Emerging Trends in The Social and Behavioral Sciences

AIDS and Social Networks

Item

Title

AIDS and Social Networks

Author

Weinreb, Alexander

Adams, Jimi

Trinitapoli, Jenny

Research Area

The Individual and Society

Topic

Health and Illness

Abstract

During the past 30 years, research on the global AIDS pandemic and on social networks has coevolved. Insights from social networks literature have advanced our understandings of AIDS; simultaneously, key empirical insights from the AIDS literature have furthered the development of social network research—especially methodologically. We elaborate on this reciprocal relationship, identifying some of the key developments and future directions for research on AIDS and on social networks generally. From existing literatures, we discuss how (i) social networks analysis was central to early attempts to understand the spread of HIV through sexual and needle‐sharing relationships; (ii) subsequent prevention efforts leveraged similar insights to different ends; (iii) social networks have been crucial in understanding patterns of care for people living with HIV/AIDS; and (iv) the structural composition of networks across international, organizational, and individual levels highlights the epidemic's global implications in ways that extend far beyond epidemiology. We contend that future research must integrate recent developments from both fields in order advance understandings. Among these, we identify as most promising: (i) a move from static modeling approaches toward research emphasizing the dynamic properties of networks; (ii) a shifting focus from single networks in isolation (e.g., sexual transmission networks) to the analysis of multiplex networks (i.e., those involving multiple relationship types represented simultaneously); and (iii) an acknowledgment—conceptual and methodological—of the “vertical” embeddedness of networks. Continued advances in this area will require the gathering of high quality social network data specifically designed to address such questions.

Related Essays

Inefficiencies in Health Care Provision (Economics), James F. Burgess et al.

Problems Attract Problems: A Network Perspective on Mental Disorders (Psychology), Angélique Cramer and Denny Borsboom

Self‐Fulfilling Prophesies, Placebo Effects, and the Social‐Psychological Creation of Reality (Sociology), Alia Crum and Damon J. Phillips

The Development of Social Trust (Psychology), Vikram K. Jaswal and Marissa B. Drell

Emerging Trends in Social Network Analysis of Terrorism and Counterterrorism (Sociology), David Knoke

Network Research Experiments (Methods), Allen L. Linton and Betsy Sinclair

Immigrant Health Paradox (Sociology), Kyriakos S. Markides and Sunshine Rote

Rationing of Health Care (Sociology), David Mechanic

The Role of School‐Related Peers and Social Networks in Human Development (Psychology), Chandra Muller

Health and Social Inequality (Sociology), Bernice A. Pescosolido

Social Relationships and Health in Older Adulthood (Psychology), Theodore F. Robles and Josephine A. Menkin

The Role of Cultural, Social, and Psychological Factors in Disease and Illness (Sociology), Robert A. Scott

Incarceration and Health (Sociology), Christopher Wildeman

Identifier

etrds0007

extracted text

AIDS and Social Networks
ALEXANDER WEINREB, jimi adams, and JENNY TRINITAPOLI

Abstract
During the past 30 years, research on the global AIDS pandemic and on social
networks has coevolved. Insights from social networks literature have advanced
our understandings of AIDS; simultaneously, key empirical insights from the AIDS
literature have furthered the development of social network research—especially
methodologically. We elaborate on this reciprocal relationship, identifying some
of the key developments and future directions for research on AIDS and on
social networks generally. From existing literatures, we discuss how (i) social
networks analysis was central to early attempts to understand the spread of HIV
through sexual and needle-sharing relationships; (ii) subsequent prevention efforts
leveraged similar insights to different ends; (iii) social networks have been crucial
in understanding patterns of care for people living with HIV/AIDS; and (iv)
the structural composition of networks across international, organizational, and
individual levels highlights the epidemic’s global implications in ways that extend
far beyond epidemiology. We contend that future research must integrate recent
developments from both fields in order advance understandings. Among these, we
identify as most promising: (i) a move from static modeling approaches toward
research emphasizing the dynamic properties of networks; (ii) a shifting focus from
single networks in isolation (e.g., sexual transmission networks) to the analysis of
multiplex networks (i.e., those involving multiple relationship types represented
simultaneously); and (iii) an acknowledgment—conceptual and methodological—of
the “vertical” embeddedness of networks. Continued advances in this area will
require the gathering of high quality social network data specifically designed to
address such questions.

INTRODUCTION
The following facts are widely circulating but bear repetition. AIDS is
the pandemic of the era. Even with uncertainty about when HIV, the
virus that causes AIDS, first emerged, since it first came to broad public
attention in 1982, it has been responsible for more than 25 million deaths. A
disproportionate number of these deaths have occurred in the sub-Saharan
African (SSA) “AIDS belt” that extends from southern into eastern African
countries, and that includes populations where more than 20% of adults
are HIV-positive. Other significant AIDS clusters have formed around gay
Emerging Trends in the Social and Behavioral Sciences. Edited by Robert Scott and Stephen Kosslyn.
© 2015 John Wiley & Sons, Inc. ISBN 978-1-118-90077-2.

1

2

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

communities, IV drug-users and prostitutes, in both more and less developed
countries. Globally, about 30 million people are infected with HIV.
While AIDS research has been engaged with the social network paradigm
since the 1980s, the form that engagement has taken across disciplines has
varied dramatically. Epidemiologists have drawn on network approaches to
model the spread of HIV; demographers and public health researchers have
explored how new information about prevention spreads and whether, if at
all, this affects subsequent HIV incidence. Sociologists and economists have
described the social support mechanisms used to care for the sick and their
survivors; and public policy and international aid scholars have examined
the networks among international donors, local nongovernmental organizations (NGOs), and religious organizations that have emerged in response
to AIDS. Yet across these disciplines, advances in social networks analysis
(SNA) have also been accelerated by this focus on AIDS. In a sense, SNA and
AIDS research have “grown up” in parallel, each making important contributions to the other.
In this essay, we outline the foundational and contemporary claims of these
literatures at the nexus of AIDS and social networks, and we point to promising avenues for future research. The literature on “AIDS and social networks”
is sometimes more focused on HIV than AIDS and throughout we follow
scholarly convention using “HIV” when looking at transmission and prevention, and “AIDS” when examining treatment, social support, or policy
in general. It should be clear that we are not restricting ourselves to a particular discipline. Rather, we see utility in bridging—as network theorists
would describe it—different disciplines. This orientation reflects our collective professional experience. Two of us are social demographers whose primary area of research over the past few years has been AIDS in Africa; the
third is a social networks researcher with interests in Africa and the United
States. All of us, however, have considerable experience collaborating with
scholars trained in sociology, economics, epidemiology, public health, and
anthropology—each of which has developed its own approach(es) to understanding and analyzing networks and AIDS.
Notwithstanding our commitment to bridging disciplines, we do
not—indeed, cannot—include everything. Most notably, we exclude
much research on AIDS where an underlying relational focus that intimates a
network approach is not reflected in actual analysis of networks. Examples
include much of the research on: how religion influences people’s sexual
behavior or what people know about AIDS; the consequences of marriage
and divorce both for individual risk and population-level epidemics; how
trust in partnerships influences condom use; relationship quality and condoms; and the relationships that structure caregiving obligations in the wake
of AIDS. The increasing number of papers on topics like these reflects the

AIDS and Social Networks

3

spread of a relational paradigm through the social and behavioral sciences,
especially in social and behavioral research on health. However, they are not
social networks papers in the normal sense of that term.
FOUNDATIONAL RESEARCH
We identify four main streams of foundational research on AIDS and social
networks.
Epidemiologists were the vanguard: the first to apply social networks
approaches to AIDS. From their earliest attempts to identify “patient zero”
in the United States, their goal has been to model the spread of HIV. Their
underlying proposition and finding: the structure of a social network affects
both who you meet and what you do with that person once you’ve met them.
For epidemiologists, this do refers primarily to the behaviors that directly
influence the likelihood of transmission—in the case of HIV, through one
of two principal transmission modes: sex and intravenous drug-use (IDU).
Networks profoundly affect whether, under what conditions, and how
quickly people proceed from meeting each other to having sex or using
drugs; networks further influence the type of sex and kind of drug use—both
of which matter for transmission. Concurrency provides one salient example
of how structured patterns of behaviors can alter the trajectories of an
epidemic above and beyond the constitutive individuals and individual
behaviors. In a series of simulations, Martina Morris et al. have shown that,
at the population level, the arrangement of partnerships is more salient
than their mere number. Where even a moderate number of partnerships
overlap in time (i.e., are “concurrent”), epidemics are much larger than
in populations with the exact same number of partnerships, differently
arranged to minimize time overlaps (such as in serial monogamy).
In offering better explanations for variation in the shape and magnitude of
AIDS epidemics, network approaches have forced an important theoretical
turn, with implications that are much broader than the particular case of
HIV: they demand that we step back from ideas about “risk” that focus
singularly on individuals and their characteristics (both socio-demographic
and behavioral) and, instead, emphasize the salience of relationships and their
characteristics. For example, the simple statement—“even monogamous
women had very large sexual networks,” written by Kay Johnson et al. about
their research in Peru—illustrates this theoretical turn. The idea is virtually
incomprehensible where we conceive of, or model, risk as an individual
property—an implicit assumption in much research on health behavior. It
makes perfect sense, however, when we think about an HIV infection as
arising from a series of conditional probabilities associated with one’s own
partners and one’s partners’ partners.

4

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Other foundational findings from SNA research on the spread of HIV
also bear broader theoretical and practical implications. One concerns the
important role of “bridgers” or “bridge populations”—those individuals
or groups that link otherwise isolated components of a network. Another
challenges traditional “risk factor” models of infection. In conceptualizing
the joint effects of multiple risks, early research assumed that each additional risk factor provided an additive effect on infection. In reality, however,
risky behaviors often co-occur, and although the joint effects are difficult
to disentangle empirically, scholars now agree that their combination is
more often multiplicative or exponential rather than linear. The underlying
argument is simple but important: simultaneity raises issues about the very
meaning of risk and how to target it. For example, it may be less profitable
to think of risky behaviors as measurable phenomena than to think of “risk”
more broadly—as a fuzzy, latent construct that captures a wide range of
associated behaviors.
A second stream of foundational social networks research focused on
HIV prevention. This followed closely on the heels of the literature on
transmission. In sub-Saharan Africa, the epidemic’s epicenter, this approach
essentially cut-and-pasted models of diffusion developed in the mid-1990s
to track the spread of information and ideas about contraception and family
limitation. This new prevention literature stemmed from a theoretical and
explanatory literature on the spread of innovative ideas through networks
that had roots in early work on communications, anthropological work
on social change, and demography’s paradigm-shifting European Fertility
Project. Coincidentally, much of this early work was developed in the same
areas of Central Africa that would later have the heaviest concentrations
of HIV. The prevention literature profited from the rapid publication of a
series of papers in the mid-1990s that clarified conceptually muddy terms
in two important ways. First, these papers made the important conceptual
distinction between two types of network effects—social learning (the flow
of information about how to reduce risk) and social influence (moral and
political evaluation of those risk-reduction behaviors). Second, in work
associated with Hans-Peter Kohler et al., researchers demonstrated that they
could leverage measures of network density to distinguish one from the
other.
The literature on networks and HIV prevention made a few other important contributions. First was its underlying recognition of a structural
parallel between networks that increase transmission risk and networks that
facilitate the flow of information that may ultimately prevent infection. This
structural parallel helps explain the “first in, first out” rule for understanding
the waves of HIV prevalence by social class in developing countries with the
worst generalized epidemics: in these settings, the earliest rise and fall in HIV

