University of Southern California Dissertations and Theses

HIV/AIDS preventive behavior in Botswana: Trends and determinants at the turn of the 21st century

(USC Thesis Other)

HIV/AIDS preventive behavior in Botswana: Trends and determinants at the turn of the 21st century

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content HIV/AIDS PREVENTIVE BEHAVIOR IN BOTSWANA: TRENDS AND
DETERMINANTS AT THE TURN OF THE 21st CENTURY
by
Boga Fidzani
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(SOCIOLOGY)
August 2003
Copyright 2003 Boga Fidzani
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3116698
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy
submitted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
®
UMI
UMI Microform 3116698
Copyright 2004 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFORNIA 90007
This dissertation, written by
fbo£f/V P / b 1
under the direction of hM Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
Dean of Graduate Studies
Date ...^8HBta a 12r a > 2003
DISSERTATION CO! : t t e e
Chairperson X
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Dedication
To my son Wedu, member of a generation whose Life
Expectancy is staked at as low as 56 years at the turn
of the 21st Century.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
iii
Acknowledgements
I would like to thank University of Southern
California Professor Emeritus David Heer for
encouraging me to enroll in the Sociology doctoral
program at USC, and for assisting me to obtain the
William and Flora Hewlett Foundation Research
Assistantship through the USC Population Research
Laboratory. He was also the first person to moot the
idea of an AIDS-related topic for my dissertation,
which led to this thesis. I am also greatly indebted
to the USC Graduate School for awarding me the Final
Year Dissertation Fellowship, which enabled me to
complete this project.
Many people contributed to the success of this
project. I would like to acknowledge the contribution
of my advisor Professor Angela James, her unwavering
support and guidance as well as for instilling in me a
great sense of urgency to complete my studies on time.
Other members of my committee Professors Edward
Ransford and Merril Silverstein were also instrumental
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in steering this project, providing invaluable
insights, going through long drafts at very short
notice. The formative stages also benefited from
contributions of Professors Dowell Myers and Kelly
Musick.
The Government of Botswana, specifically the
Central Statistics Office (CSO), allowed me to use
data from the Botswana AIDS Impact Survey (BAIS). My
colleagues at the Ministry of Agriculture, Ms Bonnake
Tsimako and Mr. Michael Manowe were instrumental in
securing the data for me. Ms Toziba Botana from CSO
was very forthcoming in data sharing, which smoothed
up the analysis process.
Gagolelwe Gasefete, CSO Librarian, was very
helpful in securing critical literature at short
notice. My niece, Beauty Nthele, was always at hand
to photocopy items identified at the University of
Botswana's Botswana Collections section. My sister-
in-law, Babakisi Fidzani, introduced me to the highly
modernized University of Botswana Library system.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V
Stephen Mpofu kindly borrowed books on my behalf from
the Oregon State University's library. David Tshere
and Goabaone Raseto accommodated me on the numerous
occasions that I returned to USC at various stages of
this project. It would also be a great omission not
to mention the day-to-day support I got from Batswana
in Los Angeles, California, and Corvallis, Oregon.
Special thanks to my friend Ms Kris Marsh; she rescued
me by correcting the final draft in order to satisfy
stringent Graduate School requirements after my
departure from USA.
Last but far from least, I am indebted to my
wife, Lily, for her unwavering support. She was not
only a pillar of strength but also a source of
immeasurable inspiration. That she could juggle the
challenging roles of student, wife, new mother and
editor so graciously propelled me to complete this
program in the record time that I did.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
VI
TABLE OF CONTENTS
Dedication ii
Acknowledgements iii
List of Tables x
List of Figures xii
Abstract xv
CHAPTER 1: INTRODUCTION 1
1.0 Background 1
1.1 The Research Problem 5
1.2 Justification of the Study 8
1.3 Objectives of the Study 11
1.4 Conceptual Framework 12
1.4.1 Perceived Susceptibility 15
1.4.2 Perceived Severity 16
1.4.3 Perceived Benefits versus Perceived
Costs of Preventive Action 17
1.4.4 Perceived Self-Efficacy 19
1.4.5 Control Variables 20
1.4.6 Preventive Health Behavior versus Health-
Protective Behavior 21
1.4.7 Critiques of the Health Belief Model 24
1.5 Preliminary Hypotheses of the Study 27
CHAPTER 2: METHODOLOGY 2 9
2.1 Data 2 9
2.1.1 Trends in Preventive Health Behavior 31
2.1.2 Modeling Preventive Health Behavior 32
2.1.2.1 The Dependent Variables 33
2.1.2. 2. Explanatory Variables 37
2.1.2.2.1 Perceived Susceptibility 38
2.1.2. 2. 2 Perceived Severity 39
2.1.2.2.3 Perceived Costs/Barriers 40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
v ii
2.1.2. 2.4 Perceived Benefits 42
2.1.2.2.5 Perceived Self-efficacy 43
2.1.2.3 Control Variables 46
2.2 Limitations of the Study 47
CHAPTER 3: REVIEW OF LITERATURE 51
3.1 The Global Spread of HIV/AIDS 51
3.2 HIV/AIDS in Botswana 57
3.2.1 The Impact of HIV/AIDS on the Demography
of Botswana 60
3.2.1 HIV/AIDS and Mortality in Botswana 64
3.2.1.2 HIV/AIDS and Population Growth in
Botswana 67
3.3 Botswana's Health Policy 72
3.4 Botswana's response to HIV/AIDS 77
3.5 HIV/AIDS-Related Knowledge, Attitudes,
and Practice 80
3.6 The Dynamics of HIV/AIDS in Botswana 85
3.7 Restatement of Hypotheses 96
3.7.1 Condom Use 97
3.7.2 HIV/AIDS Testing 101
3.7.2Number of Sexual Partners 104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
v iii
CHAPTER 4: TRENDS IN HIV KNOWLEDGE,
ATTITUDES AND PRACTICE 108
4.1 Knowledge of HIV/AIDS and Prevention 111
4.2 Attitudes towards People Living With
HIV/AIDS 122
4.3 Sexual Behavior/Practices 128
CHAPTER 5: Determinants of HIV/AIDS
Preventive Health Behavior 135
5.1 Condom Use 140
5.1.1 Perceptions, Age and Condom Use 157
5.1.1.1'Age and Perceived Susceptibility to
HIV/AIDS 160
5.1.1.2 Age and Perceived Severity of HIV/AIDS 161
5.1.1.3 Age and Perceived Costs of Condom Use 162
5.1.1.4 Age and Perceived Benefits of Condom Use 163
5.1.1.5 Age and Perceived Self-Efficacy towards
Condom Use 164
5.1.1.6 Summary 165
5.1.2 Perceptions, Socioeconomic Status and
Condom Use 166
5.1.2.1 SES, Perceived Susceptibility to
HIV/AIDS and Condom Use 168
5.1.2.2 SES, Perceived Severity of HIV/AIDS
and Condom Use 169
5.1.2.3 SES, Perceived Costs of Condom Use
and Condom Use 17 0
5.1.2.4 SES, Perceived Benefits of Condom
Use and Condom Use 171
5.1.2.1 Summary 172
5.1.3 Perceptions, Gender and Condom Use 173
5.1.3.1 Gender, Perceived Susceptibility to
HIV/AIDS and Condom Use 174
5.1.3.2 Gender, Perceived Severity to HIV/AIDS
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ix
and Condom Use 17 6
5.1.3.3 Gender, Perceived Cost of Condom Use
and Condom Use 177
5.1.3.4 Gender, Perceived Benefits of Condom Use
and Condom Use 17 8
5.1.3.5 Gender, Perceived Self-Efficacy towards
Condom Use and Condom Use 17 9
5.1.3.5 Summary 180
5.2 HIV Testing 181
5.3 Number of Sexual Partners 190
CHAPTER 6: Discussion, Conclusions and
Recommendations 201
6.1 Condom Use 201
6.2 HIV/AIDS Testing 210
6.3 Number of Sexual Partners 214
6.4 Summary 215
Bibliography 219
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
X
Table 1:
Table 2:
Table 3:
Table 4:
Table 5:
Table 6:
Table 7:
Table 8:
List of Tables
Page
Model Specification and Perception
Indicators Identified in the Botswana
AIDS Impact Survey 4 5
Summary of Descriptive Statistics for
Variables Used in the Regression
Analyses 138
Logistic Regression Model for the Odds
of Using a Condom: Different Scenarios
under the Health Belief Model 143
Logistic Regression Model for the Odds
Of Using a Condom Accounting for
Clustering 150
Age-Specific Logistic Regression Model
for the Odds of Using a Condom:
Different Scenarios under the Health
Belief Model 159
SES-specific Logistic Regression Model
for the Odds of Using a Condom: Different
Scenarios under the Health Belief
Model 167
Sex-specific Logistic Regression Model for
the Odds of Using a Condom:
Different Scenarios under the Health
Belief Model 173
Logistic Regression Model for the Odds of
Getting an HIV/AIDS Test: Different
Scenarios under the Health Belief
Model 182
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
x i
Table 9: Logistic Regression Model for the Odds of
Getting an HIV/AIDS Test Accounting
for Clustering 186
Table 10: Logistic Regression Model for the Odds
of Having One Sexual Partner: Different
Scenarios under the Health Belief
Model 191
Table 11: Logistic Regression Model for the Odds of
Multiple Sexual Relationships Taking
Clustering into Account 195
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
x ii
Figure
Figure
Figure
Figure
Figure
Figure
Figure
Figure
Figure
Figure
List of Figures
1: Diagrammatic Scheme of the Health Belief
Model
2: Population of Botswana 1904 - 2001
3: Total Fertility Rate - Botswana:
1984-1996
4: Number of Deaths - 12 Months before
the 1991 Population & Housing Census
by Age
5: Number of Deaths 2 Months before the
1998 Botswana Demographic Survey by Age
6: Total Projected Population Growth Rates:
1991-2009
7: Botswana Projected Population Growth:
1991- 2010 "With No-AIDS" vs "With AIDS
and No-Intervention" Scenarios
8: Percent Distribution of Women Aged
15-49 who had ever heard about HIV/AIDS
by Age; 1988 vs 2001
9: Percent Distribution of Women Aged
15-49 who had ever Heard about HIV/AIDS
by Education: 1988 vs 2001
10: Percent Distribution of Women Aged
15-49 who had Ever Heard about HIV/AIDS
Residence: 1988 vs 2001
Page
18
57
59
61
62
64
66
105
107
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 11: Percent Distribution of Women Aged
15-49 who had Ever Heard about HIV/AIDS
and Believed Condom Use is an Effective
Prevention Method: 1988 vs 2001
Figure 12: Percentage of Women Aged 15-49 who
had ever Heard about HIV/AIDS and
Believed "Limiting partners" is an
Effective Prevention Method by Age:
1988 vs 2001
Figure 13: Percentage of Women Aged 15-4 9 who
had ever Heard about HIV/AIDS and
Believed "Limiting partners" is an
Effective Prevention Method vs "Both
Partners have no Other Partners" by
Age: 1988 vs 2001
Figure 14: Percent Distribution of Women Aged
15-49 who had Ever Heard about HIV/AIDS
and Believed Avoiding "Dirty" Needles is
an Effective Prevention Method by Age:
1988 vs 2001
Figure 15: Percent Distribution of Women Aged
15-49 who believed that a person with
HIV/AIDS should continue going to
school/teach by Age: 1988 vs 2001
Figure 16: Percent Distribution of Women Aged
15-49 who believed that a person
with HIV/AIDS should continue
to go to school/teach by Residence:
1988 vs 2001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
x iv
Figure 17: Percent Distribution of Women Aged
15-49 who believed that a person
with HIV/AIDS should continue to
go to school/teach, by Educational
Attainment: 1988 vs 2001 119
Figure 18: Percent Distribution of Women Aged
15-49 years Currently Using Modern
Contraception: 1984, 1988,and 1996 120
Figure 19: Percent Distribution of 15-49 years
By Current Use of Specific
Contraception Method: 1984, 1988,
and 1996 122
Figure 20: Percent Distribution of all Women
by Age and Current Use of the Condom:
1988 and 1996 124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
XV
ABSTRACT
This project seeks to establish why high
knowledge levels about HIV/AIDS have not translated
into high adoption levels of HIV/AIDS preventive
measures in Botswana. It uses Botswana Family Health
Survey of 1988 (BFHS II) and Botswana AIDS Impact
Survey (BAIS) of 2001, and demonstrates that knowledge
about HIV/AIDS was already high in Botswana in 1988,
but had risen even higher by 2001.
