for studying social phenomena, this “ghettoization” of the perspective has limited its ability to inform the ways that social scientists study the variety of topics in which they are interested –
164 despite offering substantial alterations to the way social scientists approach their research topics.
Future investigations of the HIV epidemic both within the sociology of religion and those engaged in other sub-disciplines could similarly benefit from taking a relational-structural approach to the way scholars approach their research.
For example, further investigating how religious association (both affiliation differences, and intensity of participation) differentially contributes to where individuals draw their sexual partners is a ready extension to the work in Chapter 6.
Further, while many scholars build on the importance of network “hubs” – whether in explicitly network terms, or indirectly through discussions of the role of prostitutes or other presumed high-risk actors – in their models, investigating how religious associations are differentially associated with gender-discrepancies in network positions could help provide explanations of how women are at higher risk of HIV- contraction, despite consistently reporting lower numbers of sexual partners.
Outside of the sociology of religion, a burgeoning literature focuses on the role of marriage in increasing HIV-risk. Marriage is a fundamentally relational phenomenon, though it is frequently modeled as an individual characteristic much as any other held status. If instead we apply a relational frame to the way researchers examine marital contributions to the HIV-epidemic, we may be able to better understand how marriage itself is a substantial contributor to the HIV-epidemic.
Many other possibilities exist, and scholars in a wide variety of scholarly endeavors could similarly benefit from the approach elaborated here. By moving beyond thinking of networks as an alternate way to investigate the same questions
165 possible in other research strategies, scholars in other sociological sub-disciplines may also take advantage of the fundamentally different perspective on constructing the questions they ask through taking an SNA approach to their research. Whether this shift similarly alters the findings and implications of the prevailing knowledge for other substantive questions – in the same way that it does here – is something that can only be determined once researchers have applied this alternative research approach to the topics they study.
Policy Extensions
Each of the sections described above have direct implications for the way that existing prevention strategies can more readily evince desired reductions in new HIV cases. The first section primarily reorients how policies are implemented, while the second forces some additions to the predominant ABC paradigm, which drives most of what existing prevention efforts suggest.
As for the development of prevention strategies through religious organizations - denominations in Malawi do not appear to be self-organized in a sufficiently hierocratic way to funnel prevention resources through them – whether information or finances. While virtually all denominations in Malawi have formulated a response to the epidemic, these strategies have rarely impacted the efforts taken at the local level.1 Any future efforts that continue to funnel resources initially through the centralized headquarters of these organizations would ignore previous failures of this strategy. Organizations disseminating funding or other resources through religious channels may want to engage local contexts more directly, particularly by training – or seeking the cooperation of – local religious leaders themselves, rather
166 than through denominational proxies. If such direct contacts are not reasonably available, international donors and national NGOs should – at minimum – more vigorously pursue feedback on their existing efforts from multiple levels within the participating organizations, and not simply from those at the top of religious denominations.
Despite increasing recognition that supra-individual properties have substantial influence both on individuals’ HIV risk and the potential for epidemic spread, prevention strategies continue to focus almost exclusively on individual behavior changes. While existing network literature incorporates the fact that who partners are can be as influential in shaping an individual’s risk, as how many partners they have, prevention strategies rarely attempt to incorporate such alter characteristics in their efforts. While religious organizations do appear to wield some influence over the behaviors of their participants, they also occupy a unique position to potentially influence one major arena governing with whom those behaviors take place. Recent evidence suggests that discordant-status mixing between marital partners may be substantially contributing to spreading the present epidemic. Marriages, while not an exclusively religious domain in rural Malawi, almost always involves an ankhoswe, many of whom are from the congregation of those who intend to marry. One strategy that some religious leaders have recently adopted is to encourage HIV testing before marriage, so that potential marriage partners can be aware of their potential partner’s present HIV status. This should be a higher prevention priority, and is one that religious leaders may be uniquely positioned to encourage. Entering into marriage with both parties HIV-negative provides a context within which fidelity – the “B” within the ABCs – can actually protect an individual from contracting the disease. 167 168 NOTES
1 Again, this is not unique to religious organizations. Previous research has demonstrated similar activity in national governments generating no impact on local level processes (Guari 2006; USAID 2003).
169
APPENDIX A
CONGREGATION LEADER – QUALITATIVE TRAINING GUIDE
170 MRP QUALITATIVE TRAINING
Interviewers should understand the 4 key topics that we are most interested in knowing more about. The goal of these interviews is to engage in a
“conversation” with the leader about their religious life, their role as religious leader, and the problems facing them and their congregation. This is not a guide for the interviewer to follow, but gives some suggestions and examples of what we are interested in for each of the four main topics.
1. PERSONAL RELIGIOUS HISTORY of the congregation leader –
DENOMINATIONAL AND SPIRITUAL – from birth to just before he
became the leader of this congregation. This needs to be done in as much
detail as possible . What about personal spiritual growth, periods of doubt,
how were they resolved, etc.? Stories about personal experiences with healing,
their exposure to or interest in other religious traditions or denominations.
Their own decisions to examine or join other religions or to resist any
possibilities to change and WHY. Probe about the tensions and difficulties
that come with leaving a congregation and joining a new one. We want to get
the stories about their religious lives and the changes they’ve been through.
2. TIME AS LEADER AS CONGREGATION. How did you become leader?
What is satisfying about your work in this congregation? We want to know
about the problems in the congregation as well. How do you address such
problems when they come up? We’re interested in the leader’s role as
PASTORAL COUNSELOR. What problems do people come to you with? 171 What types of difficulties do people ask you about, and what kind of
responses do you give? SUSPENSION from church and why. Do people ever
get kicked out? Why? Is it hard for you to do this? Do you give people second
and third chances? How does this happen? SERVICES – what sort of
activities does the congregation engage in? Do you or other members from
your congregation participate in fellowships or revivals? What are they like?
COMPETITION WITH OTHER DENOMINATIONS- what brings new
members? What makes people leave? Probe for stories about who left, who
came and why. In these stories, do you see that members are being “snatched”
by other denominations? Which traditions are “snatching”? How do these
congregations attract others? When there are problems, where do you go for
help with these? Who does this leader consult with when things get difficult—
friends? Other pastors? How does he cope with these things? What about
relationships with denominational authorities or other outside groups? To
what extent these provide support, leadership, funding? Are they helpful?
Stingy? Supportive? Is money given with strings attached? How are other
decisions in the church influenced by these other relationships?
3. BIGGEST PROBLEM FACING YOUR CONGREGATION? Your religious
tradition in general? Here, we want to focus on the problems the leaders bring
up and not move too quickly to AIDS. Religious leaders have told us that their
congregations face many different problems. They may tell us that their
members are uneducated, that they are lacking food, that the building is falling
apart, or that their congregation is shrinking; they also may bring up AIDS 172 here. SIN – what does the leader believe about the relationship between sin
and the problems facing the congregation? We’re interested both in this
particular congregation and in “the Church” (or Islam) more generally.
4. SURVEY ADMINISTRATION
5. This section will be asked if sections 2 & 3 have not already lead to an in-
depth discussion about AIDS
VIEW OF AIDS. When did he first hear about it? What did he think about it
then? What did he say about it when it first came up? With his friends? With
his wife? Have his views about AIDS changed at all in more recent years?