AIDS and Social Networks

5

prevalence was associated with the wealthiest, then the middle class, then
the poor—note that a somewhat different pattern has persisted in western
countries. The insight that HIV infection and prevention networks run in
parallel undergirds a key policy effort: “peer-driven intervention,” which is
more effective than standard street-based outreach for preventing infection.
It also helps explain the highly variable record of African countries in combating AIDS. For example, in their comparative study of Uganda, Kenya,
Tanzania, Malawi, Zambia, and Zimbabwe, Daniel Low-Beer and Rand
Stoneburner identify unique patterns of communication about HIV through
Ugandan social networks. They argue that this reflects important differences
in approaches to HIV taken by national political leadership—to which
we add religious leadership. Specifically, intensive long-term (vertical)
communication about HIV from national leaders in Uganda legitimized and
facilitated widespread local (horizontal) communication among community
members.
A related contribution has focused on how networks structure perceptions
of risk. This has been especially valuable for understanding AIDS-related
behaviors in developing countries, where high levels of uncertainty and limited sources of information force people to use networks to both evaluate
risk and vet information about ways to minimize it. These network effects
can be seen even when controlling for unobserved factors that may affect the
structure and composition of the social networks.
A third stream of research on AIDS and social networks has focused on care
and social support for people living with HIV and AIDS (PLWHA). There
has long been a careful distinction in this literature between emotional care
and physical or task-related support. A divergence can be seen, however,
between the literatures on support in wealthy western settings from support
in poor African or Asian settings. In the first it is largely focused on support for PLWHA. In the second, it is equally focused on PLWHA and their
children. This difference is not surprising. In developed countries, AIDS has
disproportionately affected a childless, gay population with small extended
families and access to at least a partial publicly funded safety net. In contrast,
in most countries in SSA the epidemic was, from the beginning, primarily
heterosexual, and it disproportionately affected young parents embedded in
much larger extended families, who were active in local religious congregations, and living in settings devoid of any publicly funded safety net.
The existing literature on social support networks reflects these contextual differences in the sociodemographic characteristics of PLWHA. In the
US, support networks tend to be constituted by a constellation of activists
and other PLWHAs—though more recent literature notes the fragility of this
type of network where people are increasingly affected by age-related comorbidities. In SSA, both emotional and physical support for PLWHA and their

6

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

dependents is sourced from within the extended family or from religious
congregations, as documented by Jenny Trinitapoli and Alexander Weinreb.
These same networks make arrangements to support survivors—typically
children—when the PLWHA dies.
A fourth stream of research on AIDS and social networks has focused
directly on network composition and characteristics, with implications
for prevention, treatment, and social support. There are two quite different
strands of literature here. The first builds on a long tradition in the social
sciences that seeks to identify the effects of larger structural characteristics on the behavior of individuals (e.g., from Durkheim’s On Suicide)
and on networks. In addition, relevant here is the emerging discourse
in modern medicine, which reframes disease as a biosocial phenomenon
rather than simply an outcome of molecular changes. In both cases, larger
structural characteristics and arrangements act as “distal determinants” of
infection—types of “structural violence”—that affect HIV transmission and
treatment by influencing the size and composition of networks. An example:
high risk behavior, including sexual mobility and IV drug use, can be a
predictable byproduct of social disintegration and residential instability and
landholding patterns, or of discriminatory labor market or incarceration
practices. When network stability is compromised, HIV spreads more easily
through one of two mechanisms: (i) older authority structures (i.e., hierarchical networks) are undermined, freeing youth from community sanctions that
may have protected them by proscribing high levels of sexual mobility and
(ii) new peer-driven (i.e., horizontal networks) behavioral models diffuse
alongside new inequalities in wealth, new patterns of consumption-driven
behavior, and the higher-risk activities associated with both.
CUTTING-EDGE RESEARCH
The first 30 years of research on AIDS and social networks established
the contours of engagement across a number of disciplines. Part of this
groundwork involved rapid improvements in modeling the spread of the
HIV (the virus), the spread and adoption of new information, ideas and
behavior associated with HIV prevention, and of modes of social support
for those infected with HIV or affected by AIDS. There is a notable, but often
overlooked, consequence here: by boosting the amount of methodological
and empirical research on networks, AIDS changed disciplines. Recent
advances in epidemiological models of infection, for example, are rooted
in a network paradigm. Moreover, in demography and sociology, HIV
reinvigorated and accelerated the movement of social networks toward the
disciplinary center.

AIDS and Social Networks

7

One stream of contemporary cutting-edge research—focused on sampling
and response error in networks data—cuts across these disciplinary boundaries. This is connected to a long-standing concern in the networks literature
with data quality. It also builds on the recognition that as important a role as
formal models and simulations have played in many key SNA developments
that have emerged from the AIDS literature, their underlying claims are best
tested empirically.
A characteristic difficulty with AIDS in this regard is that people at greatest
risk of HIV infection (e.g., sex workers and IV drug users) tend to be in “hidden” or “hard-to-reach” populations: they are more mobile and marginalized
and thus are rarely found in household samples. They are also less trusting
of authorities in general and are, therefore, reluctant to be interviewed even
when they are found. The greater the magnitude of each of these problems
during data collection, is the less representative the resulting data are.
To address this sampling problem, researchers have developed alternative
sampling strategies. One is time-location sampling. By allowing researchers
to target recruitment in geographically defined hotspots (e.g., bath houses,
and red-light districts) where risky behaviors are concentrated, time-location
sampling directly addresses longstanding critiques of aspatial modeling
paradigms used in the AIDS literature and empirical social science in
general. Another approach is to use a link-tracing-based design, long used
in collecting data on infectious diseases. One recent adaptation of these
strategies—respondent-driven sampling (RDS)—leverages network-based
referral approaches to approximate random sample characteristics. Developed by Douglas Heckathorn et al., RDS explicitly acknowledges the unequal
and non-random tendency of how individuals form social connections. It
argues that, by following referral chains through a series of iterations,
researchers can reach an equilibrium in which the sample composition is
independent of the characteristics of initially sampled “seeds.” RDS has
important weaknesses, such as the inability to tap into portions of the population that are completely isolated from the seeds—this is less problematic
in time-location sampling—and potentially dramatic fluctuations in sample
characteristics from seemingly minor violations of the model’s assumptions.
Despite these, RDS has become de rigueur for sampling hidden and hard to
reach populations, including those at high risk for contracting HIV.
The second strand of methodological research to have emerged in the past
few years reflects growing anxiety about data quality. Because the smallest
units of analysis in social networks focus on relationships—which necessarily include two individuals—social network data are uniquely situated to
evaluate the validity of self-reported information in ways other types of data
cannot explore. For example, some have used such multiply reported data
to demonstrate how reliably individual partnerships are reported both by

8

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

members of those relationships and even by other individuals not part of
the relationship, but connected (directly or indirectly) to the people who are.
Given that small local changes in networks can have substantial implications
for global connectivity of networks among the full population, researchers
must critically evaluate ways to improve reporting. A variety of technological, passive and active participatory strategies have begun to do exactly that:
some use new types of survey instruments such as relationship histories;
others are focusing more on new types of self-interviewing (eliminating survey error associated with interviewers); still others are using new forms of
observational data, for example, by mapping network connections using cellphone data.
A second stream of cutting-edge research examines a completely different
type of AIDS network: vertical institutional networks extending from donors
to local NGOs involved in the “global governance of AIDS.” There are two
important contextual factors here; the first is global. Since the 1970s, a new
institutional configuration has emerged involving funders, development
professionals, international non-governmental organizations (INGO), and
their local agents (NGO). In more general social science this is a particular
type of “transboundary formation.” The second contextual change is specific
to poorer countries with generalized AIDS epidemics: the massive mobilization of international donors around AIDS over the past decade. The United
States President’s Emergency Plan for AIDS Relief (PEPFAR), for example,
was established in 2004 and committed to spending US$15 billion over its
first 5 years. Its reauthorization in 2008—to run from 2009 to 2013—increased
its budget to a maximum of US$48 billion (though some of those resources
are devoted to malaria and TB). The global fund to fight AIDS, tuberculosis,
and malaria—known simply as the global fund (GF)—was founded in 2002.
By December 2009, the GF had approved proposals totaling US$19.2 billion,
virtually all of it for HIV and TB. Billions more AIDS-designated dollars
have been disbursed by the Bill and Melinda Gates Foundation, by the
World Bank’s Multicountry AIDS Program (MAP), and by many other
bilateral agencies.
In settings with limited access to capital, these new institutional configurations have made available massive amounts of resources and have attracted
considerable attention from international and local NGO entrepreneurs, and
from local technocratic elites. The result has been the emergence of new vertical networks that link these major bilateral organizations and global donors
to: (i) INGOs to whom the actual running of programs is subcontracted, (ii)
national NGOs, and (iii) local NGOs and development committees, including
those run by religious organizations. Recent work by Susan Watkins and Ann

AIDS and Social Networks

9

Swidler on how policies filtered through these networks demonstrate enormous differences between their design at the “top” and their implementation
on the ground.
This body of work further shows how these resources are used as “instruments of patronage” and as opportunities to further other parochial
interests. Notably here, in the case of AIDS, the “institutional isomorphism”
imposed by funders is offset by a reshaping or “translation” process. That
is, policies may be nearly identical in their original form, but as they diffuse
down through a network they are repeatedly modified rather than being
directly reproduced. There is an ironic structural parallel here between, on
the one hand, the mutations that have changed HIV over time, resulting in an
ever-growing array of HIV subtypes and recombinations and, on the other
hand, the way that information is transmitted but constantly reconfigured
as it works its way through the network. Cumulatively, these changes lead
to dramatically different modes of implementation and outcomes on the
ground, some of them much more apposite to local setting.
The ways in which vertical networks both distribute resources and (usually
informally) reshape policy is both interesting and important. First, it points
to a unique type of AIDS network that has come into being for one of two reasons. One is the international commitment to combating AIDS. Another—as
cynical as it sounds—is because the magnitude of resources available to AIDS
has allowed for the full expression of development professionals’ scavenger
instincts, enabling them to use AIDS to perpetuate both their institutions and
their own livelihoods. A second reason the vertical networks are important is
that they hold out the tantalizing promise of integrated analyses of different
types of networks: networks through which viruses and information move;
networks through which social support is provided; and networks through
which policy responses are channeled. We return to this promise below.
KEY ISSUES FOR FUTURE RESEARCH
As work on AIDS and social networks moves into its fourth decade, a number
of areas are primed for important contributions. Some of these reflect recent
advances in the social networks literature that have not yet been translated
into work on AIDS. Others address things the networks literature could productively adapt from AIDS research, and still others reflect perspectives and
tools that are new to both fields.
Perhaps the most important developments in social networks literature
over the past decade center on the (re)introduction of temporal dynamics
into network analytic strategies and the development of statistical models
for handling network data. While each of these have some roots in literature
on HIV/AIDS, the latest developments, associated especially with James