Analysis also indicates that relative to 1988,
use of condoms had risen by 2001. However, overall
use rates remained low at slightly over 10% among
women aged 15 to 49. Attitudes towards people living
with HIV/AIDS also changed remarkably between the two
time points, tending towards greater acceptance.
BAIS is also used to establish relationships
between perceptions specified by the Health Belief
Model (HBM) and health behavioral outcomes of condom
use, HIV testing, and limiting the number of sexual
partners to one.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
x v i
Results show that different perceptions are
significantly associated with different health
behaviors; for condom use the perception of cost is
highly significant, while for getting an HIV test it
is the severity perception. Costs associated with
having one partner are also significant as a
determinant of the number of partners.
Perceptions also interact with some control
variables. Costs associated with condom use differ
significantly with age; the odds of condom use are
significantly higher among older people who do not
perceive costs of condom use relative to those who do.
Among the youth, this relationship is not significant.
Also lower SES groups perceive significantly less
severity of HIV/AIDS than the higher SES bracket.
Finally, the odds of condom use are significantly
higher among men who perceive themselves to be at risk
of infection relative to those who do not, but this
perception is not a significant predictor among women.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1
CHAPTER 1
INTRODUCTION
1.0 Background
The last two decades of the twentieth century-
witnessed an outbreak of one of the most devastating
diseases humanity has ever known, AIDS, short for
Acquired Immuno Deficiency Syndrome, caused by the HIV
virus. While its origins have remained largely
controversial one thing is clear, AIDS has
dramatically reversed gains made in many spheres
across the socio-economic spectrum. In some countries
life expectancy at birth has declined considerably;
morbidity and mortality rates have risen, causing
shrinkages in the certain age groups, which has in
turn slackened population growth (United Nations
Development Programme [UNDP], 2000).
On the economic front, countries have lost
skilled manpower and general labor due to HIV/AIDS
(Joint United Programme on HIV/AIDS [UNAIDS], 2000).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
The challenge is particularly staggering for
developing countries because their skilled manpower
shortages date back to the pre-HIV/AIDS days. Labor
performance and productivity have also taken a hard
knock as absenteeism at the workplace rose while
infected people seek medical attention (BIDPA, 2000),
a long and painful stage that usually precedes death.
The effects of HIV/AIDS have also been felt at
the household level. Many families have lost
breadwinners, leaving behind dependents and forcing
some governments to introduce welfare schemes to
assist affected households. This has not only
frustrated policies designed to eliminate poverty but
also exacerbated the incidence of poverty itself.
Also, some governments have been forced to shelve
planned development projects in order to fight the
HIV/AIDS pandemic.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
For example, Botswana's Minister of Finance and
Development Planning, stressed in the 2001 budget
speech that '...implementation of projects and the
various measures in the health sector, aimed at
dealing with the HIV/AIDS scourge, will have serious
financial and manpower implications... sacrifices will
have to be made in terms of alternative programs and
projects to be foregone in all sectors of government'
(Ministry of Finance and Development Planning [MFDP],
2001, p. 14).
AIDS however, has affected different countries
and societies differently, largely due to differences
in transmission methods and response patterns related
to the prevention of the disease, notably timing of
public intervention. Like any disease, it takes a
wide array of preventive and curative strategies to
bring it to a halt, and these have tended to vary
across countries. Awareness-creation, for example,
requires public educational campaigns to be launched
and pursued vigorously to sensitize the population at
risk.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
Not only does this require a wide range of resources,
it also needs organized institutions such as
government departments, non-governmental organizations
(NGOs), and volunteer groups to work together
systematically for optimum effectiveness.
Cures are not only expensive; they do not exist
in the case of HIV/AIDS. Life-prolonging drugs
available so far are accessible mainly in those
countries that have the technological muscle to
generate them; diffusion to the rest of world has been
slow. Not surprisingly, resource-endowed nations have
been more effective at fighting the spread of HIV/AIDS
than their poorer, less developed counterparts. For
example, according to Mann, Tarantola, and Netter
(1992) in 1990 North America was spending $2.71 per
capita on HIV/AIDS prevention programs, Oceania spent
$2.23, Europe $1.18, while sub-Saharan Africa was
spending a paltry $0.07 and Latin America $0.03.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5
Hopes for a sustained and effective campaign
against HIV/AIDS in developing nations therefore lie
with prevention of infection, rather than treatment of
already existing cases, even though this is also
critical. For this reason, this project focuses on
HIV/AIDS preventive behavior trends and underlying
determinants in one such country, Botswana. The
primary goal is to shed light on trends in prevention
as well as to investigate factors that hinder adoption
of preventive practices.
1.1 The Research Problem
Statistics from various sources have consistently
revealed very high awareness rates with respect to the
existence of HIV/AIDS, as well as its spread
mechanisms in Botswana. As early as 1988 88.3% of
women aged 15 to 49 had heard about HIV/AIDS, and more
than three quarters of them were aware that having
multiple partners increased the risk of infection
(Central Statistics Organization [CSO], 1989).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
Recently the Botswana AIDS Impact Survey (BAIS)
revealed that 94% of all people aged between 10 and 64
years had ever heard about HIV/AIDS, and 79% of the
same population mentioned at least one way of reducing
chances of infection (CSO, 2002).
Disappointingly, high awareness levels have
apparently not been translated into practice.
Specifically, high-risk behavior seems to persist, as
indicated by high incidence of unprotected sex,
multiple sexual partners, and low levels of
abstinence. For example, as much as 198 942 cases of
sexually transmitted diseases were recorded in 1999,
while 17.8% of all births during the same year
occurred to teenagers (CSO, 1999). In addition, 21.1%
of people aged 10 to 64 years, in the capital city
Gaborone, and 63.3% in small towns, continued to have
unprotected sex with their partners even when they
were aware they had genital ulcers (CSO, 2002).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
Further, turn out for free HIV tests has not been
particularly impressive, which means that many
asymptomatic cases are not aware of their status, and
therefore are not motivated to take caution even if
they could. For example, even though 59% of women
interviewed in the BAIS knew of a place to get tested
for HIV/AIDS, only 14% of them had actually gone for
testing (CSO, 2002).
This project recognizes, and attempts to address,
the huge difference between knowledge about HIV/AIDS
preventive methods and corresponding practices in
Botswana. It attempts to answer guestions such as:
why is it that risky behavior continues despite high
levels of awareness regarding the existence and
transmission of HIV/AIDS? What characteristics,
social, economic, or demographic, make individuals
less likely to conform to health professionals'
advice, and therefore predispose them to high risk of
infection, and why?
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.2 Justification of the Study
Over the past two decades or so that the world
has been aware of HIV/AIDS, a lot has been written
about this disease. Researchers from a wide spectrum
of disciplines, especially from the medical and
behavioral sciences have contributed to an ever-
increasing volume of literature. Naturally, a lot of
research was initially devoted to understanding the
transmission process of HIV as well as efforts to find
a cure. However, when it became apparent that the
cure for AIDS would not be forthcoming in the near
future, concern mounted over the possible impact of
the disease. A second dimension therefore emerged:
the impact of HIV/AIDS. Demographic and economic
models of varying complexities have been developed in
an attempt to predict the impact of HIV/AIDS,
particularly in developing countries where the disease
has had, and continues to have, devastating effects.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
These models have focused on national, household as
well as individual levels, using traditional
indicators such as the Gross Domestic Product (DGP),
mortality rates, life expectancy, unemployment levels,
orphanage rates, etc.
This research project departs from the norm in
one major way: instead of concentrating on what has
become almost a tradition for social research on
HIV/AIDS, its impact, the thesis seeks to explain
dynamics behind the phenomenal spread of the disease,
focusing specifically on Botswana. It is generally
agreed amongst both the research community and policy
making bodies that much hope for a win against
HIV/AIDS lies with effective prevention programs,
particularly at a time when an effective cure still
eludes modern science. Answers need to be found
regarding why people are still getting infected
despite wide availability of condoms; why are people
still not getting tested for HIV/AIDS when tests are
free, and the inherent benefits of early testing are
so highly publicized?
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
Why are people still maintaining multiple
relationships despite the high levels of awareness
that this increases risk of infection?
Answers to these questions will feed back into
the research process and help improve 'impact models'
because they lay a foundation for these models'
assumptions. After all, how else can assumptions over
incidence and prevalence rates covering a specified
period be made when not much is known about the
dynamics behind the transmission process in the first
place?
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11
1.3 Objectives of the Study
Broadly, this study intends to shed light on the
dynamics underlying the phenomenal spread of HIV/AIDS
in Botswana since the first case was diagnosed in 1985
till 2001. The following are its specific objectives:
i) to trace trends in HIV/AIDS knowledge
between the 1988 and 2001
ii) to trace trends in awareness regarding
HIV/AIDS transmission mechanisms between
1988 and 2001
iii) to investigate the relationship between
perceptions towards HIV/AIDS and the
adoption of medically recommended
HIV/AIDS preventive behaviors,
specifically condom use, limiting sexual
partners to one, and getting a test for
HIV
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
iv) to explore relationships between
perceptions and other factors with a goal
to furthering the understanding of
underlying dynamics of the HIV/AIDS
epidemic.
1.4 Conceptual Framework
This project adopts the Health Belief Model (HBM)
as its guiding framework, mainly because the model has
been applied in a handful of AIDS-related behavioral
studies; therefore this will facilitate comparability
between this study and others. HBM has also been
shown to be fairly parsimonious in explaining health
behavior, largely due to its cost-benefit approach
(Vanlandingham, Suprasert, Grandjean, & Sittitrai,
1995; Langlie, 1977).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
Widely believed to have been developed by
Rosenstock in the 1960s (Harris and Guten, 1979), but
with claims of having been originated by Kurt Lewin
(Bloor, 1995), HBM briefly states that preventive
health behavior is closely related with a set of
perceptions of disease. First among these is
susceptibility, that is, people need to feel
threatened by disease; in other words, they need to
perceive themselves as being at risk of getting the
disease.
Secondly, they need to perceive the consequences
of the disease as being severe. Diseases considered
lethal are more likely to trigger preventive
behavioral adjustments than minor ailments that do not
cause much damage.
Thirdly, people need to acknowledge barriers
associated with preventive behavior, that is, they
have to be aware of possible negative consequences, or
costs.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
Often, disease prevention entails foregoing some
privileges that people enjoyed prior to the discovery
of the disease, and letting go of such privileges can
prove difficult.
Fourth, there is need for appraisal of the
benefits of preventive action, against its costs.
Benefits are the motivating factors or gains that
people perceive they would derive from adopting
preventive behavior.
The model also takes into account people's
perceived personal ability to perform preventive tasks
to protect themselves from the disease, or self-
efficacy.
Lastly, the Health Belief Model takes into
account the modifying role of demographic, structural
and social psychological factors on these perceptions
(Harris and Guten, 1979).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
1.4.1 Perceived Susceptibility
According to the Health Belief Model, people need
to feel vulnerable to disease in order to be motivated
to take preventive action aimed at avoiding infection.