Views on sin (individual sin or community/collective sin). Is there
disagreement in your congregation about issues having to do with AIDS? Do
some people disagree about how to care for the sick? Whether or not to sing at
the funeral when someone has died from AIDS? Whether or not people who
are sick with AIDS should be allowed in the church? What about counseling?
Do people come to talk with you about AIDS? About AIDS-related problems
like caring for orphans? Do you teach about AIDS in your congregation?
Have you ever gone to a workshop for AIDS training? Find out EXACTLY
what happened at the AIDS training.
173
APPENDIX B
NETWORK SECTION OF CONGREGATION LEADER SURVEY
174 Congregational Leader Questionnaire, English Final Version 2005 November 11, 2005
INSTRUCTION: The following questionnaire should be administered to a leader of every group mentioned in the Village-Specific Religious Census. Where possible, and where such can be identified, the senior leader of that congregation should be interviewed.
FILL D1 – D4b PRIOR TO THE INTERVIEW D1. Village name ______D2. Village number |__|__|__| D3. Name of religious leader ______D4a. Name of Church/Mosque ______D4b. Village-specific church code [Final column, religious census]: |__|__|__|
D5. Interviewer Name ______Interviewer # |__|__|__| D6. Supervisor ______D7. Interview Date: Month [ __|__ ] Day:[ __|__] Year |__|__|__|__| D8. TIME BEGUN [ __|__ ][ __|__] (24 HOUR TIME)
FOR LOGGING PURPOSES D10. Transcriber name ______Transcriber # |__|__|__| D11. Typist name ______Typist # |__|__|__| D12. Interviewer Checked ______Date: ______
INFORMED CONSENT INTERVIEWER: READ TO THE RESPONDENT BEFORE THE BEGINNING OF THE INTERVIEW
My name is ______. I am working with a research team from the Malawi College of Medicine and several foreign universities. We are interested in religious life in this area and understand that you are one of the local religious leaders. If you don’t mind, we’d like to ask you a few questions about your church/mosque, your role as a religious leader in this community, and your community in general. All your answers will be treated with the utmost confidentiality. Is it all right if I ask you these questions?”
YES [__] NO [__]
175 Thanks for agreeing to talk with me about your church / mosque. I’d like to start by asking you about your congregation in general and also about your own religious background.
N1 IN THE PAST YEAR HAVE YOU MET WITH ANY: A. Clergy from your same denomination Yes…...... 1 No……………….0 Don’t Know….…88 B. Authorities from your same denomination Yes…...... 1 No……………….0 Don’t Know….…88 C. Clergy from other denominations Yes…...... 1 No……………….0 Don’t Know….…88 D. Authorities from other denominations Yes…...... 1 No……………….0 Don’t Know….…88 E. Missionaries (or Brothers) Yes…...... 1 No………….0 ! N1F Don’t Know..88 E2. From where?______F. NGO leaders Yes…...... 1 No…………...0 ! N1G Don’t Know..88 F2. From where?______G. Government leaders Yes…...... 1 No……………….0 Don’t Know….…88
N2 Has you congregation ever been visited by Yes…………….……....1 denominational authorities? No……………..………0 Don’t Know…………..88 N3 Has your congregation ever been visited by high- Yes…………….……....1 level government officials? No……………..………0 Don’t Know…………..88 N4 Has your congregation ever been visited by Yes…………….……....1 missionaries? No……………..………0 Don’t Know…………..88 INTERVIEWER: PROBE AS IN N1E (FOREIGN BROTHERHOOD) N5 Has your congregation been directly helped by Yes…………….……....1 any NGO programs? No……………..………0 Don’t Know…………..88
176 N6 Has your congregation been directly helped by Yes…………….……....1 any mission work? No……………..………0 Don’t Know…………..88 INTERVIEWER: PROBE AS IN N1E N7 Are there any congregations with which you Yes………………1 have participated in services or programs during No…………….…0 !N10 the past 12 month? DK……………..88
b. LIST DENOMINATION c. Freq Frequency N8 What were a. LIST NAMES Codes their names? i. ______i. ______i.____ last ii.______ii.______ii.____ week…...1 iii. ______iii.______iii.____ last iv.______iv.______iv.____ month.....2 2-5 mos v.______v.______v._____ ago.….3 vi.______vi.______vi.____ 6 mo. – 1 vii.______vii.______vii.____ yr.ago….4
N9 CHECKER QUESTION: Count number mentioned in N8a Number______
N10 Have you discussed issues of religious belief or Yes………...... 1 church doctrine with anyone outside of your No………………..0 ! N15 ! N15 own church or mosque, such as…? Don’t Know……88 LIST NAME, POSITION, & CHECKER N11 Leaders of other churches? Total # AFFILIATION Named a.______INTERVIEWER: PROBE b.______FOR NAME, POSITION, c.______AND AFFILIATION. d.______(EXAMPLE: “Joel Phiri, e.______Pastor, Ulongwe CCAP” or f.______just “Pastor, Ulongwe CCAP”) g.______
N12 Leaders of other denominations? a.______b.______INTERVIEWER: PROBE c.______FOR NAME, POSITION, d.______AND AFFILIATION. e.______(EXAMPLE: “Joel Phiri, f.______Pastor, Ulongwe CCAP” or g.______just “Pastor, Ulongwe CCAP”)
177 N13 Leaders of any NGOs? a.______b.______c.______INTERVIEWER: PROBE d.______FOR NAME, POSITION, e.______AND AFFILIATION. f.______(EXAMPLE: “Joel Phiri, site g.______coordinator, Mangochi World Vision” or just “site coordinator, Mangochi World Vision”) N14 Other individuals here in the a.______area? b.______c.______d.______INTERVIEWER: PROBE e.______FOR NAME,POSITION, f.______AFFILIATION g.______N15 Outside of formal participation a.______or doctrinal discussions, name b.______any other religious leader c.______(congregational, denominational d.______or otherwise) with whom you e.______have regular interaction. f.______g.______INTERVIEWER: PROBE FOR NAME, POSITION, AND AFFILIATION.
N16 What about issues of HIV/AIDS, have you Yes………...... 1 discussed issues of HIV/AIDS with anyone No………………..0 ! T1 outside of your own church or mosque, such Don’t Know……88 ! T1 as…? LIST NAME, POSITION, & CHECKER N17 Leaders of other Total # AFFILIATION churches/mosques? Named a.______
b.______INTERVIEWER: PROBE c.______FOR NAME, POSITION, d.______AND AFFILIATION. e.______(EXAMPLE: “Joel Phiri, f.______Pastor, Ulongwe CCAP” or g.______just “Pastor, Ulongwe CCAP”)
178 N18 Leaders of other denominations? a.______b.______INTERVIEWER: PROBE c.______FOR NAME, POSITION, d.______AND AFFILIATION. e.______(EXAMPLE: “Joel Phiri, f.______Pastor, Ulongwe CCAP” or g.______just “Pastor, Ulongwe CCAP”)
N19 Leaders of any NGOs? a.______b.______INTERVIEWER: PROBE c.______FOR NAME, POSITION, d.______AND AFFILIATION. e.______(EXAMPLE: “Joel Phiri, site f.______coordinator, Mangochi World g.______Vision” or just “site coordinator, Mangochi World Vision”)
N20 Other individuals herein the a.______area? b.______c.______INTERVIEWER: PROBE d.______FOR NAME, POSITION, e.______AND AFFILIATION. f.______g.______
179
APPENDIX C
NATIONAL LEADER INTERVIEW GUIDE
180
MRP – National Religious Leaders Interview Guide
Introduction:
Thank you for accepting to have a conversation with me on issues related to your
faith. My name is…………………I am part of a group of researchers from the
College of Medicine, University of Malawi and other researchers from America
and Israel. We are interested in knowing about the activities of your religious
organization in Malawi. I would like to record our conversation so that I do not
miss much of what we discussed. Is this okay with you?