10

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Moody, have yet to be fully embraced by AIDS researchers. The network
positional characteristics that we know influence an individual’s chance
of contracting HIV (e.g., centrality) are constantly changing. For example,
a person cannot contract HIV from their current partners’ future partners,
and s/he cannot transmit HIV to a current partner’s former partners.
Recent modeling developments clearly show that these dynamic patterns
substantially alter networks’ epidemic potential, providing a serious caution
against thinking about networks as fixed structures, even for the purposes
of modeling short-term trends. Parallel temporal dynamics exist within
the extended family structures that underlie social support systems in
developing countries. That is, people choose to maintain ties with certain
cousins, uncles and aunts, not all of them; and the absence of one of these
favored network relationships may both increase the likelihood of making
connections to someone else, but also reduce connections to others in the
family once maintained through the now absent bridging tie.
The introduction of statistical models for static or dynamic networks—
known respectively as p* or exponential random graph models (ERGM) and
stochastic actor-based models (SABM) or SIENA models—has moved the
analysis of networks from a purely descriptive endeavor to one with the
empirical wherewithal to emphasize processes. Developed by Tom Snijders
et al., these models allow for the simultaneous estimation of individual and
network-endogenous effects in observed network patterns. For instance,
while early research relied on single snapshots of a network and inferred
processes of influence from across those networks (e.g., in the dissemination
or adoption of strategies for HIV prevention), SABM are being used to disentangle processes of influence (e.g., people adopting the ideas/behaviors of
those to whom they are connected) from those of selection (e.g., individuals
forming ties to others with whom they hold similar views). New implementations of this approach will be especially beneficial to future efforts
to understand whether and how people’s strategies for avoiding infection
“work.” The dynamic and statistical advances from the modeling literature
on HIV have not yet been fully incorporated into strategies for gathering
the high-quality data we described above as necessary for advancing both
fields. Each stands to benefit substantially from future data collection
efforts that target the gathering of longitudinal data on risk-bearing and
knowledge-sharing networks within the observed populations.
AIDS-salient networks tend to be “multiplex” in nature, that is, they
involve more than one type of relationship at a time. This seemingly simple
insight has been a critical contribution of AIDS literature from early days
when, for example, it became clear that “risk” carrying networks (i.e.,
those involving sexual or needle-sharing contact) can also effectively carry
prevention efforts or can be converted into sources of support and care for

AIDS and Social Networks

11

PLWA. The formal models within SNA literature, however, still tend to focus
on a single tie type at a time (e.g., only on sexual networks or social support
networks). Yet including multiple networks has important implications
for understanding network effects in general. For example, analyses of sex
ties or needle-sharing ties alone underestimate the risk network as a whole
and misestimate how sex ties or drug ties themselves shape individual risk
patterns; tracing risk-relevant ties to the exclusion of other “social” network
ties can misrepresent the full extent of the population at risk; and treating
a social support network (in a developing country) as isomorphic with an
extended family misses crucial non-familial support from friends or local
religious organizations.
Future work should focus more explicitly on the myriad ways in which
these different types of networked relationships can combine to influence
HIV/AIDS, in part by systematically building on this idea of multiplexity.
After all, enough is now known about different types of networks, about
analyzing dynamic processes, and about how to collect high-quality data,
that we can envisage collecting and combining data that enable researchers
to estimate individual and network-endogenous effects, but this time across
multiplex networks. Only this combination of data would allow us to directly
address AIDS-related questions that lie in the social networks. For example:
What is the relationship between the structure of nuclear and extended
family, social and sexual networks, local support networks, and how do
these collectively influence HIV transmission, subsequent treatment and
social support? How do these structures change over time (or how do people
change them over time)? What happens when a best friend—or some other
key node in a friend, family, or social support network—moves or dies? What
happens to sexual networks and informal support networks as access to
antiretroviral drugs, especially the latest fixed-dose combinations, increases?
Important vertical elements should also be included in this new dynamic,
multiplex SNA frame. One would focus on structural determinants of networks themselves: from residential patterns and conditions to the quality
and intensity of connections to kin and nonkin; from involvement in institutions that facilitate new network connections (e.g., schools, churches, and
employment outside the home) to deeply entrenched cultural practices that
limit physical movement and network freedoms (e.g., gender differences in
free movement through public space embodied in purdah). Another vertical
element would extend into dynamic organizational networks. Among these
are local and national NGOs that interact with community and family-based
networks in promoting HIV prevention or organizing social support for those
directly or indirectly affected by AIDS.
A final element is related to spatial dynamics. Understanding HIV’s movement through space has been a goal of researchers since early attempts to

12

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

trace “patient zero” and map truckers’, soldiers’, and other migrants’ routes
in Africa. However, spatial modeling of AIDS networks in general has not
been at the forefront of research agendas. This is unfortunate since there are a
number of interesting spatial puzzles. Examples include: clustering of AIDS
orphans in Malawi that is completely orthogonal to HIV prevalence; clusters of AIDS support networks around particular religious congregations;
irregular diffusion of stories and ideas about AIDS within and across communities. Future research should incorporate each of these contextual layers (e.g., network-informed spatial measures or spatially informed network
measures), in order to look at the intersection of network and spatial processes and, subsequently, to directly address the ways in which they impact
outcomes of interest.
In summary, it is widely known that the social epidemiology of AIDS,
from transmission to its long gestation, is complicated. What has become
increasingly clear over the past decade is that single subsets of networks,
no matter how seemingly well circumscribed, cannot proxy for full systems
if our intention is to fully understand and document the relationship
between social networks and AIDS, whether in terms of HIV transmission,
prevention, or care. Future research should reflect this awareness. AIDS
and social networks should be treated empirically as things that occur
within a temporally dynamic, multilevel, and spatially combustible setting.
Excluding any one of these dimensions, or using instruments that artificially
limit scholars’ attention to a single type or subset of behaviorally specific
networks, or to a single period of time, misses a substantial part of the
relationship between AIDS and social networks. We need a more holistic
approach.
FURTHER READING
Bridging
Youm, Y., & Laumann, E. O. (2002). Social network effects on the transmission of
sexually transmitted diseases. Sexually Transmitted Diseases, 29, 689–97.
Care and Social Support
Trinitapoli, J., & Weinreb, A. (2012). Religion and AIDS in Africa. New York, NY:
Oxford University Press.
Concurrency
Morris, M., & Kretzchmar, M. (1997). Concurrent partnerships and the spread of HIV.
AIDS, 11, 641–48.

AIDS and Social Networks

13

Data Quality
Helleringer, S., Kohler, H.-P., Kalilani-Phiri, L., Mkandawire, J., & Armbruster, B.
(2011). The reliability of sexual partnership histories: Implications for the measurement of partnership concurrency during surveys. AIDS, 25(4), 503–11.
Infection Tracing
Potterat, J. J., Woodhouse, D. E., Muth, S. Q., Rothenberg, R. B., Darrow, W. W., Klovdahl, A. S., & Muth, J. B. (2004). Network dynamism: History and lessons of the
Colorado Springs study. In M. Morris (Ed.), Network epidemiology: A handbook for
survey design and data collection. Oxford, England: Oxford University Press.
Multiplexity
adams, j., Moody, J., & Morris, M. (2013). Sex, drugs, and race: How behaviors differentially contribute to sexually transmitted infection risk network structure. American Journal of Public Health, 103(2), 322–29.
Prevention Turn
Heckathorn, D. D., Broadhead, R. S., Anthony, D. L., & Weakliem, D. L. (1999). AIDS
and social networks: HIV prevention through network mobilization. Sociological
Focus, 32(2), 159–179.
Perception of Risk
Kohler, H.-P., Behrman, J. R., & Watkins, S. C. (2007). Social networks and HIV/AIDS
risk perceptions. Demography, 44(1), 1–33.
Statistical Models
Steglich, C., Snijders, T. A. B., & Pearson, M. (2010). Dynamic networks and behavior:
Separating selection from influence. Sociological Methodology, 40(1), 329–93.
Temporality
Moody, J. (2009). Network dynamics. In P. Hedstrom & P. S. Bearman (Eds.), The
Oxford handbook of analytical sociology. New York, NY: Oxford University Press.
“Vertical” Networks
Watkins, S. C., Swidler, A., & Hannan, T. (2012). Outsourcing social transformation:
Development NGOs as organizations. Annual Review of Sociology, 38, 285–315.

14

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

ALEXANDER WEINREB, jimi adams, JENNY TRINITAPOLI
SHORT BIOGRAPHY
Alexander Weinreb (UT-Austin), jimi adams (American University), and
Jenny Trinitapoli (Penn State) are social demographers who think about networks not only in relation to AIDS but also in relation to a variety of other
topics including religion, family structures, political processes and outcomes,
the development of disciplinary fields, and social cohesion generally. They
have lived and worked in multiple countries but as field researchers have
largely focused on collecting and analyzing network-based data in Kenya
and Malawi. Some of their work on these subjects appears in Religion and
AIDS in Africa (OUP 2012) and scholarly journals such as American Sociological Review, American Journal of Public Health, Social Networks, Social
Science and Medicine, Population and Development Review, Demographic
Research, and Field Methods.
RELATED ESSAYS
Inefficiencies in Health Care Provision (Economics), James F. Burgess et al.
Problems Attract Problems: A Network Perspective on Mental Disorders
(Psychology), Angélique Cramer and Denny Borsboom
Self-Fulfilling Prophesies, Placebo Effects, and the Social-Psychological
Creation of Reality (Sociology), Alia Crum and Damon J. Phillips
The Development of Social Trust (Psychology), Vikram K. Jaswal and Marissa
B. Drell
Emerging Trends in Social Network Analysis of Terrorism and Counterterrorism (Sociology), David Knoke
Network Research Experiments (Methods), Allen L. Linton and Betsy Sinclair
Immigrant Health Paradox (Sociology), Kyriakos S. Markides and Sunshine
Rote
Rationing of Health Care (Sociology), David Mechanic
The Role of School-Related Peers and Social Networks in Human Development (Psychology), Chandra Muller
Health and Social Inequality (Sociology), Bernice A. Pescosolido
Social Relationships and Health in Older Adulthood (Psychology), Theodore
F. Robles and Josephine A. Menkin
The Role of Cultural, Social, and Psychological Factors in Disease and Illness
(Sociology), Robert A. Scott
Incarceration and Health (Sociology), Christopher Wildeman



AIDS and Social Networks
ALEXANDER WEINREB, jimi adams, and JENNY TRINITAPOLI

Abstract
During the past 30 years, research on the global AIDS pandemic and on social
networks has coevolved. Insights from social networks literature have advanced
our understandings of AIDS; simultaneously, key empirical insights from the AIDS
literature have furthered the development of social network research—especially
methodologically. We elaborate on this reciprocal relationship, identifying some
of the key developments and future directions for research on AIDS and on
social networks generally. From existing literatures, we discuss how (i) social
networks analysis was central to early attempts to understand the spread of HIV
through sexual and needle-sharing relationships; (ii) subsequent prevention efforts
leveraged similar insights to different ends; (iii) social networks have been crucial
in understanding patterns of care for people living with HIV/AIDS; and (iv)
the structural composition of networks across international, organizational, and
individual levels highlights the epidemic’s global implications in ways that extend
far beyond epidemiology. We contend that future research must integrate recent
developments from both fields in order advance understandings. Among these, we
identify as most promising: (i) a move from static modeling approaches toward
research emphasizing the dynamic properties of networks; (ii) a shifting focus from
single networks in isolation (e.g., sexual transmission networks) to the analysis of
multiplex networks (i.e., those involving multiple relationship types represented
simultaneously); and (iii) an acknowledgment—conceptual and methodological—of
the “vertical” embeddedness of networks. Continued advances in this area will
require the gathering of high quality social network data specifically designed to
address such questions.

INTRODUCTION
The following facts are widely circulating but bear repetition. AIDS is
the pandemic of the era. Even with uncertainty about when HIV, the
virus that causes AIDS, first emerged, since it first came to broad public
attention in 1982, it has been responsible for more than 25 million deaths. A
disproportionate number of these deaths have occurred in the sub-Saharan
African (SSA) “AIDS belt” that extends from southern into eastern African
countries, and that includes populations where more than 20% of adults
are HIV-positive. Other significant AIDS clusters have formed around gay
Emerging Trends in the Social and Behavioral Sciences. Edited by Robert Scott and Stephen Kosslyn.
© 2015 John Wiley & Sons, Inc. ISBN 978-1-118-90077-2.