In other words, they need to have a reason to believe
that the disease will affect them, in one way or
another. In the case of HIV/AIDS, this means that one
has to perceive themselves or those close to them as
being at risk of HIV infection in order to engage in
preventive behavior. Inherently, this depends on the
individual's awareness regarding channels through
which the virus is spread, to enable them to make a
fairly accurate assessment of their level of risk.
People from communities where there is information and
open discussion about HIV/AIDS and its spread
mechanisms can therefore be reasonably expected to
engage in preventive activities more readily than
those from closed communities where hardly any
discussion is encouraged or practiced.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16
Basset and Mhloyi (1991) provide a useful example of
women in Zimbabwe who, after learning about the close
correlation between HIV transmission and sexually
transmitted diseases (STD), are understood to have
started inspecting men for signs of genital ulcers
before agreeing to have sex with them.
1.4.2 Perceived Severity
Appraised severity of the disease is believed to
contribute to the adoption of preventive health
behavior under the HBM. According to this concept,
diseases considered severe by an individual are likely
to trigger more deliberate and resolute action on the
part of that individual to protect themselves from
getting infected, relative to those that they rate as
not so severe.
HIV/AIDS is an incurable disease, and up to now
there is still not much hope that a cure will be found
in the near future; this reality has been given high
prominence in Botswana's anti-AIDS campaign.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17
Against this background, it can generally be expected
that most Batswana's perception of this disease will
be that of high severity. It will therefore be
interesting to see if such a perception motivates
individuals towards adoption of preventive health
behaviors.
1.4.3 Perceived Benefits versus Perceived Costs of
Preventive Action
The Health Belief Model further informs us that
for people to take preventive health action the
benefits accruing from such action must be perceived
to outweigh its costs. In the context of HIV/AIDS,
individuals need to see the benefits of using a
condom, as opposed to non-use; sticking to one sexual
partner as opposed to numerous partners; and knowing
their HIV status rather than staying ignorant.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
18
These are fundamental issues in HIV/AIDS prevention,
because it is here that various underlying factors,
ranging from attitudes to poverty, intersect and
determine the type of behavior taken. For example,
where there is a popular belief that condom use
reduces sexual pleasure, or that knowing one's HIV
status, especially if positive, quickens the rate of
bodily deterioration, or where poverty drives people
to seek multiple partners, adoption of preventive
health behavior may face insurmountable setbacks.
It is implied in this aspect of the Health Belief
Model that the health system should facilitate
adoption of preventive behavior by making information
available to the general public to help them make
these choices. Questions regarding relevance,
accuracy and clarity of public campaign messages must
necessarily be raised at this stage; do people
understand messages from health personnel? Is the
information contained in these messages accurate so
that the public may be reasonably expected to make
informed choices?
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
19
1.4.4 Perceived Self-Efficacy
The Health Belief Model recognizes the
significance of behavioral cues to action, bringing
into the picture issues of self-efficacy and how they
influence preventive behavioral adoption. After an
individual has assessed his/her own susceptibility and
disease severity, benefits and costs of action, the
next stage is concerned with whether or not to take
action. This last part depends solely on the
individual's perceived ability to perform the task at
hand, say use a condom. How does the individual view
condom use vis-a-vis personal circumstances such as
competence to use it, partner's reaction, disposal of
a used condom, etc? How much control does the
individual have over their partner's willingness to
use a condom?
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
1.4.5 Control Variables
A set of demographic, structural and social
psychological factors is conceptualized to have an
intervening effect on the explanatory power of these
perceptions under HBM. For example, preventive health
behavior is likely to differ according to the age of
the individual involved, their sex, education, and
related factors. It may vary depending on structural
factors such as the focus of health policy and the
distribution of health facilities, accessibility of
information, etc; and social factors like gender
relations and healing options available to
communities. The following figure is a diagrammatic
representation of the HBM.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
21
Figure 1: Diagrammatic Scheme of the Health Belief
Model
Independent M ediating Outcome
P e r c e p t i o n s Preventive H ealth
B ehavior
• Susceptibility
/ M odifying Factors \
• Severity
/ \ • Condom Use
• Costs
/ • Demographic \
• Single
• Benefits
• Structural \
Partnerships
• Self-Efficacy
—
• Social
• Get
\ / HIV/AIDS
Test
1.4.6 Preventive Health Behavior versus Health-
Protective Behavior
It is necessary at this stage to define
preventive health behavior (PHB), and differentiate it
from Health-Protective Behavior (HPB). Harris and
Guten (1979) quote Kasl and Cobb defining PHB as 'any
activity undertaken by a person believing himself to
be healthy, for the purpose of preventing disease or
detecting it in an asymptomatic stage' (p. 17).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
22
This definition identifies two conditions, at least
one of which must be met for PHB to be satisfied:
direct contact with health care professionals and/or
compliance with health care professionals'
recommendations. Direct contact requires physical
interaction with agents of health care such as
hospital visits, home visits, immunization, etc
whereas compliance is concerned more with the
influence that agents of the health system have on the
recipients of health care, with or without physical
interaction. In other words, this definition limits
itself to issues of accessibility of and response to
the health system.
Health-Protective Behavior (HPB) on the other
hand, has been defined by Harris and Guten (1979) as
'any behavior performed by a person, regardless of his
or her perceived or actual health status, in order to
protect, promote, or maintain his or her health,
whether or not such behavior is objectively effective
toward that end' (p. 18).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23
The main difference between the two lies in the
medical worthiness or effectiveness of the behavior
adopted. While PHB emphasizes access and conformity
to medically approved methods, HPB transcends the
medical horizon and encompasses activities that the
individual believes possess a health protective
potential, with or without medical verification.
This project is concerned with evaluating
response to health policy, which outlines a set of
medically approved activities with the overarching
objective to protect a whole nation against HIV/AIDS;
therefore it adopts preventive health behavior as its
operational definition of health behavior.
Specifically, the preventive dimension of the Botswana
AIDS policy emphasizes abstinence, consistent condom
use, having one sexual partner at a given time, and
testing for HIV in order to know one's infection
status, all of which are medically founded. Further,
information about these practices is transmitted
through contact, direct or indirect, between agents of
health care and its recipients.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
24
The extent to which this information will trigger
required behavior depends entirely on the public's
readiness to conform, which in turn depends on factors
specified by the HBM.
1.4.7 Critiques of the Health Belief Model
Despite its relative parsimony and general
consistency the HBM has been criticized for a number
of shortcomings. Vanlandingham and his colleagues
(1995) question the ability of individuals to evaluate
potential consequences of certain behavior as well as
to use such evaluations to make decisions about
behavioral change. This challenges the premise that
individuals' perceptions about costs and benefits of
adopting preventive health behavior play a role in the
resultant behavior.
The cost-benefit approach has also been
criticized for its failure to recognize the role of
external forces in determining behavior.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25
Specifically, concern here lies with the inability of
HBM to incorporate into its framework the effect of
social networks in behavior formation. This is a
critical issue especially in instances where the
behavior in question is subject to influences beyond
the individual's control. A good example is sexual
behavior among adolescents, a phenomenon that is
closely linked to peer group influence.
Some researchers have proposed alternative
theories as a way around this shortcoming. One such
theory is the Theory of Reasoned Action (TRA), whose
efficiency to handle social network effects was well
demonstrated by Vanlandingham et al (1995). Langlie
(1997) proposed what she termed the Social Network
Model of Preventive Health Behavior, which also proved
useful in accounting for network effects. This
project had wished to employ either the TRA or social
network model, but data constraints do not permit
network analysis.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26
Another area of concern about HBM is, as stated
by Bloor (1995) '...its tendency to conceive of risk
behavior as a volitional individual act... for the
analysis of behavior which self-evidently involves two
parties not one individual and which is characterized
by constraint, not free choice' (p. 90). Bloor
challenges the whole premise of HBM, and even goes on
to suggest that instead of using perceptions as
independent determinants of health behavior,
consideration must be made to treat them as dependent
outcomes. The argument is that perceptions upon which
HBM is based are shaped by social circumstances of
individuals, and these must be the focus for
investigation, so that variations in perception can be
accounted for. While this argument makes sense, it
does not dispute the existence of a relationship
between perceptions and health behavior. Indeed it is
true that even after explaining the root cause of
perceptions, their association with different health
behavior as propounded by the HBM, will remain.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
27
In fact, in a study of male university students,
soldiers, clerks, and laborers in Thailand,
Vanlandingham et al (1995) report highly statistically
significant associations between preventive health
behavior (condom use) and perceptions of
susceptibility, severity, costs, and benefits.
1.5 Preliminary Hypotheses of the Study
Closely guided by the HBM, the following preliminary
hypotheses can now be stated:
i) People who feel susceptible to HIV/AIDS,
perceive it as severe, perceive less cost with
condom use, perceive great benefit in condom
use, and have some sense of self-efficacy
regarding use of condoms will be more likely to
use condoms.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
28
ii) People who feel susceptible to HIV/AIDS,
perceive it as severe, perceive less cost
associated with testing for HIV, perceive great
benefit in HIV tests, and have some sense of
self-efficacy with regard to HIV testing will
be more likely to get an HIV/AIDS test.
iii) People who feel susceptible to HIV/AIDS,
perceive it as severe, perceive less cost
associated with limiting partners, perceive
great benefit in having one sexual partner, and
have some sense of self-efficacy with regard to
number of sexual partners will be more likely
to avoid multiple partnerships.
These hypotheses will be reviewed and restated later
to incorporate issues raised in the literature review.
This will situate the hypotheses in the context of
Botswana thus facilitate analysis accordingly.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
CHAPTER 2
METHODOLOGY
2.1 Data
This study relies only on secondary data, the
first and primary source being the Botswana AIDS
Impact Survey (BAIS) of 2001, while the second is the
Botswana Family Health Surveys of 1984, 1988, and
1996. The BAIS data was collected in January 2001 by
the government of Botswana through the CSO with the
main objective being to 'provide up-to-date
information for assessing the impact of HIV/AIDS
pandemic at household level in Botswana' (CSO, 2002;
p. 21). It is a nationally representative household
level probability sample of 2 203, with a response
rate of 88%. All males and females aged between 10
and 64 found in the selected households were also
interviewed individually, yielding a total of 4 494
eligible respondents, out of whom 4 278 were
successfully interviewed (CSO, 2002).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
The fact that men were included in the sample adds
strength to this analysis because most demographic
research tends to focus only on women.
Analysis for this project is limited to 15 to 49
year-olds, because this age bracket contains the core
of the 'sexually-active' population. Further,
analysis is restricted to those who answered, "Yes" to
the question: "Have you ever had sexual intercourse?"
This allows the study to focus only on sexually
experienced individuals since sex is the main
transmission channel of HIV in Botswana. Besides, the
nature of dependent variables used in this analysis
(Condom Use, Ever Tested for HIV, and Number of Sexual
Partners) dictates that only people who have engaged
in sexual activity before be included. These
restrictions yield a total of 2 466, still substantial
to permit the execution of detailed statistical
procedures.
The Botswana Family Health Surveys series is more
useful for tracing trends in HIV/AIDS knowledge,
attitudes and practices over the years.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31
Especially useful is the information relating to
contraceptive use, particularly condoms because this
has direct implications for the adoption of HIV/AIDS
preventive health behaviors.
2.1.1 Trends in Preventive Health Behavior
There is no single systematic data source to meet
the requirements for this type of analyses in
Botswana. Yet this is a critical condition for
understanding the direction of behavior trends as well
as assessing the effectiveness of HIV/AIDS-related
policy intervention.