Institutional Level
Section 1 - History:
a. Get the history of the religious faith/Organisation IN MALAWI; how it came
about; when it was introduced in the country; coverage (geographic and
membership); find out if membership is growing or shrinking? Why? How
does he feel about it? How does this particular denomination bring new
members into the faith?
b. Financing – ask how the denomination was and is financed; (probe how much
comes from weekly service collections); what are the organization's finances
used for? How do those funds come from / go to the congregations? Any
changes? What caused the changes?
181
Section 2 - Doctrine:
KEY PROBES
– Distinctions
– particular beliefs (healing, born again, etc.)
Ask for an overview of the doctrine of the particular organization. Focus on what makes it unique or different from other denominations/sects? Ask about healing, belief in being born again, evangelization, prosperity, casting evil spirits etc. How
does the denomination faith effectively evangelize
(congregation/revivals/fellowships/etc.)?
Section 3 - Networking:
KEY PROBES (for each network partner discussed):
- Intended collaboration (ought) vs. actual
collaboration (is)
- Tensions that arise as a result of collaborations a. Connections with local congregations (their branches): (ask specifically how
they collaborate with them. How money flows (which direction? For what
purpose?). [Probes] b. Connection with other religious organizations. Who are they working with?
Which denominations and organisations does he think is growing? Why? How
does he feel about it? Does he think there is competition between faith-based
182 organizations/denominations/religions? Does he think healing is part of the
competition? [Probes]
c. Connections with IFBOs and NGOs: What partnership? Find out if there are
any tensions (ie. Monetary or ideological). [Probes] d. Connections with Government: What departments the organisations
collaborate with? What people? What projects? Ask which areas his
organisation works in collaboration with Government and donors? What type
of support the organisation receives from Government. Ask what Government
policies the organisation has influenced in the past. (If not mentioned ask
about HIV/AIDS policy). Find out if there are any tensions/problems (i.e.
Monetary or ideological). [Probes] e. Connections with international organizations of their denomination: What
international organizations does the organisation collaborate with? What type
of partnership (training, monetary, technical). Find out if there are any
tensions (i.e. Monetary or ideological). [Probes]
Section 4 - Individual level:
a- Personal History: Ask about his appointment. When he
became a leader? How he became a leader? If had any
training, what type of training? Have there been any changes
in his spirituality (increased/decreased belief in God's
intervention/ how God's power affects day to day life).
183 b- Leadership: What are his responsibilities? How often does
he visit local branches of his organization (denomination)?
How often he conducts/attends leadership meetings (at local
level, regional level, national and international level)? How
does he handle disciplinary issues?
c- Frequency of preaching/meetings: (at own branch and other
branches); Find out if he conducts/participates in joint
evangelical meetings with his branch leaders or other
denominational leaders? How effective does he think these
meetings are? Visits to local congregations?
d- [As time permits]: Healing incidences: Has there been
anybody who has been healed due to preaching/teachings
from your denomination? Tell me what happened.
Section 5 - Problems facing the denomination / organisation:
KEY PROBES (after HIV raised as a topic)–
- Content of HIV/AIDS-related messages
- Source of information
- How disseminated a) Ask what is the biggest problem the denomination/organisation is currently
facing? [Probe about HIV/AIDS]. When did he hear about AIDS first? What
were his views about it (doctrine issues come up here)? Did he think of it as a
problem of individual sin or collective sin? How does he view this problem in
184 relation to God (judgement)? How do people (in their organization) react to
people who they think have AIDS? Do people agree on how to handle a
person with AIDS, like praying for the deceased? Has anybody asked for help
in relation to AIDS?
b) Ask about his views about condom use/ family planning/divorce? Has he
been to any workshop related to AIDS? What issues were covered? What
interventions or education services does your denomination/organisation
provide? Who provides these services? How the providers are prepared for
service provision? How training materials are developed. Let him mention any
training/education materials the denomination/organisation has developed in
the past 5-10 years. Are these implemented at the national/local level? How
information passed to the local congregations/members?
[If not mentioned above] ask about his views about orphan care? What is the role of the denomination/organisation?
185
APPENDIX D
GLOBAL-TO-LOCAL SUPPLY CHAIN OF HIV INTERVENTION
A CONCEPTUAL MODEL
186
Source (Dionne 2007)
187
REFERENCES
Adamczyk, Amy, and Jacob Felson. 2006. "Friends' Religiosity and First Sex." Social Science Research 35:924-947.
adams, jimi. 2007. "Damned if You Do, Damned if You Don't: Religion and HIV- Risk Network Structure." in Annual Meetings of the Population Association of America. New York, NY. adams, jimi, and James Moody. 2007. "To Tell the Truth? Measuring Concordance in Multiply-Reported Network Data." Social Networks 29:44-58. adams, jimi, and Jenny Trinitapoli. 2007. "Network Data from Religious Leaders: Congregations and HIV in Rural Malawi." Pp. forthcoming in Applications of Social Network Analysis, edited by Uwe Serdült and Volker Täube: Universität Zürich.
Adimora, Adaora A., and Victor J. Schoenbach. 2005. "Social Context, Sexual Networks, and Racial Disparities in Rates of Sexually Transmitted Infections." Journal of Infectious Diseases 191:S115-S122.
Agadjanian, Victor. 2001. "Religion, Social Milieu and the Contraceptive Revolution." Population Studies 55:135-148.
—. 2005. "Gender, Religious Involvement and HIV/AIDS Prevention in Mozambique." Social Science & Medicine 61:1529-1539.
Agha, Sohail, Paul Hutchinson, and Thankian Kusanthan. 2006. "The effects of religious affiliation on sexual initiation and condom use in Zambia." Journal of Adolescent Health 38:550-555.
Ammerman, Nancy T. 1997. Congregation and Community. New Brunswick, NJ: Rutgers University Press.
Anglewicz, Philip, jimi adams, Francis Obare, S. C. Watkins, and H.-P. Kohler. 2006. "The Malawi Diffusion and Ideational Change Project 2004-06: Data collection, data quality and analyses of attrition." Philadelphia: Population Studies Center, University of Pennsylvania.
188 Anglewicz, Philip and Hans-Peter Kohler. 2005. “Overestimating HIV Infection: The Construction of Subjective Probabilities of HIV Infection in Rural Malawi.” Paper presented at the Meeting of the International Union for the Study of Population (IUSSP). June, Tours, France.