1

2

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

communities, IV drug-users and prostitutes, in both more and less developed
countries. Globally, about 30 million people are infected with HIV.
While AIDS research has been engaged with the social network paradigm
since the 1980s, the form that engagement has taken across disciplines has
varied dramatically. Epidemiologists have drawn on network approaches to
model the spread of HIV; demographers and public health researchers have
explored how new information about prevention spreads and whether, if at
all, this affects subsequent HIV incidence. Sociologists and economists have
described the social support mechanisms used to care for the sick and their
survivors; and public policy and international aid scholars have examined
the networks among international donors, local nongovernmental organizations (NGOs), and religious organizations that have emerged in response
to AIDS. Yet across these disciplines, advances in social networks analysis
(SNA) have also been accelerated by this focus on AIDS. In a sense, SNA and
AIDS research have “grown up” in parallel, each making important contributions to the other.
In this essay, we outline the foundational and contemporary claims of these
literatures at the nexus of AIDS and social networks, and we point to promising avenues for future research. The literature on “AIDS and social networks”
is sometimes more focused on HIV than AIDS and throughout we follow
scholarly convention using “HIV” when looking at transmission and prevention, and “AIDS” when examining treatment, social support, or policy
in general. It should be clear that we are not restricting ourselves to a particular discipline. Rather, we see utility in bridging—as network theorists
would describe it—different disciplines. This orientation reflects our collective professional experience. Two of us are social demographers whose primary area of research over the past few years has been AIDS in Africa; the
third is a social networks researcher with interests in Africa and the United
States. All of us, however, have considerable experience collaborating with
scholars trained in sociology, economics, epidemiology, public health, and
anthropology—each of which has developed its own approach(es) to understanding and analyzing networks and AIDS.
Notwithstanding our commitment to bridging disciplines, we do
not—indeed, cannot—include everything. Most notably, we exclude
much research on AIDS where an underlying relational focus that intimates a
network approach is not reflected in actual analysis of networks. Examples
include much of the research on: how religion influences people’s sexual
behavior or what people know about AIDS; the consequences of marriage
and divorce both for individual risk and population-level epidemics; how
trust in partnerships influences condom use; relationship quality and condoms; and the relationships that structure caregiving obligations in the wake
of AIDS. The increasing number of papers on topics like these reflects the

AIDS and Social Networks

3

spread of a relational paradigm through the social and behavioral sciences,
especially in social and behavioral research on health. However, they are not
social networks papers in the normal sense of that term.
FOUNDATIONAL RESEARCH
We identify four main streams of foundational research on AIDS and social
networks.
Epidemiologists were the vanguard: the first to apply social networks
approaches to AIDS. From their earliest attempts to identify “patient zero”
in the United States, their goal has been to model the spread of HIV. Their
underlying proposition and finding: the structure of a social network affects
both who you meet and what you do with that person once you’ve met them.
For epidemiologists, this do refers primarily to the behaviors that directly
influence the likelihood of transmission—in the case of HIV, through one
of two principal transmission modes: sex and intravenous drug-use (IDU).
Networks profoundly affect whether, under what conditions, and how
quickly people proceed from meeting each other to having sex or using
drugs; networks further influence the type of sex and kind of drug use—both
of which matter for transmission. Concurrency provides one salient example
of how structured patterns of behaviors can alter the trajectories of an
epidemic above and beyond the constitutive individuals and individual
behaviors. In a series of simulations, Martina Morris et al. have shown that,
at the population level, the arrangement of partnerships is more salient
than their mere number. Where even a moderate number of partnerships
overlap in time (i.e., are “concurrent”), epidemics are much larger than
in populations with the exact same number of partnerships, differently
arranged to minimize time overlaps (such as in serial monogamy).
In offering better explanations for variation in the shape and magnitude of
AIDS epidemics, network approaches have forced an important theoretical
turn, with implications that are much broader than the particular case of
HIV: they demand that we step back from ideas about “risk” that focus
singularly on individuals and their characteristics (both socio-demographic
and behavioral) and, instead, emphasize the salience of relationships and their
characteristics. For example, the simple statement—“even monogamous
women had very large sexual networks,” written by Kay Johnson et al. about
their research in Peru—illustrates this theoretical turn. The idea is virtually
incomprehensible where we conceive of, or model, risk as an individual
property—an implicit assumption in much research on health behavior. It
makes perfect sense, however, when we think about an HIV infection as
arising from a series of conditional probabilities associated with one’s own
partners and one’s partners’ partners.

4

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Other foundational findings from SNA research on the spread of HIV
also bear broader theoretical and practical implications. One concerns the
important role of “bridgers” or “bridge populations”—those individuals
or groups that link otherwise isolated components of a network. Another
challenges traditional “risk factor” models of infection. In conceptualizing
the joint effects of multiple risks, early research assumed that each additional risk factor provided an additive effect on infection. In reality, however,
risky behaviors often co-occur, and although the joint effects are difficult
to disentangle empirically, scholars now agree that their combination is
more often multiplicative or exponential rather than linear. The underlying
argument is simple but important: simultaneity raises issues about the very
meaning of risk and how to target it. For example, it may be less profitable
to think of risky behaviors as measurable phenomena than to think of “risk”
more broadly—as a fuzzy, latent construct that captures a wide range of
associated behaviors.
A second stream of foundational social networks research focused on
HIV prevention. This followed closely on the heels of the literature on
transmission. In sub-Saharan Africa, the epidemic’s epicenter, this approach
essentially cut-and-pasted models of diffusion developed in the mid-1990s
to track the spread of information and ideas about contraception and family
limitation. This new prevention literature stemmed from a theoretical and
explanatory literature on the spread of innovative ideas through networks
that had roots in early work on communications, anthropological work
on social change, and demography’s paradigm-shifting European Fertility
Project. Coincidentally, much of this early work was developed in the same
areas of Central Africa that would later have the heaviest concentrations
of HIV. The prevention literature profited from the rapid publication of a
series of papers in the mid-1990s that clarified conceptually muddy terms
in two important ways. First, these papers made the important conceptual
distinction between two types of network effects—social learning (the flow
of information about how to reduce risk) and social influence (moral and
political evaluation of those risk-reduction behaviors). Second, in work
associated with Hans-Peter Kohler et al., researchers demonstrated that they
could leverage measures of network density to distinguish one from the
other.
The literature on networks and HIV prevention made a few other important contributions. First was its underlying recognition of a structural
parallel between networks that increase transmission risk and networks that
facilitate the flow of information that may ultimately prevent infection. This
structural parallel helps explain the “first in, first out” rule for understanding
the waves of HIV prevalence by social class in developing countries with the
worst generalized epidemics: in these settings, the earliest rise and fall in HIV

AIDS and Social Networks

5

prevalence was associated with the wealthiest, then the middle class, then
the poor—note that a somewhat different pattern has persisted in western
countries. The insight that HIV infection and prevention networks run in
parallel undergirds a key policy effort: “peer-driven intervention,” which is
more effective than standard street-based outreach for preventing infection.
It also helps explain the highly variable record of African countries in combating AIDS. For example, in their comparative study of Uganda, Kenya,
Tanzania, Malawi, Zambia, and Zimbabwe, Daniel Low-Beer and Rand
Stoneburner identify unique patterns of communication about HIV through
Ugandan social networks. They argue that this reflects important differences
in approaches to HIV taken by national political leadership—to which
we add religious leadership. Specifically, intensive long-term (vertical)
communication about HIV from national leaders in Uganda legitimized and
facilitated widespread local (horizontal) communication among community
members.
A related contribution has focused on how networks structure perceptions
of risk. This has been especially valuable for understanding AIDS-related
behaviors in developing countries, where high levels of uncertainty and limited sources of information force people to use networks to both evaluate
risk and vet information about ways to minimize it. These network effects
can be seen even when controlling for unobserved factors that may affect the
structure and composition of the social networks.
A third stream of research on AIDS and social networks has focused on care
and social support for people living with HIV and AIDS (PLWHA). There
has long been a careful distinction in this literature between emotional care
and physical or task-related support. A divergence can be seen, however,
between the literatures on support in wealthy western settings from support
in poor African or Asian settings. In the first it is largely focused on support for PLWHA. In the second, it is equally focused on PLWHA and their
children. This difference is not surprising. In developed countries, AIDS has
disproportionately affected a childless, gay population with small extended
families and access to at least a partial publicly funded safety net. In contrast,
in most countries in SSA the epidemic was, from the beginning, primarily
heterosexual, and it disproportionately affected young parents embedded in
much larger extended families, who were active in local religious congregations, and living in settings devoid of any publicly funded safety net.
The existing literature on social support networks reflects these contextual differences in the sociodemographic characteristics of PLWHA. In the
US, support networks tend to be constituted by a constellation of activists
and other PLWHAs—though more recent literature notes the fragility of this
type of network where people are increasingly affected by age-related comorbidities. In SSA, both emotional and physical support for PLWHA and their

6

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

dependents is sourced from within the extended family or from religious
congregations, as documented by Jenny Trinitapoli and Alexander Weinreb.
These same networks make arrangements to support survivors—typically
children—when the PLWHA dies.
A fourth stream of research on AIDS and social networks has focused
directly on network composition and characteristics, with implications
for prevention, treatment, and social support. There are two quite different
strands of literature here. The first builds on a long tradition in the social
sciences that seeks to identify the effects of larger structural characteristics on the behavior of individuals (e.g., from Durkheim’s On Suicide)
and on networks. In addition, relevant here is the emerging discourse
in modern medicine, which reframes disease as a biosocial phenomenon
rather than simply an outcome of molecular changes. In both cases, larger
structural characteristics and arrangements act as “distal determinants” of
infection—types of “structural violence”—that affect HIV transmission and
treatment by influencing the size and composition of networks. An example:
high risk behavior, including sexual mobility and IV drug use, can be a
predictable byproduct of social disintegration and residential instability and
landholding patterns, or of discriminatory labor market or incarceration
practices. When network stability is compromised, HIV spreads more easily
through one of two mechanisms: (i) older authority structures (i.e., hierarchical networks) are undermined, freeing youth from community sanctions that
may have protected them by proscribing high levels of sexual mobility and
(ii) new peer-driven (i.e., horizontal networks) behavioral models diffuse
alongside new inequalities in wealth, new patterns of consumption-driven
behavior, and the higher-risk activities associated with both.
CUTTING-EDGE RESEARCH
The first 30 years of research on AIDS and social networks established
the contours of engagement across a number of disciplines. Part of this
groundwork involved rapid improvements in modeling the spread of the
HIV (the virus), the spread and adoption of new information, ideas and
behavior associated with HIV prevention, and of modes of social support
for those infected with HIV or affected by AIDS. There is a notable, but often
overlooked, consequence here: by boosting the amount of methodological
and empirical research on networks, AIDS changed disciplines. Recent
advances in epidemiological models of infection, for example, are rooted
in a network paradigm. Moreover, in demography and sociology, HIV
reinvigorated and accelerated the movement of social networks toward the
disciplinary center.