This project has settled for archival data
obtainable from two major sources, namely the second
Botswana Family Health Survey of 1988 (BFHS-II), as
well as the Botswana AIDS Impact Survey (BAIS) of
2001. The strategy is to identify two related
questions in each survey, and compare responses at
each of the two time points to see how much change has
occurred and the direction of change.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
32
Three major aspects are covered under this section:
knowledge of HIV/AIDS, attitude towards people living
with HIV/AIDS, and sexual behavior.
To facilitate comparability, the analysis will
use only women aged 15 to 45 for the knowledge and
attitudes aspects. This is necessitated by the fact
that BFHS-II used this age group as its target
population.
Sexual behavior focuses primarily on family
planning practices in general, and condom use in
particular, with a goal to detect fluctuations in use
over the years. A lot more data is readily available
for family planning than the other two aspects, so
analysis here will cover more time points.
2.1.2 Modeling Preventive Health Behavior
As stated in the objectives section, one of the
key goals of this project is to identify factors that
deter individuals from adopting recommended HIV/AIDS
preventive methods.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33
Data for these analyses was obtained from the BAIS of
2001, unarguably the richest and most recent data set
on HIV/AIDS in Botswana to date.
2.1.2.1 The Dependent Variables
Three separate dependent variables are
identified, consistent with the content of anti-AIDS
campaign messages in Botswana. For years now, health
professionals and other community leaders have been
urging people to abstain from sex, be faithful to
their partner, and use condoms whenever they engage in
sexual acts. This is popularly known as the ABCs of
HIV/AIDS prevention, 'A' for Abstain, 'B' for Be
faithful, and 'C' for Condom use.
Abstinence was however not adequately addressed
in the BAIS data. This thesis therefore chose to
replace abstinence with HIV/AIDS testing, another
component of the anti-AIDS campaign that is fast
gaining momentum.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34
It is widely believed that people who are aware of
their HIV status are better positioned to adopt
preventive behavior than those who are not, for many
reasons. First, such knowledge may motivate them to
adopt safer sex practices, especially when reinforced
with effective counseling. Secondly, knowledge of
their status may help HIV-positive people live longer,
especially if they adopt healthy living habits and/or
have access to anti-retroviral cocktails. Third, HIV-
positive people who are aware of their status are in a
better position to avoid pregnancy thus contribute to
the reduction of pediatric AIDS. Forth, HIV-positive
people may be motivated to participate actively in the
fight against HIV/AIDS, giving it impetus and adding
to its overall effectiveness. Lastly, testing could
motivate those not already infected to stay negative,
thus also contributing to the control of the epidemic.
Getting an HIV test is therefore a critical indicator
of preventive health.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
35
The other dependent variables are condom use and
number of sexual partners over the twelve months prior
to the date of interview. Ideally, condom use would
be best studied by incorporating consistency of use
because only consistent users can be sure of
protection, but this information was not collected
during the BAIS. Even though this is a serious
limitation, a lot can still be learnt from factors
that influence condom use in general, especially for a
country where no such analyses have ever been done
before.
Number of sexual partners is another indicator of
the prevalence of risky behavior, because the risk of
infection increases with each sexual encounter,
especially where the 'sexual network' is wide. The
'stick to one partner' campaign has been running in
Botswana's mass media for more than a decade now, and
ideally the incidence of multiple sexual partnerships
should have declined by now.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
36
Yet this does not seem to be completely the case, as
shown by recent statements expressing concern in the
local media (Mmegi 25-31 October, 2002). This project
therefore has three dependent variables, measured as
dummies:
a) Condom Use - "Used" versus "Otherwise": the
question asked is "Did you use a condom the
first time you had sex with this (most
recent) partner?" 'Used' refers to those
who answered "Yes" whereas 'Otherwise'
refers to those who answered 'No' and those
who could not remember.
b) HIV/AIDS testing - "Ever Tested" versus
"Otherwise". The question is: "I don't want
to know the results but have you ever been
tested to see if you have HIV, the virus
that causes AIDS?" with "Yes/No" choice.
c) Number of Sexual Partners - a dummy variable
is created from the question: "In the last
12 months how many people overall have you
had sex with?"
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
37
The reference category consists of people who
had only one partner. Indeed this question is
very sensitive and might not be free from bias
but overall, it still remains a good indicator
of the prevalence of multiple sexual
relationships.
All three models will use binary logistic
regression to yield the odds of using a condom or
getting an HIV/AIDS test or having more than one
sexual partner for given specific perceptions and
controlling for specified modifying factors.
2.1.2.2. Explanatory Variables
Explanatory variables are categorized according
to the perceptions specified by the Health Belief
Model; susceptibility, severity, costs, benefits, and
self-efficacy. Questions relating to each of the above
were identified as presented in the following
subsections.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38
2.1.2.2.1 Perceived Susceptibility
The question identified as an indicator for
susceptibility is: "do you think your partner has
other partners?" Even though suspecting infidelity
may not necessarily trigger feelings of
susceptibility, at least not as strong and immediate
as those of jealousy, this question was chosen because
of two main reasons; firstly, because in Botswana
infection among the sexually active population, which
is the target group for this analysis, is mainly
through heterosexual contact. Staying with one
faithful partner therefore has a good potential to
minimize chances of infection, so that where it does
not happen people have a good reason to get worried
over possible HIV/AIDS infection.
Secondly, the proportion answering "Yes" to a
related question "Can people reduce their chances of
getting HIV/AIDS by having only one partner who has no
other partners?" was 84%, indicating high awareness on
the risks inherent in promiscuity.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39
This measure is used in all three models because the
feeling of susceptibility can equally lead to condom
use, getting an HIV test, or reduction of sexual
partners.
2.1.2.2.2 Perceived Severity
Perception of HIV/AIDS severity is measured by a
person's response to the question "Do you personally
know anyone who has HIV or who has died from AIDS?"
Knowing someone with HIV/AIDS, or who died from it,
can make all the difference in people's perception of
the disease because then they would appreciate the
symptoms and agony associated with it. This question
also cuts across all the models.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
40
2.1.2.2.3 Perceived Costs/Barriers
Costs are model-specific. For example, it is
would not make sense that the cost of using condoms,
say loss of sexual pleasure, be perceived as also
applying to getting an HIV test. For the condom use
model the question used to depict costs of condom use
is "Do you think that women should always be allowed
to buy condoms?" This question reflects cultural
barriers that may exist regarding condom use, and is
therefore useful in gauging the extent of their
influence in this phenomenon.
For the HIV test model the question asked was "If
you chose to be tested for HIV, the virus that causes
AIDS, and were told after the test that you had HIV,
would you tell anyone the results?" The rationale is
that people who foresee high costs associated with
being HIV-positive would answer in the negative, and
would thus be least likely to go for a test after all.
It is often argued that there is a substantial amount
of stigma attached to HIV/AIDS.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
41
Fears have also been expressed that stigmatization
poses the danger of driving people living with
HIV/AIDS underground to avoid social and other sorts
of discrimination. This element is expected to have a
bearing on whether one gets a test or not.
Specifically, people who felt they could tell someone
else their test results even if they tested positive
are expected to show a higher likelihood to have
tested for HIV than those who felt otherwise.
For the model on number of sexual partners the
question asked was "In the last 12 months have you
exchanged gifts or money for sex?" People who answer
in the affirmative would generally be expected to
attach higher costs to partner number reduction since
they realize material benefits out of such
relationships, relative to those answering in the
negative.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
42
2.1.2.2.4 Perceived Benefits
Perceptions of benefits are also specific to the
models. For condom use, the indicator for benefit is
the response to the question: "Using condoms can
reduce people's chances of becoming infected with
HIV", with which the respondent agreed or did not
agree.
The second model uses the question "Can a person
who looks healthy be infected with HIV/AIDS?" People
answering "No" to this question are more likely never
to go for a test until they show some signs of illness
than those answering in the affirmative.
For Model 3, the question is "Can people reduce
their chances of getting HIV/AIDS by having only one
partner who has no other partners?"
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
43
2.1.2.2.5 Perceived Self-efficacy
Ideally, issues of self-efficacy must also be
largely model-specific, but no specific questions
could be identified to satisfy this condition.
Fortunately, a general question was found that
reflects an individual's overall attitude towards
HIV/AIDS: "Is there anything a person can do to reduce
their chances of becoming infected with HIV, the virus
that causes HIV?" Langlie (1977) calls this the
"perceived internal locus of control" or PIC, a
reflection of alienation or powerlessness, and sums it
up nicely thus:
The hypothesis is that persons who view
themselves as having some control over what
happens to them are likely to perceive action as
efficacious and are more likely to perceive and
process information relevant to engaging in
specific actions.1
1 Jean K. Langlie; Social Networks, Health Beliefs, and Preventive
Health Behavior", Journal of Health and Social Behavior 18(3)
244-260
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44
People who feel there is something that one can
do to avoid infection are therefore more likely to be
those who generally perceive action as efficacious,
giving them some sense of control over their destiny
with respect to HIV/AIDS infection. The following
table summarizes the three models and indicators of
perceptions for each concept.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
Table 1: Model Specification and Perception Indicators
Identified in the Botswana AIDS Impact Survey
Perception
Dependent Variable
Condom Use HIV Testing Sexual
Partners
Did you use a
condom the first
time you had sex
with this (most
recent) partner?
I don't want to
know the
results but
have you ever
been tested to
see if you have
HIV, the virus
that causes
AIDS?
In the last
12 months
with how many
people
overall have
you had sex?
Susceptibility
Do you think your
partner has other
partners?
Do you think
your partner
has other
partners?
Do you think
your partner
has other
partners?
Severity
Do you personally
know anyone who
has HIV or who
has died from
AIDS?
Do you
personally know
anyone who has
HIV or who has
died from AIDS?
Do you
personally
know anyone
who has HIV
or who has
died from
AIDS?
Barriers/Costs
Do you think
women should
always be allowed
to buy condoms?
If you chose to
be tested for
HIV, the virus
that causes
AIDS, and were
told after the
test that had
HIV, would you
tell anyone the
results?
In the last
12 months
have you
exchanged or
received
money for
sex?
Benefits
Can people reduce
their chances of
becoming infected
with HIV by using
condoms ?
Can a person
who looks
healthy be
infected with
HIV/AIDS?
Can people
reduce their
chances of
getting
HIV/AIDS by
having only
one partner?
Self-Efficacy
Is there anything
a person can do
to reduce their
chances of
becoming infected
with HIV, the
virus that causes
AIDS?
Is there
anything a
person can do
to reduce their
chances of
becoming
infected with
HIV, the virus
that causes
AIDS?
Is there
anything a
person can do
to reduce
their chances
of becoming
infected with
HIV, the
virus that
causes AIDS?
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
46
2.1.2.3 Control Variables
The Health Belief Model recognizes the fact that
health behavior does not only depend on the above
specified perceptions; in fact, it states that these
perceptions are modified by other factors. These
factors are generally of demographic, structural, and
psycho-social nature. Below is a list of variables
identified under this category:
i) Demographic factors - Age, Sex,
Educational Attainment, Marital
Status, whether couple live together
or apart, place of residence (rural
versus urban), age difference between
partners, occupation.
ii) Structural factors - these are factors
that are to a large extent beyond the
control or influence of the
individual, or system barriers.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
47
They include knowledge of an HIV/AIDS
Testing Center, Heard or seen HIV/AIDS
information in the past 4 weeks, ever
heard of any of the AIDS commemoration
events (AIDS Day, Month of Prayer,
etc) .
iii) Psycho-social factors - the social
system can also influence people's
perceptions; women's status in
society, traditional beliefs, etc.
2.2 Limitations of the Study
The fact that this study uses secondary data
exposes it to a number of constraints. First, some
variables that could have been very useful to this
project are not available in the sources used. For
example, it would have been very helpful to these
analyses if specific information relating to the
perception variables, say costs of using a condom, or
having only one sexual partner, had been available.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
48
It has been alleged in certain quarters that some men
refuse to wear condoms because they feel this reduces
sexual pleasure. Other works have suggested a close
link between poverty and the prevalence of multiple
sexual partners (UNDP, 2002), albeit with little or no
empirical basis.