Aral, S. O., N. Padian, and K. K. Holmes. 2005. "Advances in Multilevel Approaches to Understanding the Epidemiology and Prevention of Sexually Transmitted Infections and HIV: An Overview." The Journal of Infectious Diseases 191:S1-6.
Aral, S. O., and Russ Roegner. 2000. "Mathematical Modeling as a Tool in STD Prevention and Control: A Decade of Progress, a Millennium of Opportunities." Sex Transm Dis 27:556-7.
Aral, Sevgi O. 2002. "Understanding Racial-Ethnic and Societal Differences in STI." Sexually Transmitted Infections 78:2-4.
ARHAP, The African Religious Health Assets Pogram. 2006. "Appreciating Assets: The Contribution of Religion to Universal Access in Africa. Mapping, Understanding, Translating and Engaging Religious Health Assets in Zambia and Lesotho. In Suport of Universal Access to HIV/AIDS Treatment, Care and Prevention." Cape Town: World Health Organization.
Astley, W. Graham, and Paramjit S. Sachdeva. 1984. "Structural Sources of Intraorganizational Power: A Theoretical Synthesis." Academy of Management Review 9:104-113.
Barabasi, Albert-Laszlo, and Reka Albert. 1999. "Emergence of Scaling in Random Networks." Science 286:509-512.
Barrett, David B., George Thomas Kurian, and Todd M. Johnson. 2001. World Christian Encyclopedia: A Comparative Survey of Churches and Religions in the Modern World. New York: Oxford University Press.
Barry, Christine A. 1998. "Choosing Qualitative Analysis Software: Atlas/ti and Nudist Compared." Sociological Research Online 3.
Baum, Joel A. C., S. C. Clegg, C. Hardy, and W. R. Ward. 1996. Pp. chapter 2 in Organizational Ecology. Thousand Oaks, CA: Sage.
Bearman, Peter, and Hannah Bruckner. 2005. "After the Promise: the STD Consequences of Adolescent Virginity Pledges." Journal of Adolescent Health 36:271-78.
Bedaiko, Kwame. 1995. Christianity in Africa: The Renewal of a Non-Western Religion. Edinburgh, Scotland: Edinburgh University Press.
189 —. 2000. "Africa and Christianity on the Threshold of the Third Millennium: The Religious Dimension." African Affairs 99:303-323.
Behrman, Jere R., H.-P. Kohler, and S. C. Watkins. 2003. "Social Networks HIV/AIDS and Risk Perceptions." Pp. 32 in PIER Working Paper Series. Philadelphia: University of Pennsylvania.
Behrman, Jere R., Hans-Peter Kohler, and Susan Cotts Watkins. 2002. "Social Networks and Changes in Contraceptive Use Over Time: Evidence from a Longitudinal Study in Rural Kenya." Demography 39:713-738.
Bell, David C., I. D. Montoya, and John S. Atkinson. 2000. "Partner Concordance in Reports of Joint Risk Behaviors." Journal of Acquired Immune Deficiency Syndrome 25:173-181.
Berkowitz, Steven D. 1982. An Introduction to Structural Analysis: The Network Approach to Social Research. Toronto: Butterworths.
Bernard, H. R., and P. D. Killworth. 1977. "Informant accuracy in social network data II." Human Communications Research 4:3-18.
Bertrand, Jane T. 2004. "Diffusion of Innovations and HIV/AIDS." Journal of Health Communication 9:113-121.
Bignami-Van Assche, Simona, Georges Reniers, and Alexander A. Weinreb. 2003. "An Assessment of the KDICP and MDICP Data Quality." Demographic Research S1:31-76.
Black, Donald. 2000a. "Dreams of Pure Sociology." Sociological Theory 18:343-367.
—. 2000b. "The Purification of Sociology." Contemporary Sociology 29:704-709.
Boerma, J. Ties, Elizabeth Holt, and Robert Black. 2001. "Measurement of Biomarkers in Developing Countris: Opportunities and Problems." Population and Development Review 27:303-314.
Borgatti, Stephen P., and R. Cross. 2003. "A Relational View of Information Seeking and Learning." Management Science 49:432-445.
Brewer, Devon D., R. Rothenberg, J. Potterat John, and Stephen Q. Muth. 2007. "Data-Free Modeling of HIV Transmission in Sub-Saharan Africa." Sex Transm Dis 34:54-56.
Brewer, Devon D., Richard B. Rothenberg, Stephen Q. Muth, John M. Roberts, and J. Potterat John. 2006. "Agreement in Reported Sexual Partnership Dates and Implications for Measuring Concurrency." Sexually Transmitted Diseases 33:277-283.
190 Brody, Stuart. 2004. "Declining HIV rates in Uganda: due to cleaner needles, not abstinence or condoms." International Journal of STD & AIDS 15:440-441.
Caldwell, John C., I. O. Orubuloye, and Pat Caldwell. 1999. "Obstacles to Behavioural Change to Lessen the Risk of HIV Infection in the African Epidemic: Nigerian Research." in Resistances to Behavioural Change to Reduce HIV/AIDS Infection in Predominantly Heterosexual epidemics in Third World Countries, edited by John C. Caldwell, Pat Caldwell, John Anarfi, Kofi Awusabo-Asare, James Ntozi, I. O. Orubuloye, Jeff Marck, Wendy Cosford, Rachel Colombo, and Elaine Hollings. Canberra: Health Transition Centre, National Centre for Epidemiology and Population Health, The Australian National University.
Case, Anne, Christina Paxson, and Joseph Ableidinger. 2004. "Orphans in Africa: Parental Death, Poverty, and School Enrollment." Demography 41:483-508.
Chatters, Linda M. 2000. "Religion and Health: Public Health Research and Practice." Annual Review of Public Health 21:335-67.
Chaves, M., M. E. Konieczny, K. Beyerlein, and E. Barman. 1999. "The National Congregations Study: Background, methods, and selected results." Journal for the Scientific Study of Religion 38:458-476.
Chaves, Mark. 2004. Congregations in America. Cambridge, MA: Harvard University Press.
—. 1993. “Denominations as Dual Structures: An Organizational Analysis.” Sociology of Religion 54(2):147-169.
Clark, Shelley. 2004. "Early Marriage and HIV Risks in Sub-Saharan Africa." Studies in Family Planning 35:149-160.
Clark, Shelley, H.-P. Kohler, and Michelle Poulin. 2006. "Crossing the Marriage Threshold in a High HIV Setting." Unpublished Manuscript. Boston: Boston University.
Coleman, James S. 1986. "Social Theory, Social Research and a Theory of Action." American Journal of Sociology 91:1309-1335.
—. 1988. "Social Capital in the Creation of Human Capital." American Journal of Sociology 94:S95-S120.
Coleman, James S., E. Katz, and H. Menzel. 1966. "The diffusion of an innovation among physicians." Sociometry 20:253-270.
Collins, Randall. 1988. The Sociology of Philosophies: A Global Theory of Intellectual Change. Cambridge, MA: Harvard University Press.
191 Crampin, Amelia C., Judith R. Glynn, Bagrey M. M. Ngwira, Frank D. Mwaungulu, Jörg M. Pönnighaus, David K. Wanrdorff, and Paul E. M. Fine. 2003. "Trends and Measurement of HIV Prevalence in Northern Malawi." AIDS 17:1817- 1825.