AIDS and Social Networks

7

One stream of contemporary cutting-edge research—focused on sampling
and response error in networks data—cuts across these disciplinary boundaries. This is connected to a long-standing concern in the networks literature
with data quality. It also builds on the recognition that as important a role as
formal models and simulations have played in many key SNA developments
that have emerged from the AIDS literature, their underlying claims are best
tested empirically.
A characteristic difficulty with AIDS in this regard is that people at greatest
risk of HIV infection (e.g., sex workers and IV drug users) tend to be in “hidden” or “hard-to-reach” populations: they are more mobile and marginalized
and thus are rarely found in household samples. They are also less trusting
of authorities in general and are, therefore, reluctant to be interviewed even
when they are found. The greater the magnitude of each of these problems
during data collection, is the less representative the resulting data are.
To address this sampling problem, researchers have developed alternative
sampling strategies. One is time-location sampling. By allowing researchers
to target recruitment in geographically defined hotspots (e.g., bath houses,
and red-light districts) where risky behaviors are concentrated, time-location
sampling directly addresses longstanding critiques of aspatial modeling
paradigms used in the AIDS literature and empirical social science in
general. Another approach is to use a link-tracing-based design, long used
in collecting data on infectious diseases. One recent adaptation of these
strategies—respondent-driven sampling (RDS)—leverages network-based
referral approaches to approximate random sample characteristics. Developed by Douglas Heckathorn et al., RDS explicitly acknowledges the unequal
and non-random tendency of how individuals form social connections. It
argues that, by following referral chains through a series of iterations,
researchers can reach an equilibrium in which the sample composition is
independent of the characteristics of initially sampled “seeds.” RDS has
important weaknesses, such as the inability to tap into portions of the population that are completely isolated from the seeds—this is less problematic
in time-location sampling—and potentially dramatic fluctuations in sample
characteristics from seemingly minor violations of the model’s assumptions.
Despite these, RDS has become de rigueur for sampling hidden and hard to
reach populations, including those at high risk for contracting HIV.
The second strand of methodological research to have emerged in the past
few years reflects growing anxiety about data quality. Because the smallest
units of analysis in social networks focus on relationships—which necessarily include two individuals—social network data are uniquely situated to
evaluate the validity of self-reported information in ways other types of data
cannot explore. For example, some have used such multiply reported data
to demonstrate how reliably individual partnerships are reported both by

8

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

members of those relationships and even by other individuals not part of
the relationship, but connected (directly or indirectly) to the people who are.
Given that small local changes in networks can have substantial implications
for global connectivity of networks among the full population, researchers
must critically evaluate ways to improve reporting. A variety of technological, passive and active participatory strategies have begun to do exactly that:
some use new types of survey instruments such as relationship histories;
others are focusing more on new types of self-interviewing (eliminating survey error associated with interviewers); still others are using new forms of
observational data, for example, by mapping network connections using cellphone data.
A second stream of cutting-edge research examines a completely different
type of AIDS network: vertical institutional networks extending from donors
to local NGOs involved in the “global governance of AIDS.” There are two
important contextual factors here; the first is global. Since the 1970s, a new
institutional configuration has emerged involving funders, development
professionals, international non-governmental organizations (INGO), and
their local agents (NGO). In more general social science this is a particular
type of “transboundary formation.” The second contextual change is specific
to poorer countries with generalized AIDS epidemics: the massive mobilization of international donors around AIDS over the past decade. The United
States President’s Emergency Plan for AIDS Relief (PEPFAR), for example,
was established in 2004 and committed to spending US$15 billion over its
first 5 years. Its reauthorization in 2008—to run from 2009 to 2013—increased
its budget to a maximum of US$48 billion (though some of those resources
are devoted to malaria and TB). The global fund to fight AIDS, tuberculosis,
and malaria—known simply as the global fund (GF)—was founded in 2002.
By December 2009, the GF had approved proposals totaling US$19.2 billion,
virtually all of it for HIV and TB. Billions more AIDS-designated dollars
have been disbursed by the Bill and Melinda Gates Foundation, by the
World Bank’s Multicountry AIDS Program (MAP), and by many other
bilateral agencies.
In settings with limited access to capital, these new institutional configurations have made available massive amounts of resources and have attracted
considerable attention from international and local NGO entrepreneurs, and
from local technocratic elites. The result has been the emergence of new vertical networks that link these major bilateral organizations and global donors
to: (i) INGOs to whom the actual running of programs is subcontracted, (ii)
national NGOs, and (iii) local NGOs and development committees, including
those run by religious organizations. Recent work by Susan Watkins and Ann

AIDS and Social Networks

9

Swidler on how policies filtered through these networks demonstrate enormous differences between their design at the “top” and their implementation
on the ground.
This body of work further shows how these resources are used as “instruments of patronage” and as opportunities to further other parochial
interests. Notably here, in the case of AIDS, the “institutional isomorphism”
imposed by funders is offset by a reshaping or “translation” process. That
is, policies may be nearly identical in their original form, but as they diffuse
down through a network they are repeatedly modified rather than being
directly reproduced. There is an ironic structural parallel here between, on
the one hand, the mutations that have changed HIV over time, resulting in an
ever-growing array of HIV subtypes and recombinations and, on the other
hand, the way that information is transmitted but constantly reconfigured
as it works its way through the network. Cumulatively, these changes lead
to dramatically different modes of implementation and outcomes on the
ground, some of them much more apposite to local setting.
The ways in which vertical networks both distribute resources and (usually
informally) reshape policy is both interesting and important. First, it points
to a unique type of AIDS network that has come into being for one of two reasons. One is the international commitment to combating AIDS. Another—as
cynical as it sounds—is because the magnitude of resources available to AIDS
has allowed for the full expression of development professionals’ scavenger
instincts, enabling them to use AIDS to perpetuate both their institutions and
their own livelihoods. A second reason the vertical networks are important is
that they hold out the tantalizing promise of integrated analyses of different
types of networks: networks through which viruses and information move;
networks through which social support is provided; and networks through
which policy responses are channeled. We return to this promise below.
KEY ISSUES FOR FUTURE RESEARCH
As work on AIDS and social networks moves into its fourth decade, a number
of areas are primed for important contributions. Some of these reflect recent
advances in the social networks literature that have not yet been translated
into work on AIDS. Others address things the networks literature could productively adapt from AIDS research, and still others reflect perspectives and
tools that are new to both fields.
Perhaps the most important developments in social networks literature
over the past decade center on the (re)introduction of temporal dynamics
into network analytic strategies and the development of statistical models
for handling network data. While each of these have some roots in literature
on HIV/AIDS, the latest developments, associated especially with James

10

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Moody, have yet to be fully embraced by AIDS researchers. The network
positional characteristics that we know influence an individual’s chance
of contracting HIV (e.g., centrality) are constantly changing. For example,
a person cannot contract HIV from their current partners’ future partners,
and s/he cannot transmit HIV to a current partner’s former partners.
Recent modeling developments clearly show that these dynamic patterns
substantially alter networks’ epidemic potential, providing a serious caution
against thinking about networks as fixed structures, even for the purposes
of modeling short-term trends. Parallel temporal dynamics exist within
the extended family structures that underlie social support systems in
developing countries. That is, people choose to maintain ties with certain
cousins, uncles and aunts, not all of them; and the absence of one of these
favored network relationships may both increase the likelihood of making
connections to someone else, but also reduce connections to others in the
family once maintained through the now absent bridging tie.
The introduction of statistical models for static or dynamic networks—
known respectively as p* or exponential random graph models (ERGM) and
stochastic actor-based models (SABM) or SIENA models—has moved the
analysis of networks from a purely descriptive endeavor to one with the
empirical wherewithal to emphasize processes. Developed by Tom Snijders
et al., these models allow for the simultaneous estimation of individual and
network-endogenous effects in observed network patterns. For instance,
while early research relied on single snapshots of a network and inferred
processes of influence from across those networks (e.g., in the dissemination
or adoption of strategies for HIV prevention), SABM are being used to disentangle processes of influence (e.g., people adopting the ideas/behaviors of
those to whom they are connected) from those of selection (e.g., individuals
forming ties to others with whom they hold similar views). New implementations of this approach will be especially beneficial to future efforts
to understand whether and how people’s strategies for avoiding infection
“work.” The dynamic and statistical advances from the modeling literature
on HIV have not yet been fully incorporated into strategies for gathering
the high-quality data we described above as necessary for advancing both
fields. Each stands to benefit substantially from future data collection
efforts that target the gathering of longitudinal data on risk-bearing and
knowledge-sharing networks within the observed populations.
AIDS-salient networks tend to be “multiplex” in nature, that is, they
involve more than one type of relationship at a time. This seemingly simple
insight has been a critical contribution of AIDS literature from early days
when, for example, it became clear that “risk” carrying networks (i.e.,
those involving sexual or needle-sharing contact) can also effectively carry
prevention efforts or can be converted into sources of support and care for

AIDS and Social Networks

11

PLWA. The formal models within SNA literature, however, still tend to focus
on a single tie type at a time (e.g., only on sexual networks or social support
networks). Yet including multiple networks has important implications
for understanding network effects in general. For example, analyses of sex
ties or needle-sharing ties alone underestimate the risk network as a whole
and misestimate how sex ties or drug ties themselves shape individual risk
patterns; tracing risk-relevant ties to the exclusion of other “social” network
ties can misrepresent the full extent of the population at risk; and treating
a social support network (in a developing country) as isomorphic with an
extended family misses crucial non-familial support from friends or local
religious organizations.
Future work should focus more explicitly on the myriad ways in which
these different types of networked relationships can combine to influence
HIV/AIDS, in part by systematically building on this idea of multiplexity.
After all, enough is now known about different types of networks, about
analyzing dynamic processes, and about how to collect high-quality data,
that we can envisage collecting and combining data that enable researchers
to estimate individual and network-endogenous effects, but this time across
multiplex networks. Only this combination of data would allow us to directly
address AIDS-related questions that lie in the social networks. For example:
What is the relationship between the structure of nuclear and extended
family, social and sexual networks, local support networks, and how do
these collectively influence HIV transmission, subsequent treatment and
social support? How do these structures change over time (or how do people
change them over time)? What happens when a best friend—or some other
key node in a friend, family, or social support network—moves or dies? What
happens to sexual networks and informal support networks as access to
antiretroviral drugs, especially the latest fixed-dose combinations, increases?
Important vertical elements should also be included in this new dynamic,
multiplex SNA frame. One would focus on structural determinants of networks themselves: from residential patterns and conditions to the quality
and intensity of connections to kin and nonkin; from involvement in institutions that facilitate new network connections (e.g., schools, churches, and
employment outside the home) to deeply entrenched cultural practices that
limit physical movement and network freedoms (e.g., gender differences in
free movement through public space embodied in purdah). Another vertical
element would extend into dynamic organizational networks. Among these
are local and national NGOs that interact with community and family-based
networks in promoting HIV prevention or organizing social support for those
directly or indirectly affected by AIDS.
A final element is related to spatial dynamics. Understanding HIV’s movement through space has been a goal of researchers since early attempts to

12

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

trace “patient zero” and map truckers’, soldiers’, and other migrants’ routes
in Africa. However, spatial modeling of AIDS networks in general has not
been at the forefront of research agendas. This is unfortunate since there are a
number of interesting spatial puzzles. Examples include: clustering of AIDS
orphans in Malawi that is completely orthogonal to HIV prevalence; clusters of AIDS support networks around particular religious congregations;
irregular diffusion of stories and ideas about AIDS within and across communities. Future research should incorporate each of these contextual layers (e.g., network-informed spatial measures or spatially informed network
measures), in order to look at the intersection of network and spatial processes and, subsequently, to directly address the ways in which they impact
outcomes of interest.
In summary, it is widely known that the social epidemiology of AIDS,
from transmission to its long gestation, is complicated. What has become
increasingly clear over the past decade is that single subsets of networks,
no matter how seemingly well circumscribed, cannot proxy for full systems
if our intention is to fully understand and document the relationship
between social networks and AIDS, whether in terms of HIV transmission,
prevention, or care. Future research should reflect this awareness. AIDS
and social networks should be treated empirically as things that occur
within a temporally dynamic, multilevel, and spatially combustible setting.
Excluding any one of these dimensions, or using instruments that artificially
limit scholars’ attention to a single type or subset of behaviorally specific
networks, or to a single period of time, misses a substantial part of the
relationship between AIDS and social networks. We need a more holistic
approach.
FURTHER READING
Bridging
Youm, Y., & Laumann, E. O. (2002). Social network effects on the transmission of
sexually transmitted diseases. Sexually Transmitted Diseases, 29, 689–97.
Care and Social Support
Trinitapoli, J., & Weinreb, A. (2012). Religion and AIDS in Africa. New York, NY:
Oxford University Press.
Concurrency
Morris, M., & Kretzchmar, M. (1997). Concurrent partnerships and the spread of HIV.
AIDS, 11, 641–48.