Even though this study identified proxy variables
to test such hypotheses, it still feels much remains
to be gained from using direct measures for such
concepts. In other words, more specific and direct
questions could have been preferred to those used
here, data permitting. For example, instead of using
the question "do you think your partner has other
partners?" to indicate susceptibility, a more direct
question such as "do you consider yourself at some
risk for contracting HIV/AIDS" could have been more
precise and a more valid measure of susceptibility.
However, indicators used here are considered good
proxies for the corresponding concepts.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
49
Further, this analysis could have benefited
immensely from data documenting the consistency of
condom use among those indicating having used this
method beginning from a specified reference point.
Effective prevention of HIV infection requires
consistent condom use, and answering "yes" to the
question "Did you use a condom the first time you had
sex with this (most recent) partner?", used as the
dependent variable in the first model here, does not
enable us to distinguish between habitual and
intermittent users. Obviously, the level of risk of
HIV infection between the two groups is not the same;
this study cannot account for this difference due to
lack of information.
Related to the above limitation is the issue of
reliability of data on self-reported private behavior
such as sexual practices, especially in a cultural
setting like Botswana's where open discussion of sex
is taboo.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50
Data on sexual behavior therefore faces the challenge
posed by complete reliance of the researcher on what
the respondent is willing to divulge regarding their
sexual practices without the possibility of cross
checking it for truthfulness. Of particular concern
here is the fact that the respondent may be tempted to
tell the researcher what is socially expected, eg,
condom use, which may not be the same as what they
practice.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
51
CHAPTER 3
REVIEW OF LITERATURE
3.1 The Global Spread of HIV/AIDS
HIV/AIDS was first brought to the attention of
the world in the early 1980s when unusual
opportunistic infections were observed among 26 gay
men in New York and 5 in Los Angeles (Walker, 1998) .
Before long, similar cases were reported in other
parts of the developed world, notably Western Europe
and Australia. At this stage, the disease was thought
to be a 'gay plague' , and was actually called GRID,
short for gay-related immunodeficiency (Fee & Krieger,
1994) . It was not until a few years later that its
incidence among non-gay populations was noticed, this
being among predominantly heterosexual societies of
Central Africa and Haiti. The World Health
Organization immediately stepped up its surveillance
program to monitor the spread of this new disease, and
over most of the 1980s the number of reported cases
doubled annually (Bongaarts, 1996).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
52
It is possible that even with such an exponential
growth of reported cases, the incidence of HIV/AIDS
was most probably under-reported, owing largely to
lack of diagnostic equipment in many parts of the
world at that time.
There has been a lag in the timing of HIV/AIDS
between regions of the world. As already noted, early
cases of the disease were first noticed in the early
1980s in most developed countries, and shortly
thereafter in sub-Saharan Africa. However, it was not
until almost a decade later that it was recorded in
Asia (Mann et al, 1992).
HIV/AIDS prevalence has since spread
dramatically, reaching all parts of the world over the
last decade. Leading by far in this respect is
Africa, even though Asia and Latin America have also
seen a considerable rise in their rates (UNAIDS,
2001) . Prevalence rates in North America and Western
Europe leveled off in this period (Bongaarts, 1996).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
Bongaarts further asserts that sooner or later,
prevalence rates in the other regions will ultimately
reach a plateau because 'every population is a mixture
of more or less frequently interacting subgroups with
widely varying infection risks' (p. 26). In fact, it
is upon this interaction and the size of these
subgroups that the overall effect of HIV/AIDS will
ultimately depend.
The fact that each population has its own pattern
of infection, which in turn depends on specific
attributes like its age distribution, settlement
patterns, transmission mechanisms, and socio-cultural
factors means that different societies will be
affected differently. Countries with young
populations such as those of sub-Saharan Africa and
other developing countries will probably take much
longer for their rates to level off because of the
huge size of their risk populations, and as a result
stand to lose large portions of their populations.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
54
Within countries, urban areas have been found to
be the concentration centers of HIV/AIDS, an
indication of an uneven distribution of high-risk
groups such as prostitutes, drug users, even gay
populations between urban and rural areas. For
example, a 1991 survey in Rwanda found that prevalence
in the urban areas was as high as 25% whereas in the
rural areas it stood at only 2.2% (Mann et al, 1992).
With respect to socio-cultural factors,
Caldwell, Caldwell and Quiggin (1989) reported that
sexual behavior in sub-Saharan was characterized by
frequent contacts with persons other than a permanent
partner, which they also associated with a relatively
high incidence of sexually transmitted diseases
(STDs). Other studies (United Nations, 1989) have
also reported low rates of condom use in this part of
Africa, which further exacerbates the problem.
These factors combine to paint gloomy prospects for
such countries. At the national level, they will
inevitably suffer heavier mortality as the large
numbers of infected people succumb to HIV/AIDS.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
55
This invariably includes substantial losses in labor,
given the fact that sexually active age groups contain
large numbers of the labor force, a critical input in
national development.
At the household level, heavy dependencies will
be created by the loss of income earners, forcing
governments to divert resources from development
projects to relief programs. Even then the
psychosocial needs of these dependents will remain
unmet because relief effort is limited in this regard
(Walker, 1998). The opportunity cost of HIV/AIDS will
therefore take a multidimensional form given the wide-
ranging nature of its scope.
This contrasts sharply with countries such as the
United States and Australia where high-risk groups
consist mainly of a very small proportion of the
population, gays and intravenous drug users. Even if
prevalence rates had not stabilized, the effects of
the epidemic in these countries would still not be as
far-reaching as it is in developing countries.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
56
Bloom and Carliner (1988) assessed potential economic
costs of HIV/AIDS in the United States and came to the
conclusion that the epidemic would not cost much
relative to normal US medical spending. The same
cannot be said about developing regions of the world,
which are struggling to supply their populations with
the most basic care possible for people living with
HIV/AIDS. The World Bank (1999) estimates that of
the 30 million people who have contracted HIV/AIDS
since the beginning of the epidemic, 90% were in
developing countries.
The fact that no cure has yet been found for
HIV/AIDS poses a big challenge for the whole world in
general, and developing countries in particular. The
future of their populations will depend heavily on
sexual behavioral change among high-risk groups,
because developing nations cannot afford available
treatment cocktails due to prohibitive prices.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
57
The path for health policy in these countries is
therefore clear-cut, to build high HIV/AIDS awareness
among their populations and promote effective HIV/AIDS
preventive health behavior.
3.2 HIV/AIDS in Botswana
The first case of AIDS to be reported in Botswana
occurred in 1985 (UNDP, 2000; NACA, 2002), and since
then prevalence rates have risen at an alarming rate.
In 1992, national prevalence rates were estimated at
just under 15% among pregnant women, but by 1999 they
had skyrocketed to approximately 38% (UNDP, 2000), the
highest in the whole world. An estimated 300,000 out
of 1.6 million Batswana were living with HIV/AIDS in
1999. Worse still, more than one in four people aged
15 to 49 were estimated to be HIV-positive by 1999
(UNDP, 2000).
Predominantly, most HIV transmission in Botswana
occurs through heterosexual sex, but there is also a
significant passage from mother to child (UNDP, 2000).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
58
So serious are these two types of transmission that
government was forced to intervene in two major ways
in the recent past: first, the prevention of mother-
to-child transmission (PMTCT) program was introduced
in 2000 to combat vertical transmission of HIV (UNDP,
2000). Through this initiative, expectant women
testing HIV-positive are given free drugs that stop
passage of HIV from mother to child during pregnancy.
New mothers are also encouraged not to breastfeed
their children following birth.
The second major step is to make antiretroviral
(ARV) therapy available to all HIV-positive citizens,
free of charge. Launched in 2002 and supported by
large American corporations, this program is scheduled
for a gradual roll out from district to district until
the whole country is covered (MFDP, 2002).
Not much is known about transmission resulting
from homosexual contact, mainly because homosexuality
is largely ostracized and draws substantial stigma
(UNDP, 2000).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59
This probably explains the rather uncharacteristic
silence of government policy on this issue, despite
mass media reports to the effect that homosexual
relationships are rife in prisons (Mmegi, 2002) .
Government has put in place strict measures to
curb transmission of HIV through blood transfusion,
starting as early as 1986 (Stegling, 2000). According
to the Ministry of Health (2001), it is government
policy that 'all blood that is donated for transfusion
must be screened for HIV antibodies' (pg 4).
3.2.1 The Impact of HIV/AIDS on the Demography of
Botswana
This section takes a brief look at the effect
HIV/AIDS has had on the demographic make-up of
Botswana, focusing particularly on mortality and
population growth.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
60
Figure 2: Population of Botswana: 1911-2001
1800000
1600000
1400000
g 1200000
^ 1000000
r 0
00000
I—I
i !
3
a
0
01
600000
400000
200000
CkV oT> rtV
Year
Figure 2 shows the population of Botswana as
enumerated at various intervals since 1911. As can be
clearly seen, the population grew quite rapidly in
recent times, especially since 1971. This pattern is
not particularly visible for the earlier years,
probably more a reflection of inaccuracies in the data
than lack of growth; methods applied during these
censuses were crude and largely primitive.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
61
Earlier censuses were conducted by colonial
administrations whose main interest was to estimate
the tax base, thus tended to be biased towards the
labor force. These counts were characterized by
under-enumeration, mainly because no thorough effort
was made to cover all citizens (CSO, 1992) . The first
reliable statistics were collected in 1964, two years
before independence.
Between 1971 and 2001 the population of Botswana
tripled from just over half a million to more than 1.6
million, owing largely to high natural increase
resulting from an huge deficit between fertility and
mortality following a sharp decline in the latter.
This pattern is consistent with the demographic
transition of developing nations whose mortality
levels fell sharply in the middle of the twentieth
century because of modern technology imported from the
developed world, which eliminated common ills
associated with poor hygiene and general ill-health
(Weeks, 1999; Yaukey, 1990; Zopf, 1984).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
62
This led to what some demographers have labeled
the mortality revolution (Easterlin, 1986), because of
the unprecedented drop in the incidence of death. At
the same time fertility remained high in these
countries, leading to a huge rise in population
growth, thus raising fears of what came to be known as
the population explosion.
Figure 3: Total Fertility Rate - Botswana: 1984-19962
C x j
7
6
5
4
3
2
1
0
1984 1988 1996
Year
■ TFR
Source: Botswana Family Health Survey III - 1996
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
63
The large deficit between mortality levels and
fertility triggered some governments to enact
population policies that sought to reduce fertility,
as a means through which to address the growing
population problem. Botswana was one such country,
and even though a formal population policy was not
pursued until the last decade of the twentieth
century, programmes were put in place early in the
post independence era to reduce fertility. In
particular, family planning was integrated into
general health services and widely promoted to enhance
adoption, especially among women of childbearing ages
(MFDP, 1997). As a result, contraceptives such as the
pill, IUD, injection, condoms, etc were provided as
part of the family planning programme.
Figure 3 shows that between 1984 and 1996 Total
Fertility Rate (TFR) fell from well above 6 to just
over 4 children per woman, a decline that probably
emanates from this effort.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
64
3.2.1.1 HIV/AIDS and Mortality in Botswana
HIV/AIDS is a killer disease; it can therefore be
expected that with the world's highest known infection
rate, Botswana should suffer heavy mortality as a
result. Figure 4 shows the number of deaths recorded
12 months before the 1991 census, by age. Interest
here should not be so much on the numbers, but on the
"U" shape of the curve, which conforms to biological
logic that the risk of death is high at younger ages,
especially during infancy, and also at older ages as
individuals approach the end of the human lifespan.