Darrow, W. W., J. J. Potterat, R. B. Rothenberg, D. E. Woodhouse, S. Q. Muth, and A. S. Klovdahl. 1999. "Using knowledge of social networks to prevent human immunodeficiency virus infections: the Colorado Springs study." Sociological Focus 32:143-158.
DeRose, Laurie F., F. Nii-Amoo Dodoo, and Vrushali Patil. 2002. "Schooling and Attitudes on Reproductive-Related Behavior in Ghana." International Journal of Sociology of the Family 30:50-65.
Derrida, Jacques. 1980. Writing and Difference. Chicago: University of Chicago Press.
Dezso, Zoltan, and A. L. Barabasi. 2002. "Halting Viruses in Scale-Free Networks."
Dezso, Zoltan, and Albert-Laszlo Barabasi. 2001. "Can we stop the AIDS epidemic?" Pp. cond-mat/0107420.
Doctor, Henry V., and Alexander A. Weinreb. 2003. "Estimation of AIDS Adult Mortality by Verbal Autopsy in Rural Malawi." AIDS 17:2509-2513.
Dodoo, F. Nii-Amoo. 1993. "Education and Changing Reproductive-Behavior in Ghana." Sociological Perspectives 36:241-256.
—.1998. "Marriage Type and Reproductive Decisions: A Comparative Study in Sub- Saharan Africa." Journal of Marriage and the Family 60:232-242.
Ellison, Christopher G. 1991. "Religious Involvement and Subjective Well-being." Journal of Health and Social Behavior 32:80-99.
Ellison, Christopher G., Jeffery A. Burr, and Patricia L. McCall. 1997. "Religious Homogeneity and Metropolitan Suicide Rates." Social Forces 76:273-299.
Ellison, Christopher G., and Jeffrey S. Levin. 1998. "The Religion-Health Connection: Evidence, Theory, and Future Directions." Health Education & Behavior 25:700-720.
Englund, Harri. 2003. "Christian Independency and Global Membership: Pentecostal Extraversions in Malawi." Journal of Religion in Africa 33:83-111.
Erickson, Bonnie H., and T.A. Nosanchuck. 1983. "Applied Network Sampling." Social Networks 5:367-382.
192 Erickson, Bonnie H., T.A. Nosanchuck, and Edward Lee. 1981. "Network Sampling in Practice: Some Second Steps." Social Networks 3:127-136.
Feidler, Klaus. 2004. "The Process of Religions Diversification in Malawi: A Reflection on Method and a First Attempt at Synthesis." Zomba: University of Malawi.
Ford, Julie, and Charles Kadushin. 2002. "Between Sacral Belief and Moral Community: A Multidimensional Approach to the Relationship between Religion and Alcohol among Whites and Blacks." Sociological Forum 17:255-279.
Freeman, Linton C. 1979. "Centrality in social networks: Conceptual Clarification." Social Networks 1:215-239.
—. 2004. The Development of Social Network Analysis: A Study in the Sociology of Science. Vancouver, BC: Empirical Press.
Friedkin, Noah E. 1991. "Theoretical Foundations for Centrality Measures." American Journal of Sociology 96:1478-1504.
—. 1998. A Structural Theory of Social Influence. Cambridge: Cambridge University Press.
Friedman, Samuel R., Alan Neaigus, Benny Jose, Richard Curtis, Marjorie Goldstein, Gilbert Ildefonso, Richard B. Rothenberg, and Don C. Des Jarlais. 1997. "Sociometric Risk Networks and Risk for HIV Infection." American Journal of Public Health 87:1289-1296.
Garner, Robert C. 2000. "Safe Sects? Dynamic Religion and AIDS in South Africa." Journal of Modern African Studies 38:41-69.
Gauri, Varun, and Evan S. Lieberman. 2006. "Boundary Institutions and HIV/AIDS Policy in Brazil and South Africa." Studies in Comparative International Development 41:47-73.
Gerland, Patrick. 2006. "Effect os Social Interactions on Individual AIDS-Prevention Attitudes and Behaiors in Rural Malawi." Pp. 269 in Sociology: Princeton University.
Gisselquist, David, J. Potterat John, and Stuart Brody. 2004. "Response: Debate about iatrogenic HIV transmission should not be a pretext for inaction." International Journal of STD & AIDS 15:623-625.
Gisselquist, David, Richard Rothenberg, John J. Potterat, and Ernest Drucker. 2002. "HIV infections in sub-Saharan Africa not explained by sexual or vertical transmission." International Journal of STD & AIDS 13:657-666.
193 Granovetter, Mark. 1973. "The Strength of Weak Ties." American Journal of Sociology 81:1287-1303.
—. 1977. "Network Sampling: Some First Steps." American Journal of Sociology 81:1287-1303.
—. 1985. "Economic Action and Social Structure: The problem of Embeddedness." American Journal of Sociology 91:481-510.
Grassly, N. C., G. Garnett, Bernhard Schwartlander, S. Gregson, and Roy M. Anderson. 2001. "The Effectiveness of HIV Prevention and the Epidemiological COntext." Bulletin of the World HEalth Organization 79:1121-1135.
Gray, Peter B. 2004. "HIV and Islam: is HIV prevalence lower among Muslims?" Social science & medicine 58:1751-1756.
Gray, Ronald H., Maria J. Wawer, Rob Brookmeyer, Nelson K. Sewankambo, David Serwadda, Fred Wabwire-Mangen, Tom Lutalo, Xianbin Li, Thomas vanCott, Thomas C. Wuinn, and the Rakai Project Team. 2001. "Probability of HIV-1 transmission per coital act in monogamous, heterosexual, HIV-1 discordant couples in Rakai, Uganda." The Lancet 357:1149-53.
Green, Edward C. 2003a. "Faith-Based Organizations: Contributions to HIV Prevention." Washington DC: USAID/Washington and The Synergy Project, TvT Associates.
—. 2003b. Rethinking AIDS Prevention: Learning from Successes in Developing Countries. Westport, CT: Praeger Publishers.
Green, Edward C., Vinand Nantulya, Rand Stoneburner, and John Stover. 2002. "What Happened in Uganda? Declining HIV Prevalence, Behavior Change, and the National Response." in Project Lessons Learned Case Study, edited by Janice A. Hogle. Washington, D.C.: U.S. Agency for International Development.
Gregson, Simon, Tom Zhuwau, Roy M. Anderson, Tendayi Chimbadzwa, and Stephen K. Chiwandiwa. 1995. "Age and Religion Selection Biases in HIV-1 Prevalence Data from Antenatal Clinics in Manicaland, Zimbabwe." Central African Journal of Medicine 41:339-345.
Gunter, M., and L. Hue. 2000. "Jamaican Religious Cultre and its Role in Acceleration of HIV/AIDS and stigmatization of PLWAs." in XIIIth International AIDS Conference. Durban, South Africa.