AIDS and Social Networks

13

Data Quality
Helleringer, S., Kohler, H.-P., Kalilani-Phiri, L., Mkandawire, J., & Armbruster, B.
(2011). The reliability of sexual partnership histories: Implications for the measurement of partnership concurrency during surveys. AIDS, 25(4), 503–11.
Infection Tracing
Potterat, J. J., Woodhouse, D. E., Muth, S. Q., Rothenberg, R. B., Darrow, W. W., Klovdahl, A. S., & Muth, J. B. (2004). Network dynamism: History and lessons of the
Colorado Springs study. In M. Morris (Ed.), Network epidemiology: A handbook for
survey design and data collection. Oxford, England: Oxford University Press.
Multiplexity
adams, j., Moody, J., & Morris, M. (2013). Sex, drugs, and race: How behaviors differentially contribute to sexually transmitted infection risk network structure. American Journal of Public Health, 103(2), 322–29.
Prevention Turn
Heckathorn, D. D., Broadhead, R. S., Anthony, D. L., & Weakliem, D. L. (1999). AIDS
and social networks: HIV prevention through network mobilization. Sociological
Focus, 32(2), 159–179.
Perception of Risk
Kohler, H.-P., Behrman, J. R., & Watkins, S. C. (2007). Social networks and HIV/AIDS
risk perceptions. Demography, 44(1), 1–33.
Statistical Models
Steglich, C., Snijders, T. A. B., & Pearson, M. (2010). Dynamic networks and behavior:
Separating selection from influence. Sociological Methodology, 40(1), 329–93.
Temporality
Moody, J. (2009). Network dynamics. In P. Hedstrom & P. S. Bearman (Eds.), The
Oxford handbook of analytical sociology. New York, NY: Oxford University Press.
“Vertical” Networks
Watkins, S. C., Swidler, A., & Hannan, T. (2012). Outsourcing social transformation:
Development NGOs as organizations. Annual Review of Sociology, 38, 285–315.

14

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

ALEXANDER WEINREB, jimi adams, JENNY TRINITAPOLI
SHORT BIOGRAPHY
Alexander Weinreb (UT-Austin), jimi adams (American University), and
Jenny Trinitapoli (Penn State) are social demographers who think about networks not only in relation to AIDS but also in relation to a variety of other
topics including religion, family structures, political processes and outcomes,
the development of disciplinary fields, and social cohesion generally. They
have lived and worked in multiple countries but as field researchers have
largely focused on collecting and analyzing network-based data in Kenya
and Malawi. Some of their work on these subjects appears in Religion and
AIDS in Africa (OUP 2012) and scholarly journals such as American Sociological Review, American Journal of Public Health, Social Networks, Social
Science and Medicine, Population and Development Review, Demographic
Research, and Field Methods.
RELATED ESSAYS
Inefficiencies in Health Care Provision (Economics), James F. Burgess et al.
Problems Attract Problems: A Network Perspective on Mental Disorders
(Psychology), Angélique Cramer and Denny Borsboom
Self-Fulfilling Prophesies, Placebo Effects, and the Social-Psychological
Creation of Reality (Sociology), Alia Crum and Damon J. Phillips
The Development of Social Trust (Psychology), Vikram K. Jaswal and Marissa
B. Drell
Emerging Trends in Social Network Analysis of Terrorism and Counterterrorism (Sociology), David Knoke
Network Research Experiments (Methods), Allen L. Linton and Betsy Sinclair
Immigrant Health Paradox (Sociology), Kyriakos S. Markides and Sunshine
Rote
Rationing of Health Care (Sociology), David Mechanic
The Role of School-Related Peers and Social Networks in Human Development (Psychology), Chandra Muller
Health and Social Inequality (Sociology), Bernice A. Pescosolido
Social Relationships and Health in Older Adulthood (Psychology), Theodore
F. Robles and Josephine A. Menkin
The Role of Cultural, Social, and Psychological Factors in Disease and Illness
(Sociology), Robert A. Scott
Incarceration and Health (Sociology), Christopher Wildeman


AIDS and Social Networks
ALEXANDER WEINREB, jimi adams, and JENNY TRINITAPOLI

Abstract
During the past 30 years, research on the global AIDS pandemic and on social
networks has coevolved. Insights from social networks literature have advanced
our understandings of AIDS; simultaneously, key empirical insights from the AIDS
literature have furthered the development of social network research—especially
methodologically. We elaborate on this reciprocal relationship, identifying some
of the key developments and future directions for research on AIDS and on
social networks generally. From existing literatures, we discuss how (i) social
networks analysis was central to early attempts to understand the spread of HIV
through sexual and needle-sharing relationships; (ii) subsequent prevention efforts
leveraged similar insights to different ends; (iii) social networks have been crucial
in understanding patterns of care for people living with HIV/AIDS; and (iv)
the structural composition of networks across international, organizational, and
individual levels highlights the epidemic’s global implications in ways that extend
far beyond epidemiology. We contend that future research must integrate recent
developments from both fields in order advance understandings. Among these, we
identify as most promising: (i) a move from static modeling approaches toward
research emphasizing the dynamic properties of networks; (ii) a shifting focus from
single networks in isolation (e.g., sexual transmission networks) to the analysis of
multiplex networks (i.e., those involving multiple relationship types represented
simultaneously); and (iii) an acknowledgment—conceptual and methodological—of
the “vertical” embeddedness of networks. Continued advances in this area will
require the gathering of high quality social network data specifically designed to
address such questions.

INTRODUCTION
The following facts are widely circulating but bear repetition. AIDS is
the pandemic of the era. Even with uncertainty about when HIV, the
virus that causes AIDS, first emerged, since it first came to broad public
attention in 1982, it has been responsible for more than 25 million deaths. A
disproportionate number of these deaths have occurred in the sub-Saharan
African (SSA) “AIDS belt” that extends from southern into eastern African
countries, and that includes populations where more than 20% of adults
are HIV-positive. Other significant AIDS clusters have formed around gay
Emerging Trends in the Social and Behavioral Sciences. Edited by Robert Scott and Stephen Kosslyn.
© 2015 John Wiley & Sons, Inc. ISBN 978-1-118-90077-2.

1

2

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

communities, IV drug-users and prostitutes, in both more and less developed
countries. Globally, about 30 million people are infected with HIV.
While AIDS research has been engaged with the social network paradigm
since the 1980s, the form that engagement has taken across disciplines has
varied dramatically. Epidemiologists have drawn on network approaches to
model the spread of HIV; demographers and public health researchers have
explored how new information about prevention spreads and whether, if at
all, this affects subsequent HIV incidence. Sociologists and economists have
described the social support mechanisms used to care for the sick and their
survivors; and public policy and international aid scholars have examined
the networks among international donors, local nongovernmental organizations (NGOs), and religious organizations that have emerged in response
to AIDS. Yet across these disciplines, advances in social networks analysis
(SNA) have also been accelerated by this focus on AIDS. In a sense, SNA and
AIDS research have “grown up” in parallel, each making important contributions to the other.
In this essay, we outline the foundational and contemporary claims of these
literatures at the nexus of AIDS and social networks, and we point to promising avenues for future research. The literature on “AIDS and social networks”
is sometimes more focused on HIV than AIDS and throughout we follow
scholarly convention using “HIV” when looking at transmission and prevention, and “AIDS” when examining treatment, social support, or policy
in general. It should be clear that we are not restricting ourselves to a particular discipline. Rather, we see utility in bridging—as network theorists
would describe it—different disciplines. This orientation reflects our collective professional experience. Two of us are social demographers whose primary area of research over the past few years has been AIDS in Africa; the
third is a social networks researcher with interests in Africa and the United
States. All of us, however, have considerable experience collaborating with
scholars trained in sociology, economics, epidemiology, public health, and
anthropology—each of which has developed its own approach(es) to understanding and analyzing networks and AIDS.
Notwithstanding our commitment to bridging disciplines, we do
not—indeed, cannot—include everything. Most notably, we exclude
much research on AIDS where an underlying relational focus that intimates a
network approach is not reflected in actual analysis of networks. Examples
include much of the research on: how religion influences people’s sexual
behavior or what people know about AIDS; the consequences of marriage
and divorce both for individual risk and population-level epidemics; how
trust in partnerships influences condom use; relationship quality and condoms; and the relationships that structure caregiving obligations in the wake
of AIDS. The increasing number of papers on topics like these reflects the

AIDS and Social Networks

3

spread of a relational paradigm through the social and behavioral sciences,
especially in social and behavioral research on health. However, they are not
social networks papers in the normal sense of that term.
FOUNDATIONAL RESEARCH
We identify four main streams of foundational research on AIDS and social
networks.
Epidemiologists were the vanguard: the first to apply social networks
approaches to AIDS. From their earliest attempts to identify “patient zero”
in the United States, their goal has been to model the spread of HIV. Their
underlying proposition and finding: the structure of a social network affects
both who you meet and what you do with that person once you’ve met them.
For epidemiologists, this do refers primarily to the behaviors that directly
influence the likelihood of transmission—in the case of HIV, through one
of two principal transmission modes: sex and intravenous drug-use (IDU).
Networks profoundly affect whether, under what conditions, and how
quickly people proceed from meeting each other to having sex or using
drugs; networks further influence the type of sex and kind of drug use—both
of which matter for transmission. Concurrency provides one salient example
of how structured patterns of behaviors can alter the trajectories of an
epidemic above and beyond the constitutive individuals and individual
behaviors. In a series of simulations, Martina Morris et al. have shown that,
at the population level, the arrangement of partnerships is more salient
than their mere number. Where even a moderate number of partnerships
overlap in time (i.e., are “concurrent”), epidemics are much larger than
in populations with the exact same number of partnerships, differently
arranged to minimize time overlaps (such as in serial monogamy).
In offering better explanations for variation in the shape and magnitude of
AIDS epidemics, network approaches have forced an important theoretical
turn, with implications that are much broader than the particular case of
HIV: they demand that we step back from ideas about “risk” that focus
singularly on individuals and their characteristics (both socio-demographic
and behavioral) and, instead, emphasize the salience of relationships and their
characteristics. For example, the simple statement—“even monogamous
women had very large sexual networks,” written by Kay Johnson et al. about
their research in Peru—illustrates this theoretical turn. The idea is virtually
incomprehensible where we conceive of, or model, risk as an individual
property—an implicit assumption in much research on health behavior. It
makes perfect sense, however, when we think about an HIV infection as
arising from a series of conditional probabilities associated with one’s own
partners and one’s partners’ partners.