Meanwhile, the ages in between are generally
characterized by low mortality and therefore symbolize
the prime of human existence.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
Figure 4: Number of Deaths 12 Months before the 1991
Population Census by Age3
2000
1600
to
£ 1400
t o
g 1200
^ 1000
< D
X!
3 600
S
400
200
3 3 3 3 3
X
Age _____________
—♦— Number of
____________ Deaths
The above picture contrasts sharply with that
portrayed by Figure 5, which presents the number of
deaths recorded only two months before the 1998 BDS.
3
Source: Botswana Family Health Survey III - 1996
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
66
In addition to the huge difference in the number of
deaths occurring in only two months in 1998 compared
to those in twelve months in 1991, it can also been
noticed that the shape of the curve depicted in 1991
had been distorted by 1998. Specifically, the "U"
shape was replaced by a "W"-type curve indicating a
rise in the incidence of death in middle life.
Figure 5: Number of Deaths 2 Months before the 1998
Botswana Demographic Survey by Age4
-P
f t
O
f-4
000
7000
6000
5000
4000
3000
2000
1000
0
P>
X
Age
-Number of
Deaths
4
Source: Botswana Demographic Survey (BDS) - 1998
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
67
This increase cannot be accidental; in 1991 deaths
resulting from HIV/AIDS had not yet started, probably
due to the fact that this was during the early years
of the outbreak, and also because of the long
incubation period of the disease. By 1998 however,
most infected cases had reached the clinical stage,
leading to an increase in deaths among the sexually
active population, mostly found in the middle ages.
3.2.1.2 HIV/AIDS and Population Growth in Botswana
The above scenario has triggered interest among
demographers on what the impact of the sudden reversal
of gains against mortality over the past several
decades will be in the future, especially with respect
to population growth. The next section assesses some
of the literature available on this issue.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
68
The common approach used by the scant analyses
that have been conducted on this topic compares the
"No HIV/AIDS" and "AIDS No Intervention" scenarios,
with the result that the picture painted depicts two
extremes that, in the case of Botswana at least, are
both impossible since HIV/AIDS is a big problem, and
government is taking bold steps to fight it. In other
words both scenarios do not reflect the situation on
the ground. It would therefore be helpful to view the
figures presented here as indicating both "best case"
(No AIDS) and "worst case" (AIDS No Intervention)
scenarios, with a strong likelihood that reality will
fall somewhere within that continuum as government
efforts take effect.
Figure 6 shows that without HIV/AIDS, Botswana's
population growth rate would have continued to fall,
albeit gently across the years. In fact, between 1991
and 2009 it would have dropped from approximately 2.7%
to 2%, probably owing largely to continued declines in
fertility.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
69
Figure 6: Total Projected Population Growth Rates:
1991-20095
CL)
-p
( 0
■p
s
o
P
o
-0.5
-1
1991 1995 2000 2005 2009
- ♦ — Growth Rate (No AIDS)
•— Growth Rate (AIDS-No
Intervention)________
The "AIDS No Intervention" scenario however
presents a totally different picture, as the country
would achieve zero population growth (ZPG) before
2005. By 2009, the growth rate would be negative.
5 Source: United Nations Development Programme - 2000
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
70
For a nation of less than 2 million people, this is
surely an undesirable result, which gives Botswana
more reason to fight the disease.
Figure 7 presents the projected population, still
under the two scenarios, between 1991 and 2010. As
could be expected, the population would have grown
rather linearly during this period if there was no
HIV/AIDS, surpassing the 2 million mark before 2010.
However, with HIV/AIDS and no intervention, the
population would not grow much beyond 1.6 million,
achieved by 2005, after which it would level off and
ultimately start declining.
The projections quoted here are rather short
term, making it difficult to shed any light about the
far future. This is however not surprising given the
complicated nature of the HIV/AIDS epidemic, and the
changing dynamics created by on-going strategies
against it.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
71
Figure 7: Botswana Projected Population Growth: 1991-
2010 "With No-AIDS" vs "With AIDS versus No-
Intervention" Scenarios6
2500000
2000000
2 1500000
ft 1000000
500000
0
1991 1995 2000 2005 2010
Year
Popn (No
AIDS)
—■— Popn (AIDS)
Perhaps it would be more useful for projections not to
simply polarize analyses into "No AIDS" and "AIDS No
Intervention" scenarios because after all, it is known
that AIDS is a reality, it exists.
6 Source: United Nations Development Programme 2000
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
72
Various interventions have already been put in
place and continually reviewed to combat it. Focus
might need to be channeled towards evaluating various
strategies and their impact on population dynamics
across time, given the fact that various choices are
now available from which governments can choose. For
example, what are the possible effects of
Antiretroviral Therapy (ART) on population growth in
the next ten years, especially taking into account the
fact that the government of Botswana is currently
undertaking a programme to make these universally
accessible to all citizens?
3.3 Botswana's Health Policy
Botswana's health policy is based on the
principle of Primary Health Care (PHC) , adopted at the
Alma Ata Convention in 1978, which stresses disease
prevention and promotes community health. The Health
chapter in the National Development Plan 8, the
current government blueprint, spells out the
philosophy of Botswana's National Health Policy thus:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
73
The government shall, when planning its
activities, put health promotion and care, and
disease prevention, among its priorities, the
basic objectives of which shall be access by all
citizens of Botswana to essential health care,
whatever their own financial resources or place
of domicile, and the assurance of an equitable
distribution of health resources and utilization
of health services... In pursuance of some of the
above objectives, special measures may be taken
in respect of high-risk groups, such as children,
adolescents, pregnant women, the elderly,
disabled persons, and workers whose occupations
justify such measures.7
Effective implementation of this approach
requires, among other things, the development of
infrastructure for easy access to health care
throughout the country and, secondly, provision of
trained manpower to staff the medical facilities. In
this respect, the Primary Health Care strategy has
posed considerable budgetary challenges for
government.
In response, government adopted a hierarchical
approach to health service provision whereby
facilities are divided into nine types, ranging from a
mobile health stop to a national referral hospital,
7 National Development Plan 8(1997/98 - 2002/2003) pg 381
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
74
and these are networked through a referral system.
The distinguishing features between facilities are
physical size (number of beds), and/or the level of
health services provided; for example a health post
has no bed, is manned by one nurse and a Family
Welfare Educator (FWE) and is limited to preventive
and basic curative tasks, but a Primary Hospital Level
II, has between 20 and 70 beds and has X-ray and
surgery capacity (MFDP, 1997).
The hierarchical approach has allowed government
to reach many parts of the country by expanding basic
health services in the form of facilities and
manpower. In 1995 98% of the urban population lived
within 15 kilometers from the nearest health facility,
while the rural population was not too far behind with
88% (MFDP, 1997).
Lower level facilities have been expanded
immensely over the past two decades or so, while at
higher levels focus has been predominantly on
increasing manpower supply, especially medical
doctors, dentists and nurses.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
75
For example, between 1985 and 2000, the number of
mobile stops jumped from 375 to 712, while health
posts increased from 270 to 324. During the same
period of time only 8 additional hospitals were built,
bringing the total to 17 (CSO, 2001). However, during
this period the number of doctors increased
significantly from 1.7 doctors per ten thousand
population in 1985 to 2.9 per ten thousand population
in 2000, while the ratio of Family Welfare Educators
(FWEs) to the population dropped from 5.6 in 1985 to
4.7 per 10 000 population in 2000.
Effectively, this means that between 1985 and
2000 the provision of doctors was given higher
priority over that of health extension personnel such
as FWEs. While this might not necessarily reflect a
shift in policy focus from prevention to curative
strategies during the said period, it indicates hard
choices that the country has had to make with respect
to allocation of resources to social development. In
an ideal situation, provision of one service should
not adversely affect that of other related services.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
Since disease prevention takes the center stage in
PHC, expectation could be that provision of FWEs and
other frontline staff would be stepped up during this
period to reflect the critical role played by health
education in fighting the HIV/AIDS epidemic.
Another key feature of Botswana's health policy
is the Maternity/Child Health/Family Planning Program
(MCH/FP). Interestingly, this program was triggered
by a group of women who felt the need for family
planning services in the early post independence
period and approached government for help (CSO/MFDP;
1989; MOH, 1985) . By 1973 a full-fledged MCH/FP Unit
was established in the Ministry of Health with a goal
to promote health through participation at the home
level, again using FWEs as key agents of change (CSO,
1989) . The MCH/FP Unit was combined with Nutrition
Unit and Health Education Unit in 1979 to form what is
now the Family Health Division.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
77
Perhaps the most important point to note here is
that Family Planning has been an integral component of
health service provision from the early days of the
establishment of Botswana as an independent state, and
that an integrated approach was preferred whereby
family planning services were provided as part and
parcel of general family health instead of an isolated
service.
This project will return later to assess
attitudes towards and adoption of modern
contraception, particularly condoms, against this
background, rather than view it purely as a response
to the HIV/AIDS epidemic.
3.4 Botswana's Response to HIV/AIDS
Like in many parts of the world, Botswana
government's response was not prompt, which has been
blamed for the rapid spread of the disease in some
quarters.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
78
When the intervention was finally instituted, it came
in the form of short term plans, probably based on a
mistaken belief that HIV/AIDS was a temporary problem.
For example, the National AIDS Control Program
(NACP)'s Short Term Plan concentrated on the two-year
plan period 1987-1989, and was followed by Medium Term
Plan I, which focused on 1989-1993. Both plans
emphasized information, education and communication
(IEC) as key strategies to combat HIV/AIDS. By 1993
it began to dawn on the government that the AIDS
epidemic was not going to disappear solely on the
basis of IEC, and it is then that the Botswana
National Policy on HIV/AIDS was conceived UNDP, 2000).
Key features of the national policy on HIV/AIDS
are as follows: prevention of new infections, reducing
personal and psychosocial impacts of HIV/AIDS,
intersectoral mobilization around HIV/AIDS, provision
of care for people living with HIV/AIDS, and reduction
of socioeconomic consequences of HIV/AIDS (UNDP,
2000).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
79
This policy is more holistic in content and approach,
and has striking similarities with the principle of
Primary Health Care, which emphasizes disease
prevention and health promotion, without overlooking
treatment and care of already existing cases of
disease. The prevention dimension of the policy also
intersects considerably with the Family Planning
program, especially because both advocate, among other
methods, condom use. In this light the AIDS policy can
be seen to be building on already existing health
practices.
A multisectoral response recognizes the fact that
the epidemic is not the sole responsibility of the
health sector, and to effectively control its spread
needs concerted efforts of people from all walks of
life, literally, ranging from government departments
to the private sector and Non-governmental
Organizations (NGOs).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
80
This project focuses on the prevention dimension
of the national policy, because, like the UNDP (2000)
correctly argues, Botswana's main "..window of hope is
the 80% of the entire population, about 1.28 million
people, who are free of HIV infection" (pg 1).
Specifically, attention is focused on identifying
constraints to prevention of infection among people of
various socioeconomic characteristics with a goal to
ultimately remove these constraints by mainstreaming
them into policy targets.
3.5 HIV/AIDS-Related Knowledge, Attitudes, and
Practice
It is important for the study to take a close
look at knowledge about HIV/AIDS, attitudes towards
people living with the disease, and practices relating
to preventive action. This puts the project into
perspective and lays the foundation for all analyses
that follow in later chapters.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
81
Earliest official response to the HIV/AIDS
problem was to mount countrywide information,
education and communication campaigns (IEC) in a bid
to create awareness about the disease (UNDP, 2000).
This resulted in high levels of awareness regarding
the existence of HIV/AIDS, as consistently reflected
by available statistics. As early as 1988, 88.3% of
all women aged 15 to 4 9 had heard about AIDS (MFDP and
MOH, 1989). By 1994 awareness levels had reached over
90% (MFDP, 1997). In 2001, the level of awareness had
risen to 94% among people aged 10 and 64, even though
males reported significantly lower awareness at 91%
relative to females' 95% (CSO, 2002).