Hallett, T. B., J. Aberle-Grasse, G. Bello, L. M. Boulos, M. P. Cayemittes, B. Cheluget, J. Chipeta, R. Dorrington, S. Dube, A. K. Ekra, J. M. Garcia- Calleja, G. P. Garnett, S. Greby, S. Gregson, J. T. Grove, S. Hader, J. Hanson, 194 W. Hladik, S. Ismail, S. Kassim, W. Kirungi, L. Kouassi, A. Mahomva, L. Marum, C. Maurice, M. Nolan, T. Rehle, J. Stover, and N. Walker. 2006. "Declines in HIV prevalence can be associated with changing sexual behaviour in Uganda, urban Kenya, Zimbabwe, and urban Haiti." Sex Transm Infect 82 Suppl 1:i1-8.
Handcock, Mark S., James Holland Jones, and Martina Morris. 2003. "On "Sexual Contacts and Epidemic Thresholds," Models and Inference for Sexual Partnership Distributions."
Hannan, Michael T., and John H. Freeman. 1977. "The Population Ecology of Organizations." American Journal of Sociology 82:929-964.
Haynie, Dana. 2001. "Delinquent peers Revisited: Does Network Structure Matter?" American Journal of Sociology 106:1013-1057.
Hearn, Julie. 2002. "The 'Invisible' NGO: US Evangelical Missions in Kenya." Journal of Religion in Africa 32:32-60.
Helleringer, S., and H.-P. Kohler. 2005. "Social networks, perceptions of risk, and changing attitudes towards HIV/AIDS: New evidence from a longitudinal study using fixed-effects analysis." Population Studies 59:265-282.
Herek, Gregory M., Keith F. Widaman, and John P. Capitanio. 2005. "When Sex Equals AIDS: Symbolic Stigma and Heterosexual Adults' Inaccurate Beliefs about Sexual Transmission of AIDS." Social Problems 52:15-37.
Hill, Zelee E., John Cleland, and Mohamed M. Ali. 2004. "Religious Affiliation and Extramarital Sex Among Men in Brazil." International Family Planning Perspectives 30:20-26.
Horowitz, L. 1983. "Projective Identification in Dyads and Groups." International Journal of Group Psychotherapy 33:259-279.
Hummer, Robert A., Richard G. Rogers, Charles B. Nam, and Christopher G. Ellison. 1999. "Religious Involvement and U.S. Adult Mortality." Demography 36:273-285.
Idler, Ellen L. 1995. "Religion, Health, and Nonphysical Senses of Self." Social Forces 74:683-704.
Jenkins, Philip. 2002a. The Next Christendom: The Rise of Global Christianity. New York: Oxford University Press.
—. 2002b. The Next Christendom: The Rise of Global Christianity. New York: Oxford University Press.
195 Jenkins, Richard A. 1995. "Religion and HIV: Implications for Research and Intervention." Journal of Social Issues 51:131-144.
Jones, James, and Mark Handcock. 2003. "Sexual contacts and epidemic thresholds." Nature 423:605-606.
Jubber, Ken. 1977. "The Persecution of Jehovah's Witnesses in Southern Africa." Social Compass 24:121-134.
Keister, Lisa A. 2001. "Exchange Structures in Transition: A Longitudinal Study of Lending and Trade Relations in Chinese Business Groups." American Sociological Review 66:336-360.
Killworth, P. D., and H. R. Bernard. 1976. "Informant Accuracy in Social Network data." Human Organizations 35:269-286.
Kirungi, W. L., J. Musinguzi, E. Madraa, N. Mulumba, T. Callejja, P. Ghys, and R. Bessinger. 2006. "Trends in antenatal HIV prevalence in urban Uganda associated with uptake of preventive sexual behaviour." Sex Transm Infect 82 Suppl 1:i36-41.
Klovdahl, Alden S. 1985. "Social Networks and the spread of infectious diseases: The AIDS example." Social Science Medicine 21:1203-1216.
Kohler, H.-P. 2000. "Social Interactions and Fluctuations in Birth Rates." Population Studies 54:223-237.
Kohler, H.-P., Jere R. Behrman, and S. C. Watkins. 2000. "Empirical Assessments of Social Networks, Fertility and Family Planning Programs: Nonlinearities and Their Implications." Demographic Research 3:7.
—. 2001. "The Density of Social Networks and Fertility Decisions: Evidence from South Nyanza District, Kenya." Demography 38:43-58.
Krause, Neal, Christopher G. Ellison, and Keith M. Wulff. 1998. "Church-Based Emotional Support, Negative Interaction, and Psychological Well-Being: Findings from a National Sample of Presbyterians." Journal for the Scientific Study of Religion 37:725-741.
Kuhn, Thomas. 1970. The Structure of Scientific Revolutions. Chicago: University of Chicago Press.
Kumbasar, Ece, A. Kimball Romney, and William H. Batchelder. 1994. "Systematic Biases in Social Perception." American Journal of Sociolgy 100:477-505.
Lagarde, E., C. Enel, K. Seck, A. Gueye, J. P. Piau, G. Pison, V. Delaunay, I. Ndoye, and S. Mboup. 2000. "Religion and Protective Behaviours towards AIDS in Rural Senegal." AIDS 14:2027-2033. 196 Laumann, Edward O., and Yoosik Youm. 1999. "Racial/Ethnic Differences in the Prevalence of Sexually Transmitted Diseases in the United States: A Network Explanation." Sexually Transmitted Diseases 26:250-261.
Levin, Jeffrey S. 1994. "Religion and Health: Is There an Association, Is It Valid, and Is It Causal?" Social science & medicine 38:1475-1482.
Liebowitz, Jeremy. 2004. "The Impact of Faith-Based Organizations on HIV/AIDS Prevention and Mitigation in Africa." Health Economics and HIV/AIDS Research Division (HEARD): University of Natal.
Liljeros, F., C. R. Edling, H. E. Stanley, Y. Aberg, and L. A. N. Amaral. 2003. "Distributions of Number of Sexual Partnerships Have Power Law Decaying Tails and Finite Variance."
Liljeros, Fredrik, Christofer R. Edling, Luis A. Nunes Amaral, H. Eugene Stanley, and Yvonne Aberg. 2001. "The Web of Human Sexual Contacts." Nature 411:907-908.
Lin, Nan Cook Karen Burt Ronald S. 2001. Social Capital: Theory and Research. Hawthorne, NY: Aldine de Gruyter.
Marsden, Peter V. 1987. "Core Discussion Networks of Americans." American Sociological Review 52:122-131.
—. 1990. "Network Data and Measurement." Annual Review of Sociology 16:435- 463.
—. 2003. "Interviewer Effects in Measuring Network Size Using a Single Name Generator." Social Networks 25:1-16.
Martin, H. Gayle. 2003. "A Comparative Analysis of the Financing of HIV/AIDS Programmes: In Botswana, Leosotho, Mozambique, South Africa, Swaziland and Zimbabwe." in Social Aspects of HIV/AIDS and Health Research Program, edited by Human Sciences Research Council.
Mayhew, Bruce. 1980. "Structuralism versus Individualism: Part 1, Shadowboxing in the Dark." Social Forces 59:335-375.
—. 1981. "Structuralism versus Individualism: Part II, Ideological and Other Obfuscations." Social Forces 59:627-648.
McPherson, J. Miller. 1982. "Hypernetwork Sampling: Duality and Differentiation among Voluntary Organizations." Social Networks 3:225-249.
Merton, Robert K. 1968. Social Theory and Social Structure. New York: The Free Press.