4

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Other foundational findings from SNA research on the spread of HIV
also bear broader theoretical and practical implications. One concerns the
important role of “bridgers” or “bridge populations”—those individuals
or groups that link otherwise isolated components of a network. Another
challenges traditional “risk factor” models of infection. In conceptualizing
the joint effects of multiple risks, early research assumed that each additional risk factor provided an additive effect on infection. In reality, however,
risky behaviors often co-occur, and although the joint effects are difficult
to disentangle empirically, scholars now agree that their combination is
more often multiplicative or exponential rather than linear. The underlying
argument is simple but important: simultaneity raises issues about the very
meaning of risk and how to target it. For example, it may be less profitable
to think of risky behaviors as measurable phenomena than to think of “risk”
more broadly—as a fuzzy, latent construct that captures a wide range of
associated behaviors.
A second stream of foundational social networks research focused on
HIV prevention. This followed closely on the heels of the literature on
transmission. In sub-Saharan Africa, the epidemic’s epicenter, this approach
essentially cut-and-pasted models of diffusion developed in the mid-1990s
to track the spread of information and ideas about contraception and family
limitation. This new prevention literature stemmed from a theoretical and
explanatory literature on the spread of innovative ideas through networks
that had roots in early work on communications, anthropological work
on social change, and demography’s paradigm-shifting European Fertility
Project. Coincidentally, much of this early work was developed in the same
areas of Central Africa that would later have the heaviest concentrations
of HIV. The prevention literature profited from the rapid publication of a
series of papers in the mid-1990s that clarified conceptually muddy terms
in two important ways. First, these papers made the important conceptual
distinction between two types of network effects—social learning (the flow
of information about how to reduce risk) and social influence (moral and
political evaluation of those risk-reduction behaviors). Second, in work
associated with Hans-Peter Kohler et al., researchers demonstrated that they
could leverage measures of network density to distinguish one from the
other.
The literature on networks and HIV prevention made a few other important contributions. First was its underlying recognition of a structural
parallel between networks that increase transmission risk and networks that
facilitate the flow of information that may ultimately prevent infection. This
structural parallel helps explain the “first in, first out” rule for understanding
the waves of HIV prevalence by social class in developing countries with the
worst generalized epidemics: in these settings, the earliest rise and fall in HIV

AIDS and Social Networks

5

prevalence was associated with the wealthiest, then the middle class, then
the poor—note that a somewhat different pattern has persisted in western
countries. The insight that HIV infection and prevention networks run in
parallel undergirds a key policy effort: “peer-driven intervention,” which is
more effective than standard street-based outreach for preventing infection.
It also helps explain the highly variable record of African countries in combating AIDS. For example, in their comparative study of Uganda, Kenya,
Tanzania, Malawi, Zambia, and Zimbabwe, Daniel Low-Beer and Rand
Stoneburner identify unique patterns of communication about HIV through
Ugandan social networks. They argue that this reflects important differences
in approaches to HIV taken by national political leadership—to which
we add religious leadership. Specifically, intensive long-term (vertical)
communication about HIV from national leaders in Uganda legitimized and
facilitated widespread local (horizontal) communication among community
members.
A related contribution has focused on how networks structure perceptions
of risk. This has been especially valuable for understanding AIDS-related
behaviors in developing countries, where high levels of uncertainty and limited sources of information force people to use networks to both evaluate
risk and vet information about ways to minimize it. These network effects
can be seen even when controlling for unobserved factors that may affect the
structure and composition of the social networks.
A third stream of research on AIDS and social networks has focused on care
and social support for people living with HIV and AIDS (PLWHA). There
has long been a careful distinction in this literature between emotional care
and physical or task-related support. A divergence can be seen, however,
between the literatures on support in wealthy western settings from support
in poor African or Asian settings. In the first it is largely focused on support for PLWHA. In the second, it is equally focused on PLWHA and their
children. This difference is not surprising. In developed countries, AIDS has
disproportionately affected a childless, gay population with small extended
families and access to at least a partial publicly funded safety net. In contrast,
in most countries in SSA the epidemic was, from the beginning, primarily
heterosexual, and it disproportionately affected young parents embedded in
much larger extended families, who were active in local religious congregations, and living in settings devoid of any publicly funded safety net.
The existing literature on social support networks reflects these contextual differences in the sociodemographic characteristics of PLWHA. In the
US, support networks tend to be constituted by a constellation of activists
and other PLWHAs—though more recent literature notes the fragility of this
type of network where people are increasingly affected by age-related comorbidities. In SSA, both emotional and physical support for PLWHA and their

6

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

dependents is sourced from within the extended family or from religious
congregations, as documented by Jenny Trinitapoli and Alexander Weinreb.
These same networks make arrangements to support survivors—typically
children—when the PLWHA dies.
A fourth stream of research on AIDS and social networks has focused
directly on network composition and characteristics, with implications
for prevention, treatment, and social support. There are two quite different
strands of literature here. The first builds on a long tradition in the social
sciences that seeks to identify the effects of larger structural characteristics on the behavior of individuals (e.g., from Durkheim’s On Suicide)
and on networks. In addition, relevant here is the emerging discourse
in modern medicine, which reframes disease as a biosocial phenomenon
rather than simply an outcome of molecular changes. In both cases, larger
structural characteristics and arrangements act as “distal determinants” of
infection—types of “structural violence”—that affect HIV transmission and
treatment by influencing the size and composition of networks. An example:
high risk behavior, including sexual mobility and IV drug use, can be a
predictable byproduct of social disintegration and residential instability and
landholding patterns, or of discriminatory labor market or incarceration
practices. When network stability is compromised, HIV spreads more easily
through one of two mechanisms: (i) older authority structures (i.e., hierarchical networks) are undermined, freeing youth from community sanctions that
may have protected them by proscribing high levels of sexual mobility and
(ii) new peer-driven (i.e., horizontal networks) behavioral models diffuse
alongside new inequalities in wealth, new patterns of consumption-driven
behavior, and the higher-risk activities associated with both.
CUTTING-EDGE RESEARCH
The first 30 years of research on AIDS and social networks established
the contours of engagement across a number of disciplines. Part of this
groundwork involved rapid improvements in modeling the spread of the
HIV (the virus), the spread and adoption of new information, ideas and
behavior associated with HIV prevention, and of modes of social support
for those infected with HIV or affected by AIDS. There is a notable, but often
overlooked, consequence here: by boosting the amount of methodological
and empirical research on networks, AIDS changed disciplines. Recent
advances in epidemiological models of infection, for example, are rooted
in a network paradigm. Moreover, in demography and sociology, HIV
reinvigorated and accelerated the movement of social networks toward the
disciplinary center.

AIDS and Social Networks

7

One stream of contemporary cutting-edge research—focused on sampling
and response error in networks data—cuts across these disciplinary boundaries. This is connected to a long-standing concern in the networks literature
with data quality. It also builds on the recognition that as important a role as
formal models and simulations have played in many key SNA developments
that have emerged from the AIDS literature, their underlying claims are best
tested empirically.
A characteristic difficulty with AIDS in this regard is that people at greatest
risk of HIV infection (e.g., sex workers and IV drug users) tend to be in “hidden” or “hard-to-reach” populations: they are more mobile and marginalized
and thus are rarely found in household samples. They are also less trusting
of authorities in general and are, therefore, reluctant to be interviewed even
when they are found. The greater the magnitude of each of these problems
during data collection, is the less representative the resulting data are.
To address this sampling problem, researchers have developed alternative
sampling strategies. One is time-location sampling. By allowing researchers
to target recruitment in geographically defined hotspots (e.g., bath houses,
and red-light districts) where risky behaviors are concentrated, time-location
sampling directly addresses longstanding critiques of aspatial modeling
paradigms used in the AIDS literature and empirical social science in
general. Another approach is to use a link-tracing-based design, long used
in collecting data on infectious diseases. One recent adaptation of these
strategies—respondent-driven sampling (RDS)—leverages network-based
referral approaches to approximate random sample characteristics. Developed by Douglas Heckathorn et al., RDS explicitly acknowledges the unequal
and non-random tendency of how individuals form social connections. It
argues that, by following referral chains through a series of iterations,
researchers can reach an equilibrium in which the sample composition is
independent of the characteristics of initially sampled “seeds.” RDS has
important weaknesses, such as the inability to tap into portions of the population that are completely isolated from the seeds—this is less problematic
in time-location sampling—and potentially dramatic fluctuations in sample
characteristics from seemingly minor violations of the model’s assumptions.
Despite these, RDS has become de rigueur for sampling hidden and hard to
reach populations, including those at high risk for contracting HIV.
The second strand of methodological research to have emerged in the past
few years reflects growing anxiety about data quality. Because the smallest
units of analysis in social networks focus on relationships—which necessarily include two individuals—social network data are uniquely situated to
evaluate the validity of self-reported information in ways other types of data
cannot explore. For example, some have used such multiply reported data
to demonstrate how reliably individual partnerships are reported both by

8

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

members of those relationships and even by other individuals not part of
the relationship, but connected (directly or indirectly) to the people who are.
Given that small local changes in networks can have substantial implications
for global connectivity of networks among the full population, researchers
must critically evaluate ways to improve reporting. A variety of technological, passive and active participatory strategies have begun to do exactly that:
some use new types of survey instruments such as relationship histories;
others are focusing more on new types of self-interviewing (eliminating survey error associated with interviewers); still others are using new forms of
observational data, for example, by mapping network connections using cellphone data.
A second stream of cutting-edge research examines a completely different
type of AIDS network: vertical institutional networks extending from donors
to local NGOs involved in the “global governance of AIDS.” There are two
important contextual factors here; the first is global. Since the 1970s, a new
institutional configuration has emerged involving funders, development
professionals, international non-governmental organizations (INGO), and
their local agents (NGO). In more general social science this is a particular
type of “transboundary formation.” The second contextual change is specific
to poorer countries with generalized AIDS epidemics: the massive mobilization of international donors around AIDS over the past decade. The United
States President’s Emergency Plan for AIDS Relief (PEPFAR), for example,
was established in 2004 and committed to spending US$15 billion over its
first 5 years. Its reauthorization in 2008—to run from 2009 to 2013—increased
its budget to a maximum of US$48 billion (though some of those resources
are devoted to malaria and TB). The global fund to fight AIDS, tuberculosis,
and malaria—known simply as the global fund (GF)—was founded in 2002.
By December 2009, the GF had approved proposals totaling US$19.2 billion,
virtually all of it for HIV and TB. Billions more AIDS-designated dollars
have been disbursed by the Bill and Melinda Gates Foundation, by the
World Bank’s Multicountry AIDS Program (MAP), and by many other
bilateral agencies.
In settings with limited access to capital, these new institutional configurations have made available massive amounts of resources and have attracted
considerable attention from international and local NGO entrepreneurs, and
from local technocratic elites. The result has been the emergence of new vertical networks that link these major bilateral organizations and global donors
to: (i) INGOs to whom the actual running of programs is subcontracted, (ii)
national NGOs, and (iii) local NGOs and development committees, including
those run by religious organizations. Recent work by Susan Watkins and Ann