In 1988, out of women aged 15 to 4 9 who had heard
about AIDS, only 42.2% of them mentioned condoms as a
way to avoid contracting the disease (CSO, 1989).
Thirteen years later 75% of 15-19 year olds, 80.5% of
20-30 year olds, and 69.1% of 31-49 year olds
mentioned the condom. Even 36.2% of 10-14 year olds
were aware that condoms were an effective way to
protect against HIV infection (CSO, 2002).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
82
Not much statistics are available on attitudes in
Botswana, largely due to the absence of comprehensive
monitoring mechanisms for the IEC campaign launched
more than a decade ago. Attitudes are critical
because they form the basis for stigmatization of
HIV/AIDS. As Anneke and Mhone (2001) caution, 'HIV
can evoke a particularly strong stigma response
because it is a sexually communicable disease, is
often fatal, has a mysterious origin and has been
associated with groups that were already stigmatized
in other ways' (p. 11). Basset and Mhloyi (1991)
argue further that the early focus of most AIDS
research on female prostitutes was fueled by the
mistaken belief that women were the only critical
vector of the HIV/AIDS outbreak.
The immediate risk attached to stigmatization is
the increased possibility of 'driving the problem
underground' (MOH, 2001), since PLWAs may get too
scared to come out in the open about their status in
fear of social exclusion and other forms of
discrimination.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
83
However, there are indications that attitudes
towards people living with HIV/AIDS are not
excessively discriminatory in Botswana. For example,
50.1% of the people aged between 10 and 64 felt that
an HIV-positive teacher must be allowed to continue
teaching if they are not sick. Further, 81.7% said
they were willing to care for a family member inside
their households if they became ill with HIV/AIDS
(CSO, 2002).
These results may be related to the fact that a
sizeable portion of respondents in the survey (27.9%)
knew someone who is HIV-positive at the time of the
interview, which might contribute to general
acceptance of the disease. There are also indications
that the number of people getting voluntarily tested
for HIV is increasing (UNDP, 2000), presumably in
response to the apparent acceptance of PLWA.
Not much information is available on HIV/AIDS-
related practices in Botswana, especially sexual
behavior.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
84
Again, this is probably due to the absence of
monitoring systems, but it may also reflect the
sensitive nature of the data required for such
insights.
The anti-AIDS campaign preaches what has come to
be known as the ABCs of HIV/AIDS prevention: abstain,
be faithful, and condom use (UNDP, 2000). However,
high rates of infection despite near universal
awareness of HIV/AIDS suggest that knowledge has not
been translated into preventive behavior.
The 2001 sentinel surveillance results indicate
that median crude HIV prevalence among pregnant women
of all ages in six districts jumped from 38.5% in 2000
to 42% in 2001. During the same period the number of
15-49 years olds infected with HIV is estimated to
have jumped from 277 637 to 317 639 (NACA, 2001), an
increase of more than 14%. Persistence of new
infections suggests that sexual practices have not yet
changed; in other words the ABCs of HIV/AIDS
prevention still eludes the sexually active population
in Botswana.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
85
UNDP (2000) however cautions that '...the ultimate
indicator of success for any HIV and AIDS prevention
programme is the extent to which rates of new
infections have been reduced' (pg 2).
3.6 The Dynamics of HIV/AIDS in Botswana
Dynamics fuelling the AIDS epidemic in Botswana
form the heart of this project. It is indeed
interesting that up to this day, not much is known
about factors that have been interacting with so much
intensity to make Botswana the world's worst HIV/AIDS
case to date. What is known comes more in the form of
descriptive-level analyses and essays, with little or
no empirical backing whatsoever. Further, not much
effort has been made in this regard to apply social
theory to the Botswana context to facilitate in-depth
analysis of forces underlying the epidemic.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
86
Similar deficiencies have been observed elsewhere
in the world, for Vanlandingham et al (1995) deplore
the absence of theoretical perspectives in most social
scientific enquiry on HIV/AIDS risk-taking behavior.
He and his colleagues are wary of simplistic analyses
consisting of ' . . . a melange of factors ranging from age
to socioeconomic status' from which '...results are
frequently presented as inventories of statistically
significant associations' (pg 196).
The need for in-depth analysis of factors behind
the HIV/AIDS in Botswana has however been recognized,
though not so widely. Stegling (2000) acknowledges
the general lack of knowledge concerning the cultural
and social factors influencing the spread of HIV/AIDS
in Botswana, and identifies it as an area in need of
further research. That challenge is taken up by the
current research project.
Of the many assertions made about HIV
transmission in Botswana, one that appears
consistently in the literature is gender inequality
(Lesetedi, 1999; Stegling, 2000; UNDP, 2000).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
87
This factor seems to operate in a multidimensional
manner because women in Botswana are disadvantaged in
many respects. First, even though things have changed
remarkably in the recent past, women were
traditionally considered minors and '...subject for life
to the authority of male guardians' (Schapera, 1976;
p. 37) . At the household level, this has left women
with no property rights, and dependent almost entirely
on their husbands or guardians. A good example is the
much contested legal provision known as the Married
Persons Act, under which husbands married in community
of property are recognized as sole administrators of
their family's property (Women's Affairs Division,
undated).
At the national level, gender inequality has kept
women out of decision-making positions for ages,
debarring them from political participation and other
institutions of key social significance. For example,
as recently as 1979, there was only one woman holding
a cabinet ministerial post, and she was the only woman
representative in a parliament consisting of 38 seats.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
88
By 1994 things had improved only marginally, as the
number increased to 4, two of them holding cabinet
posts in a parliament with 43 seats (CSO, 1998).
While issues of women empowerment have been
brought to the center stage of political debate over
the past two decades or so, inequalities created over
centuries have persisted. The absence of women in
meaningful political participation not only maintains
the status quo, it also slows down change, as men
cannot be in a hurry to implement policies that
threaten their privilege. Stegling (2000) warns that
even though the current AIDS Policy recognizes that
'empowerment of women and reduction of poverty are
crucial elements to fight the (HIV/AIDS) epidemic' (p.
5) it does not specify how this will be achieved.
The second dimension of gender inequality takes
the form of violence against women. Stegling (2000)
asserts that the problem of violence against women is
on the increase, and suspects that a high proportion
of most young women's first sexual experience is
tantamount to rape.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
89
Lesetedi (1999) argues that women in Botswana are
socialized to be sexually submissive. This leaves
women in a vulnerable position, with virtually no
bargaining power in sexual relationships, and
therefore heightens their risk of infection. Basset
and Mhloyi (1991) echo the same concern about the
situation of women in Zimbabwe, arguing that for most
women, sexual decision making is a man's domain, and
that women who refuse sex to their men have to choose
between 'social death' and 'biological death', i.e.
between death due to poverty or due to HIV/AIDS
infection.
Closely tied to violence is the issue of
intergenerational sex. Women in Botswana, especially
at the younger ages, have been found to engage in sex
with relatively older men (Lesetedi, 1999; UNDP,
2000). Basset and Mhloyi (1991) observed a similar
trend in Zimbabwe, while Idele-Akwara (2002) has made
the same observation in Kenya.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
90
Older men are likely to be more sexually experienced
and therefore bring greater risk of infection, this
way ensuring passage of the HIV virus from one
generation to the next. UNDP (2000) has been
particularly vocal against this trend and calls for a
"social revolution" where society is called upon to
despise such practices so as to save the young
generation. This is part of what has been called "the
three No's" (pg 3), (No Sex before HIV testing, No sex
without a condom, No sex outside own cohort).
Marital status is another dimension through which
gender inequalities function to influence the spread
of HIV/AIDS in Botswana. Specifically, married women
have been shown to face greater risks of infection
compared to single women (Machacha, 2001). This is
closely tied with the social status of women,
particularly the fact that women are generally
regarded as minors in marriage set-ups, which deprives
them control over decision making in the household,
including decisions about sexual relations with their
husbands.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
91
In addition to a socialization process that
propagates sexual submissiveness for women, Setswana
culture has also been shown to promote promiscuity on
the part of men, especially within marriage. Machacha
(2001) cites a common passage rite normally carried
out at weddings, whereby elderly married women
encourage the bride not to question her husband's
whereabouts regardless of how suspicious they might
be, for the sake of marital stability. The groom, on
the other hand, is taught by elderly men not to
'forget' his wife and children, which suggests that he
could have other relationships, as long as he provides
for and does not neglect his family.
The social system is therefore structured to
favor men not only in sexual decision-making but also
tends to absolve them of any accountability in terms
of sexual responsibility within marriage. Marital
status therefore becomes critical as an intervening
variable in assessing HIV preventive behavior.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
92
The essence of the aforegoing review is to show
the disadvantaged position of women in life in
general, and in sexual decision-making in particular.
All the stated factors seem to operate in concert to
place women in a precarious position with regard to
HIV infection. The sex variable as well as the age
difference between partners are therefore critical to
this analysis because they pose a strong potential to
modify the relationship between women's perceptions of
the HIV/AIDS and how they ultimately respond to it.
Another factor that has been linked to the spread
of HIV/AIDS in Botswana is the high mobility of the
population (Stegling, 2000; UNDP, 2000). Botswana has
a long history of labor migration, dating back to the
colonial era when most of Southern Africa served as a
labor pool for South African gold mines. This
migration stream has however declined, mostly due to
the rise of Botswana as a dynamic and rapidly growing
economy. Between 1977 and 1981 for example, Botswana
citizens resident in South Africa dropped from 43 159
to 23 200 (CSO, 1987), a whopping 46.2% decline.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
International labor migration has been replaced
by a set of internal migration streams, going
different directions and fueled by different forces.
For example, between 1980 and 1981, 7.36% of between-
district migration was from urban to urban, 27.54% was
from urban to rural, 36.46% was from rural to urban,
and 28.64% was from rural to rural areas (CSO, 1987).
What has not changed however is the separation of
couples that migration causes. It is not unusual for
couples to work miles from each other, only getting
together occasionally. This has been blamed largely
on government's policy of decentralized service
provision (Stegling, 2000). Botswana's government is
a major employer, and despite its good intention to
bring services such as health, education, etc closer
to the people, it invariably takes personnel away from
their families.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
94
This heightens chances of HIV infection because it
increases the risk of extra-marital sexual relations,
especially in a society such as Botswana in which
there is social acceptance of multiple sexual
relationships and childbearing out of wedlock
(Stegling, 2000).
It is therefore of great importance for this
project to take into account the proximity of sexual
partners in terms of residence, regardless of marital
status.
Age is a critical factor in HIV/AIDS
transmission. As already seen, intergenerational sex
has been identified as one of the risk factors for
infection. It is important therefore to compare
sexual practices of people who engage in sex with
partners much older than themselves to see if they are
indeed less likely to take preventive measures.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
95
Recent statistics show that HIV/AIDS awareness is
higher in urban than in rural areas (CSO, 2002). Even
though Stegling (2000) argues that it is difficult to
differentiate between 'urban' and 'rural' residents in
Botswana because of high population exchange between
the two, it is still true that most people spend most
of their time in one and not the other place. In
fact, the 1991 Population and Housing Census
classified as much as 86.4% as non-migrants, even
though it does not specify the reference period (CSO,
undated). The observed deficit between urban and
rural areas may be a reflection of the uneven
distribution of health and other facilities as well as
personnel between the two, which more often than not
favors urban areas. This is likely to impact on the
effectiveness of the anti-AIDS campaign in the two
areas, introducing differentials in the resultant
levels of awareness. It is important therefore to
control for the effects of this variation in assessing
the effects of various perceptions on health behavior.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
96
Socioeconomic status has also been linked to
fluctuations in preventive health behavior.