197 Mische, Ann, and Harrison White. 1998. "Between Conversation and Situation: Public Switching Dynamics across Network Domains." Social Research 65:695-724.
Mkandawire, Owen. 2000. ""The LIving Waters Church: A Historical, Cultural and Theological Approach: a Study of Church Growth"."
Moody, James. 1999. "SPAN: SAS Programs for Analyzing Networks." The Ohio State University.
—. 2001. "Race, school integration, and friendship segregation in America." American Journal of Sociology 107:679-716.
—. 2004. "The structure of a social science collaboration network: Disciplinary cohesion from 1963 to 1999." American Sociological Review 69:213-238.
—. 2007. "Small World Mechanisms Fail on Dynamic Networks." Durham, NC: Duke University.
Moody, James, and jimi adams. 2006. "The Relative Contribution of Sex and Drug Ties to STI-relevent Network Connectivity." in Sunbelt XXVI, Annual Meetings of the International Network for Social Network Analysis. Vancouver, BC.
Moody, James, Martina Morris, jimi adams, and Mark Handcock. 2003. "Epidemic Potential in Low Degree Networks." Under Review.
Moody, James, and Douglas R. White. 2003. "Structural cohesion and embeddedness: A hierarchical concept of social groups." American Sociological Review 68:103-127.
Moreno, Jacob L. 1934. Who Shall Survive? Washington, DC: Nervous and Mental Disease Publishing Company.
Morris, Martina. 1993. "Epidemiology and Social Networks: Modeling Structured Diffusion." Sociological Methods and Research 22:99-126.
—. 2004. Network Epidemiology: A Handbook for survey design and Data Collection. London: Oxford University Press.
Muhr, Thomas. 2004. "User's Manual for ATLAS.ti 5.0." Berlin: ATLAS.ti Scientific Software Development.
Mullins, Nicholas, and Carolyn J. Mullins. 1973. Theories and Theory Groups in Contemporary American Sociology. New York: Harper & Row.
198 NAC, National Aids Commission. 2004. "Estimating National HIV Prevalence in Malawi from Sentinel Surveillance Data: Technical Report." Lilongwe, Malawi: POLICY Project.
—. 2006. "Performance Assessment of Umbrella Organizations Report." Pp. 69. Lilongwe, Malawi: National AIDS Comission.
Newman, M. E. J. 2001. "The Structure of Scientfic collaboration networks." Porceedings of the National Acadamy of Sciences 98:404-409.
—. 2002. "Spread of Epidemic disease on Networks." Physical Review E 66 016128.
Nnko, Soori, J. Ties Boerma, Mark Urassa, Gabriel Mwaluko, and Basia Zaba. 2004. "Secretive females or swaggering males? An assessment of the quality of sexual partnership reporting in rural Tanzania." Social science & medicine 59:299-310.
NSO, National Statistical Office and ORC Macro. 2005. "Malawi Demographic and Health Survey 2004." Pp. 480, edited by NSO and ORC Macro. Calverton, MD.
—. 2006. "Statistical Yearbook, 2006." Lilongwe, Malawi.
Obare, Francis. 2005. "The Effect of Non-Response on Population-Based HIV Prevalence Estimates: The Case of Rural Malawi." in SNP Working Papers. Philadelphia, PA: Population Studies Center, University of Pennsylvania.
Obare, Francis, Peter Fleming, Philip Anglewicz, Francis Martinson, Agatha Kapatuka, Susan C. Watkins, and H.-P. Kohler. 2007. "HIV Incidence and HIV Prevention in Rural Malawi: Evidence Based on Voluntary Counseling and Testing in Rural Malawi." Philadelphia: University of Pennsylvania.
Orubuloye, I. O., John C. Caldwell, and Pat Caldwell. 1993. "The Role of Religious Leaders in Changing Sexual Behaviour in Southwest Nigeria in an Era of AIDS." Health Transition Review 3:93-104.
Padgett, John F., and Christopher K. Ansell. 1993. "Robust Action and the Rise of the Medici, 1400-1434." American Journal of Sociology 98:1259-1319.
Parkhurst, Justin O. 2002. "The Ugandan Success Story? Evidence and Claims of HIV-1 Prevention." The Lancet 360:78-80.
Parry, Susan. 2003. "Responses of the Faith-Based Organizations to HIV/AIDS in Sub Saharan Africa." World Council of Churches/EHAIA.
PEPFAR, The U.S. President's Emergency Plan for AIDS Relief. 2007. "The U.S. Commitment on Global HIV AIDS."
199 Pescosolido, Bernice A. 1990. "The Social Context of Religious Integration and Suicide: Pursuing the Network Explanation." The Sociological Quarterly 31:337-357.
Pescosolido, Bernice A., and Sharon Georgianna. 1989. "Durkheim, Suicide, and Religion: Toward a Network Theory of Suicide." American Sociological Review 54:33-48.
Pfeiffer, James. 2002. "African Independent Churches in Mozambique: Healing the Afflictions of Inequality." Medical Anthropology Quarterly 16:176-99.
—. 2004. "Condom Social Marketing, Pentecostalism, and Structural Adjustment in Mozambique: A Clash of AIDS Prevention Messages." Medical Anthropology Quarterly 18:77-103.
PRB, Population Reference Bureau. 2006. "Datafinder7."
R, Development Core Team. 2006. "R: A Language and Environment for Statistical Computing." Vienna, Austria: R Foundation for Statistical Computing.
Rankin, S. H., T. Lindgren, W. W. Rankin, and J. Ng'Oma. 2005. "Donkey work: women, religion, and HIV/AIDS in Malawi." Health Care Women Int 26:4- 16.
Regnerus, M. D. 2003. "Moral communities and adolescent delinquency: Religious contexts and community social control." Sociological Quarterly 44:523-554.
Reniers, G. 2006. "HIV/AIDS Surveillance and Behavioral Change in Populations Affected by the AIDS Epidemic. Four Essays." Doctoral Dissertation in Demography & Sociology: University of Pennsylvania.
Rogers, Everett M. 1995. Diffusion of Innovations. New York: The Free Press.
Rostosky, S. S., M. D. Regnerus, and M. L. C. Wright. 2003. "Coital debut: The role of religiosity and sex attitudes in the add health survey." Journal of Sex Research 40:358-367.
Rothenberg, R., and J. Narramore. 1996. "The relevance of social network concepts to sexually transmitted disease control." Sexually transmitted diseases 23:24- 29.
SAS. 2004. "SAS Software, Version 8.2." Cary, NC: SAS Institute Inc.
Scott, John. 2000. Social Network Analysis: A Handbook. London: Sage Publications.
Seidman, Steven. 1991. "The End of Sociological Theory: The Postmodern Hope." Sociological Theory 9:131-146.
200 Singh, Susheela, Jacqueline E. Darroch, and Akinriola Bankole. 2003. "A, B and C in Uganda: The Roles of Abstinence, Monogamy and Condom Use in HIV Decline." in Occasional Report. New York: The Alan Guttmacher Institute.
Smith, K. P., and S. C. Watkins. 2005. "Perceptions of risk and strategies for prevention: responses to HIV/AIDS in rural Malawi." Social Science & Medicine 60:649-660.