AIDS and Social Networks

9

Swidler on how policies filtered through these networks demonstrate enormous differences between their design at the “top” and their implementation
on the ground.
This body of work further shows how these resources are used as “instruments of patronage” and as opportunities to further other parochial
interests. Notably here, in the case of AIDS, the “institutional isomorphism”
imposed by funders is offset by a reshaping or “translation” process. That
is, policies may be nearly identical in their original form, but as they diffuse
down through a network they are repeatedly modified rather than being
directly reproduced. There is an ironic structural parallel here between, on
the one hand, the mutations that have changed HIV over time, resulting in an
ever-growing array of HIV subtypes and recombinations and, on the other
hand, the way that information is transmitted but constantly reconfigured
as it works its way through the network. Cumulatively, these changes lead
to dramatically different modes of implementation and outcomes on the
ground, some of them much more apposite to local setting.
The ways in which vertical networks both distribute resources and (usually
informally) reshape policy is both interesting and important. First, it points
to a unique type of AIDS network that has come into being for one of two reasons. One is the international commitment to combating AIDS. Another—as
cynical as it sounds—is because the magnitude of resources available to AIDS
has allowed for the full expression of development professionals’ scavenger
instincts, enabling them to use AIDS to perpetuate both their institutions and
their own livelihoods. A second reason the vertical networks are important is
that they hold out the tantalizing promise of integrated analyses of different
types of networks: networks through which viruses and information move;
networks through which social support is provided; and networks through
which policy responses are channeled. We return to this promise below.
KEY ISSUES FOR FUTURE RESEARCH
As work on AIDS and social networks moves into its fourth decade, a number
of areas are primed for important contributions. Some of these reflect recent
advances in the social networks literature that have not yet been translated
into work on AIDS. Others address things the networks literature could productively adapt from AIDS research, and still others reflect perspectives and
tools that are new to both fields.
Perhaps the most important developments in social networks literature
over the past decade center on the (re)introduction of temporal dynamics
into network analytic strategies and the development of statistical models
for handling network data. While each of these have some roots in literature
on HIV/AIDS, the latest developments, associated especially with James

10

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Moody, have yet to be fully embraced by AIDS researchers. The network
positional characteristics that we know influence an individual’s chance
of contracting HIV (e.g., centrality) are constantly changing. For example,
a person cannot contract HIV from their current partners’ future partners,
and s/he cannot transmit HIV to a current partner’s former partners.
Recent modeling developments clearly show that these dynamic patterns
substantially alter networks’ epidemic potential, providing a serious caution
against thinking about networks as fixed structures, even for the purposes
of modeling short-term trends. Parallel temporal dynamics exist within
the extended family structures that underlie social support systems in
developing countries. That is, people choose to maintain ties with certain
cousins, uncles and aunts, not all of them; and the absence of one of these
favored network relationships may both increase the likelihood of making
connections to someone else, but also reduce connections to others in the
family once maintained through the now absent bridging tie.
The introduction of statistical models for static or dynamic networks—
known respectively as p* or exponential random graph models (ERGM) and
stochastic actor-based models (SABM) or SIENA models—has moved the
analysis of networks from a purely descriptive endeavor to one with the
empirical wherewithal to emphasize processes. Developed by Tom Snijders
et al., these models allow for the simultaneous estimation of individual and
network-endogenous effects in observed network patterns. For instance,
while early research relied on single snapshots of a network and inferred
processes of influence from across those networks (e.g., in the dissemination
or adoption of strategies for HIV prevention), SABM are being used to disentangle processes of influence (e.g., people adopting the ideas/behaviors of
those to whom they are connected) from those of selection (e.g., individuals
forming ties to others with whom they hold similar views). New implementations of this approach will be especially beneficial to future efforts
to understand whether and how people’s strategies for avoiding infection
“work.” The dynamic and statistical advances from the modeling literature
on HIV have not yet been fully incorporated into strategies for gathering
the high-quality data we described above as necessary for advancing both
fields. Each stands to benefit substantially from future data collection
efforts that target the gathering of longitudinal data on risk-bearing and
knowledge-sharing networks within the observed populations.
AIDS-salient networks tend to be “multiplex” in nature, that is, they
involve more than one type of relationship at a time. This seemingly simple
insight has been a critical contribution of AIDS literature from early days
when, for example, it became clear that “risk” carrying networks (i.e.,
those involving sexual or needle-sharing contact) can also effectively carry
prevention efforts or can be converted into sources of support and care for

AIDS and Social Networks

11

PLWA. The formal models within SNA literature, however, still tend to focus
on a single tie type at a time (e.g., only on sexual networks or social support
networks). Yet including multiple networks has important implications
for understanding network effects in general. For example, analyses of sex
ties or needle-sharing ties alone underestimate the risk network as a whole
and misestimate how sex ties or drug ties themselves shape individual risk
patterns; tracing risk-relevant ties to the exclusion of other “social” network
ties can misrepresent the full extent of the population at risk; and treating
a social support network (in a developing country) as isomorphic with an
extended family misses crucial non-familial support from friends or local
religious organizations.
Future work should focus more explicitly on the myriad ways in which
these different types of networked relationships can combine to influence
HIV/AIDS, in part by systematically building on this idea of multiplexity.
After all, enough is now known about different types of networks, about
analyzing dynamic processes, and about how to collect high-quality data,
that we can envisage collecting and combining data that enable researchers
to estimate individual and network-endogenous effects, but this time across
multiplex networks. Only this combination of data would allow us to directly
address AIDS-related questions that lie in the social networks. For example:
What is the relationship between the structure of nuclear and extended
family, social and sexual networks, local support networks, and how do
these collectively influence HIV transmission, subsequent treatment and
social support? How do these structures change over time (or how do people
change them over time)? What happens when a best friend—or some other
key node in a friend, family, or social support network—moves or dies? What
happens to sexual networks and informal support networks as access to
antiretroviral drugs, especially the latest fixed-dose combinations, increases?
Important vertical elements should also be included in this new dynamic,
multiplex SNA frame. One would focus on structural determinants of networks themselves: from residential patterns and conditions to the quality
and intensity of connections to kin and nonkin; from involvement in institutions that facilitate new network connections (e.g., schools, churches, and
employment outside the home) to deeply entrenched cultural practices that
limit physical movement and network freedoms (e.g., gender differences in
free movement through public space embodied in purdah). Another vertical
element would extend into dynamic organizational networks. Among these
are local and national NGOs that interact with community and family-based
networks in promoting HIV prevention or organizing social support for those
directly or indirectly affected by AIDS.
A final element is related to spatial dynamics. Understanding HIV’s movement through space has been a goal of researchers since early attempts to

12

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

trace “patient zero” and map truckers’, soldiers’, and other migrants’ routes
in Africa. However, spatial modeling of AIDS networks in general has not
been at the forefront of research agendas. This is unfortunate since there are a
number of interesting spatial puzzles. Examples include: clustering of AIDS
orphans in Malawi that is completely orthogonal to HIV prevalence; clusters of AIDS support networks around particular religious congregations;
irregular diffusion of stories and ideas about AIDS within and across communities. Future research should incorporate each of these contextual layers (e.g., network-informed spatial measures or spatially informed network
measures), in order to look at the intersection of network and spatial processes and, subsequently, to directly address the ways in which they impact
outcomes of interest.
In summary, it is widely known that the social epidemiology of AIDS,
from transmission to its long gestation, is complicated. What has become
increasingly clear over the past decade is that single subsets of networks,
no matter how seemingly well circumscribed, cannot proxy for full systems
if our intention is to fully understand and document the relationship
between social networks and AIDS, whether in terms of HIV transmission,
prevention, or care. Future research should reflect this awareness. AIDS
and social networks should be treated empirically as things that occur
within a temporally dynamic, multilevel, and spatially combustible setting.
Excluding any one of these dimensions, or using instruments that artificially
limit scholars’ attention to a single type or subset of behaviorally specific
networks, or to a single period of time, misses a substantial part of the
relationship between AIDS and social networks. We need a more holistic
approach.
FURTHER READING
Bridging
Youm, Y., & Laumann, E. O. (2002). Social network effects on the transmission of
sexually transmitted diseases. Sexually Transmitted Diseases, 29, 689–97.
Care and Social Support
Trinitapoli, J., & Weinreb, A. (2012). Religion and AIDS in Africa. New York, NY:
Oxford University Press.
Concurrency
Morris, M., & Kretzchmar, M. (1997). Concurrent partnerships and the spread of HIV.
AIDS, 11, 641–48.

AIDS and Social Networks

13

Data Quality
Helleringer, S., Kohler, H.-P., Kalilani-Phiri, L., Mkandawire, J., & Armbruster, B.
(2011). The reliability of sexual partnership histories: Implications for the measurement of partnership concurrency during surveys. AIDS, 25(4), 503–11.
Infection Tracing
Potterat, J. J., Woodhouse, D. E., Muth, S. Q., Rothenberg, R. B., Darrow, W. W., Klovdahl, A. S., & Muth, J. B. (2004). Network dynamism: History and lessons of the
Colorado Springs study. In M. Morris (Ed.), Network epidemiology: A handbook for
survey design and data collection. Oxford, England: Oxford University Press.
Multiplexity
adams, j., Moody, J., & Morris, M. (2013). Sex, drugs, and race: How behaviors differentially contribute to sexually transmitted infection risk network structure. American Journal of Public Health, 103(2), 322–29.
Prevention Turn
Heckathorn, D. D., Broadhead, R. S., Anthony, D. L., & Weakliem, D. L. (1999). AIDS
and social networks: HIV prevention through network mobilization. Sociological
Focus, 32(2), 159–179.
Perception of Risk
Kohler, H.-P., Behrman, J. R., & Watkins, S. C. (2007). Social networks and HIV/AIDS
risk perceptions. Demography, 44(1), 1–33.
Statistical Models
Steglich, C., Snijders, T. A. B., & Pearson, M. (2010). Dynamic networks and behavior:
Separating selection from influence. Sociological Methodology, 40(1), 329–93.
Temporality
Moody, J. (2009). Network dynamics. In P. Hedstrom & P. S. Bearman (Eds.), The
Oxford handbook of analytical sociology. New York, NY: Oxford University Press.
“Vertical” Networks
Watkins, S. C., Swidler, A., & Hannan, T. (2012). Outsourcing social transformation:
Development NGOs as organizations. Annual Review of Sociology, 38, 285–315.

14

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

ALEXANDER WEINREB, jimi adams, JENNY TRINITAPOLI
SHORT BIOGRAPHY
Alexander Weinreb (UT-Austin), jimi adams (American University), and
Jenny Trinitapoli (Penn State) are social demographers who think about networks not only in relation to AIDS but also in relation to a variety of other
topics including religion, family structures, political processes and outcomes,
the development of disciplinary fields, and social cohesion generally. They
have lived and worked in multiple countries but as field researchers have
largely focused on collecting and analyzing network-based data in Kenya
and Malawi. Some of their work on these subjects appears in Religion and
AIDS in Africa (OUP 2012) and scholarly journals such as American Sociological Review, American Journal of Public Health, Social Networks, Social
Science and Medicine, Population and Development Review, Demographic
Research, and Field Methods.
RELATED ESSAYS
Inefficiencies in Health Care Provision (Economics), James F. Burgess et al.
Problems Attract Problems: A Network Perspective on Mental Disorders
(Psychology), Angélique Cramer and Denny Borsboom
Self-Fulfilling Prophesies, Placebo Effects, and the Social-Psychological
Creation of Reality (Sociology), Alia Crum and Damon J. Phillips
The Development of Social Trust (Psychology), Vikram K. Jaswal and Marissa
B. Drell
Emerging Trends in Social Network Analysis of Terrorism and Counterterrorism (Sociology), David Knoke
Network Research Experiments (Methods), Allen L. Linton and Betsy Sinclair
Immigrant Health Paradox (Sociology), Kyriakos S. Markides and Sunshine
Rote
Rationing of Health Care (Sociology), David Mechanic
The Role of School-Related Peers and Social Networks in Human Development (Psychology), Chandra Muller
Health and Social Inequality (Sociology), Bernice A. Pescosolido
Social Relationships and Health in Older Adulthood (Psychology), Theodore
F. Robles and Josephine A. Menkin
The Role of Cultural, Social, and Psychological Factors in Disease and Illness
(Sociology), Robert A. Scott
Incarceration and Health (Sociology), Christopher Wildeman



Media

AIDS and Social Networks