Smith, Tom W. 1990. "Classifying Protestant Denominations." Review of Religious Research 31:225-245.
Spaeth, J. L., D. P. O'Rourke, Arne L. Kalleberg, David Knoke, Peter V. Marsden, and J. L. Spaeth. 1996. in Design of the National Organizations Study. Thousand Oaks, CA: Sage.
Stark, Rodney. 1996. "Why Religious Movements Succeed or Fail: A Revised General Model." Journal of Contemporary Religion 11:133-146.
Steensland, B., J. Z. Park, M. D. Regnerus, L. D. Robinson, W. B. Wilcox, and R. D. Woodberry. 2000. "The measure of American religion: Toward improving the state of the art." Social Forces 79:291-318.
Strawbridge, William J., Richard D. Cohen, Sarah J. Shema, and George A. Kaplan. 1997. "Frequent Attendance at Religious Services and Mortality over 28 Years." American Journal of Public Health 87:957-961.
Takyi, Baffour K. 2003. "Religion and Women's Health in Ghana: Insights into HIV/AIDS Preventive and Protective Behavior." Social Science and Medicine 56:1221-1234.
Taylor, Julie J. 2007. "Assisting or Compromising Intervention? The Concept of 'Culture' in Biomedical and Social Research on HIV/AIDS." Social Science & Medicine 64:965-975.
Trinitapoli, Jenny. 2005. "Malawi Religion Project (MRP) Data Collection Protocol: Religious Leaders Ethnosurvey, Sermon Reports and Lay Interviews.” Unpublished Manuscript, Univeristy of Texas.
—. 2006a. "Religion and HIV-Related Behavior Change in Sub-Saharan Africa." in Society for the Scientific Study of Religion. Portland, OR.
—. 2006b. "Religious responses to aids in sub-Saharan Africa: An examination of religious congregations in rural Malawi." Review of Religious Research 47:253-270.
—. 2007. "The Role of Religious Congregations in the AIDS Crisis of Sub-Saharan Africa." in Sociology. Austin, TX: University of Texas at Austin.
201 Trinitapoli, Jenny, and Mark D. Regnerus. 2006. "Religion and HIV Risk Behaviors Among Married Men: Initial Results from a Study in Rural Sub-Saharan Africa." Journal for the Scientific Study of Religion 45:505-528.
Tucker, James. 2002. "Becoming a Pure Sociologist." Contemporary Sociology 31:661-664.
UNAIDS. 2006a. "2006 Report on the Global AIDS Epidemic: A UNAIDS 10th Anniversary Special Edition." Pp. 590: Joint United Nations Programme on HIV/AIDS and World Health Organization.
—. 2006b. "AIDS Epidemic Update: Special Report on HIV/AIDS." Pp. 96: Joint United Nations Programme on HIV/AIDS (UNAIDS) and World Health Organization (WHO)
USAID, UNAIDS, WHO, and The Policy Project. 2003. "The Level of Effort in the National Response to HIV/AIDS: The AIDS Program Effort Index (API) 2003 Round."
Valente, T. W., S. C. Watkins, M. N. Jato, A. VanderStraten, and L. P. M. Tsitsol. 1997. "Social network associations with contraceptive use among Cameroonian women in voluntary associations." Social Science & Medicine 45:677-687.
Valente, Thomas W. 1995. Network Models of the Diffusion of Innovations. Cresskill, NJ: Hampton Press.
Valente, Thomas W., and Rebecca L. Davis. 1999. "Accelerating the Diffusion of Innovations Using Opinion Leaders." The Annals of the American Academy of the Political and Social Sciences 566:55-67.
Warner, R. Stephen. 1993. “Work in Progress Toward a New Paradigm for the Sociological Study of Religion in the United States”. American Journal of Sociology 98:1044-1093.
Wasserman, Stanley, and Katherine Faust. 1994. Social Network Analysis: Methods and Applications. New York: Cambridge University Press.
Watkins, S. C. 2004. "Navigating the AIDS epidemic in rural Malawi." Population and Development Review 30:673-693.
Watkins, Susan C. 2000. "Local and Foreign Models of Reproduction in Nyanza Province, Kenya." Population and Development Review 26:725-759.
Watkins, Susan C., jimi adams, Philip Anglewicz, Stephane Helleringer, Christopher Manyamba, James Mwera, Georges Reniers, and Alexander Weinreb. 2006. "A Rose by Any Other Name: Identifying Elusive and Eager Would-be
202 Respondents in Longitudinal Data Collection." Philadelphia: Population Studies Center, University of Pennsylvania.
Watkins, Susan C., and Chiweni Chimbwete. 2004. "Repentence and Hope Among Christians and Muslims in Rural Malawi." Religion in Malawi 11:1-13.
Watkins, Susan C., and Ina Warriner. 2003. "How do we Know we Need to Control for Selectivity?" Demographic Research Special Collection 1:4.
Watkins, Susan C., Eliya M. Zulu, Hans-Peter Kohler, and Jere R. Behrman. 2003. "Introduction to Social Interactions and HIV/AIDS in Rural Africa." Demographic Research S1:1-30.
Weber, Max. 1978. Economy and Society. Berkeley: University of California Press.
Welch, Michael R., Charles R. Tittle, and Thomas Petee. 1991. "Religion and Deviance among Adult Catholics: A Test of the "Moral Communities" Hypothesis." Journal for the Scientific Study of Religion 30:159-172.
Wellman, Barry. 1988. "Structural analysis: from method and metaphor to theory and substance." Pp. 19-61 in Social Structures: A Network Approach, edited by Barry Wellman and S. D. Berkowitz. Cambridge: Cambridge University Press.
Wellman, Barry, and S. D. Berkowitz. 1988. Social Structures: A Network Approach. Cambridge: Cambridge University Press.
White, Douglas R., and Paul Jorion. 1992. "Representing and computing kinship: a new approach." Current Anthroplogy 33:454-463.
—. 1996. "Kinship networks and discrete structure theory: applications and implicaitons." Social Networks 18:267-314.
White, Harrison C. 1963. An Anatomy of Kinship. Englewood Cliffs, NJ: Prentice Hall.
Woodhouse, D. E., R. B. Rothenberg, J. Potterat John, W. W. Darrow, S. Q. Muth, A. S. Klovdahl, H. P. Zimmerman, H. L. Rogers, T. S. Maldonado, J. B. Muth, and J. U. Reynolds. 1994. "Mapping a social network of heterosexuals at high risk of human immunodeficiency virus infection." AIDS 8:1331-1336.
World Health Organization and UNAIDS, WHO. 2003. "Reconciling Antenatal Clinic-Based Surveillance and Population-Based Survey Estimates of HIV Prevalence in sub- Saharan Africa."
Youm, Y., and E. O. Laumann. 2002. "Social Network Effects on the Transmission of Sexually Transmitted Diseases." Sexually Transmitted Diseases 29:689-697.
203 Zeggelink, Evelien P. H. 1995. "Evolving Friendship networks: an individual- oriented approach implementing similarity." Social Networks 17:83-110.
Zeggelink, Evelien P. H., Frans N. Stokman, and Gerhard G. Van De Bunt. 1996. "The Emergence of groups in the evolution of friendship networks." Journal of Mathematical Sociology 21:29-55.
204