Specifically, it has been argued that poor people
engage in behaviors that expose them to heightened
risks of HIV/AIDS infection (UNDP, 2000; Letamo and
Bainame, 1997). Women have been identified as a
particularly vulnerable group here, largely due to
their disadvantaged position in society as seen
earlier.
This project will factor socioeconomic status
into the analysis, as measured by educational
attainment. Socioeconomic status is expected to
directly vary with the probability of adopting
preventive health measures.
3.7 Restatement of Hypotheses
Having reviewed the literature on the dynamics of
HIV/AIDS in Botswana, and cultivated an understanding
of the underlying factors, this thesis ventures to
restate the original hypothesis.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
97
The main purpose of this exercise is to factor
control variables into the models so that their
influence on the original relationships postulated
under the HBM may be accounted for, given what is
already known about them.
3.7.1 Condom Use
People who feel susceptible to HIV/AIDS, perceive
it as severe, perceive less costs and high benefits in
condom use as a measure of HIV prevention, and are
confident about self-efficacy of use are more likely
to use condoms. However, this relationship will be
mitigated by age, gender-based variables (sex, age-
difference between partners, and marital status),
socioeconomic status (SES), residence, and the
availability of information on the disease.
Age is a critical variable in the sense that
different age groups may respond differently to the
same preventive technology such as the condom.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
98
For example, while the young generation may easily
accept condoms as part sexual activity, older
generations may find this difficult because they have
been exposed to a life in which condom use has never
been so necessary. To older groups condom use may
require a complete change of lifestyle, while for
those at the beginning of their sexual lives it is
simply part of the sexual act. This project therefore
hypothesizes that condom use will be higher among
young people relative to older ones.
With regard to gender relations much as women
could want to use condoms, they might not find this
easy to practice given their weak position in
negotiating safe sex, relative to their male
counterparts. Also age difference is likely to affect
condom use because as argued in the literature,
intergenerational sex is one of the key means through
which HIV is transmitted. Elderly men are especially
cited for a tendency to have sexual relations with
younger women, in return for financial and other
favors.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
99
Condom use is apparently not common in such
relationships. This variable is included in the
analysis with the expectation that it will be a
significant determinant of condom use. To put the
nature of its sexual selectivity on the spotlight
(older men's preference for younger women
predominantly), an interaction between age difference
between partners and sex of the respondent is also
introduced into the analysis.
Marital status is another variable that is likely
to impact on the relationship between perception
variables and condom use. Marital unions are
characterized by gender inequalities favoring men.
Generally, men are absolved of many responsibilities
within the household, including those pertaining to
responsible sex. This tendency is expected to extend
beyond sexual activity within marriage. It should be
borne in mind that condom use as measured in this
analysis is not restricted to sexual activity within
marriage.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
100
Marital status is also expected to interact with sex
of the respondent in the sense that women, though
married, may not have much say in sexual decisions
relative to their male counterparts.
As seen earlier, SES also has a bearing on the
adoption of preventive behavior. Some researchers
have argued that poor people tend to engage in risky
behaviors than their relatively well-off counterparts.
It could be expected therefore that the adoption of
condom use will be partially explained by this factor,
with high SES more likely to use a condom relative to
low SES.
In the same vein, condom use is likely to vary
between urban and rural dwellers, largely because of
the observed variation in awareness levels between the
two. This relationship is however not expected to
cause much reduction in the relation between the
original HBM variables and condom use because of the
high exchange of population between rural and urban
areas in Botswana.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
101
Availability of information is also likely to
explain part of the adoption of safe sexual practices,
in this case condom use. It is expected that people
who have easy access to information about HIV/AIDS
will be more likely to use condoms than those who do
not have much access.
It is therefore important to control for these
factors when testing the HBM in this context, so that
the relationships specified under the model are not
confounded by external factors.
3.7.2 HIV/AIDS Testing
People who feel susceptible to HIV/AIDS, perceive
it as severe, perceive no costs of getting a test,
high benefits in getting tested, and are confident in
self-efficacy towards getting tested are more likely
to get an HIV/AIDS test. However, this relationship
will be mitigated by age, sex, marital status,
socioeconomic status (SES), residence, and the
knowledge of a facility where a test can be obtained.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
102
Since not much literature exists on HIV testing,
this model essentially ventures into untested waters
with respect to some of the variables it employs. For
example, it generally expects that younger individuals
will be more likely to get an HIV test than older
people since they have a greater stake in life than
the former - the fact that HIV/AIDS poses a greater
danger in cutting their lives relatively shorter
should motivate them to check their status.
Marital status is also expected to have some
influence in determining whether or not one gets an
HIV test, with married people showing less likelihood
to get a test than those who are not married. This
relationship is however likely to interact with sex,
in the sense that married women could be expected to
be less likely to get a test. The reason behind this
logic is the same as that advanced for failure to use
condoms - that married women do not have the power to
turn their wishes into action since they need the
consent of their husbands, who may not approve of HIV
testing.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
103
Socioeconomic status is also expected to play a
role in HIV testing, with people of higher SES showing
more likelihood to get a test than relatively lower
SES. Generally, high SES is tied closely with high
educational attainment, which translates into higher
level of one's understanding and appreciation of the
world around them, therefore enhancing their chances
of conformity to medical messages than those in the
lower SES group.
Urban dwellers are more likely to get an HIV test
than their rural counterparts, owing largely to an
uneven distribution of testing centers in the country.
This model also introduces a structural variable
- whether one knows where to get an HIV test. It is
expected that those who know where to get a test will
show a higher likelihood to get it than those who do
not. This variable is expected to interact with place
of residence since testing facilities are unevenly
distributed between urban and rural areas, tending to
favor the former.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
104
3.7.3 Number of Sexual Partners
People who feel susceptible to HIV/AIDS, perceive
it as severe, perceive less costs and high benefits
associated with having one partner, and are confident
about self-efficacy towards having one partner are
more likely to be involved with one person at a time.
However, this relationship will be mitigated by age,
sex, SES, residence, whether one lives in the locality
with their partner, and the amount of discussion about
HIV/AIDS occurring in their community.
Age is expected to have an inverse relationship
with number of sexual partners, the logic being that
younger people are generally more sexually active, and
may therefore be reasonably expected to have more
partners. Studies have already shown that in Botswana
while the average at age first marriage is about 26
years, the average age at first sexual intercourse is
16.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
105
This means that people on average have ten years to
experiment with different partners before they make
lifetime commitments. In the process they make also
be tempted to experiment with multiple relationships.
Sex is another variable that is expected to have
a bearing to the number of sexual partnerships. As
already seen, gender relations in Botswana favor men
to the extent that even the socialization process
absolves them of sexual responsibility. Specifically,
Setswana culture does not ostracize promiscuity,
especially among men, even within marriage. Men are
therefore expected to show a significant likelihood to
engage in multiple sexual relationships than women.
Socioeconomic status is also expected to have a
significant bearing on the number of sexual partners.
A relationship between multiple sexual partners and
poverty has already been postulated in the literature,
as battering sex can be a means for the poor to make
ends meet, subjecting them to sexual exploitation by
the relatively well off.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
106
Women are particularly vulnerable in this regard,
as struggling single mothers, prostitutes, or young
girls attracted to the financial prowess of "sugar
daddies". An interaction effect between sex and SES
will therefore be introduced into the model to check
for this relationship.
Residence is another variable that might
influence the relationship between multiple sexual
relationships and variables specified under the HBM.
Rural areas are generally viewed as cultural
sanctuaries where norms and values of old are still
observed and practiced. On the other hand, cities and
towns are generally regarded as the centers of
cultural rot, where "Western" living has eroded
traditional values. For example, hardly any
prostitutes can be found in rural areas, as is the
presence of gays and lesbian populations. It can
therefore be expected that multiple relationships will
be significantly pronounced in urban than rural areas.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
107
Separation of partners, especially married
couples, has been blamed for the spread of HIV/AIDS in
Botswana, as shown in the literature. The argument
here is that when couples live apart, they face great
temptation to find sexual partners in areas within
close proximity, which in turn increases their risk
for HIV infection. As a result, this project expects
that people who do not live together or within the
same area as their partners will be more likely to
have multiple sexual relationships than those who do.
Multiple sexual relationships pose one of the
greatest risks for HIV infection; therefore it is
assumed that awareness of this risk could trigger
behavioral change. Awareness, however, is a function
of many factors, one of which is the openness of the
community within which one lives with regard to
discussing issues such as promiscuity. This dimension
is captured in the model, and it is expected that
people who engage in discussion within their community
will be less likely to have multiple relationships
than those who do not.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
108
CHAPTER 4
Trends in HIV/AIDS Knowledge, Attitudes
and Practice
This chapter presents trends in knowledge,
attitudes and practices related to HIV/AIDS, comparing
the situation in 1988 to that in 2001.
1988 marks the "early days" of HIV/AIDS in
Botswana, given the fact that the first case of
HIV/AIDS was reported in 1985. It represents a period
during which there was not as much information as
there is today, a period also characterized by laxity
in terms of government intervention. Indeed by 198 8
hardly any HIV/AIDS-related deaths had been recorded
in Botswana, therefore by then the reality of the
epidemic had not yet struck most citizens.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
109
On the other hand, 2001 represents a period
during which the country had begun to experience heavy
losses in terms of human life due to HIV/AIDS; also,
by 2001 public messages regarding HIV/AIDS
transmission and prevention methods were commonplace.
Perhaps even more important is the fact that
government's position on the issue was more committal,
with more resources devoted to fight the disease.
1988 data is obtained from the Botswana Family
Health Survey II, specifically from the country report
prepared as part of the Demographic Health Surveys
(DHS) programme. For 2001, data was obtained from the
Botswana AIDS Impact Survey (BAIS). The former
sampled only women aged between 15 and 4 9 years;
therefore to facilitate comparison women of the same
age bracket were sieved out from the BAIS for this
purpose.
This analysis is limited by data availability,
and only presents results for which the same questions
were asked in both surveys.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
110
Specifically trends in knowledge, attitudes and
practice are presented here, by age, educational
attainment and residence (urban versus rural) wherever
possible.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I l l
4.1 Knowledge of HIV/AIDS and Prevention
Figure 8: Percent Distribution of Women Aged 15-4 9 who
had ever heard about HIV/AIDS by Age; 1988 vs 2001
01988
■ 2001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
112
Figure 8 depicts awareness about HIV/AIDS by age.
The most striking observation to be made here is the
fact that awareness has always been high among
Botswana women aged 15 to 49. Even as far back as
1988, the proportion of women reporting awareness
ranged from up of 70% to just over 90%. Also, the
figure shows that these rates have increased between
1988 and 2001, with awareness nearing universality in
2001.
Another striking trend here is the shift in
variation by age between the two periods; in 1988,
younger women showed higher awareness of the existence
of HIV/AIDS than their older counterparts; in 2001
this pattern is rather distorted, showing no
outstanding differences between the age groups. This
is probably due to intensified public campaigns over
the years to create awareness about the AIDS epidemic,
and also due to the fact that the same young
generation of women observed in 1988 grew up with this
knowledge over the 13 years.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
113
Awareness varies directly with educational
attainment, as can be seen from Figure 9. In other
words awareness tends to be lower among people with
lower educational levels and higher among those with
higher educational levels. This relationship remained
unchanged even in 2001. Generally, rates are higher
for 2001 compared to 1988, among people of the same
educational level, and this is truer for lower levels
of education.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
114
Figure 9: Percent Distribution of Women Aged 15-49 who
had ever Heard about HIV/AIDS by Education: 1988 vs
2001
102
100
98
4J 94
c
CD
o 92
H
CD
Oj 90
84
Incomplete Complete Secondary
Primary Primary or Higher
Education
1988
2001
Figure 10 presents awareness by residence; whether one
resides in an urban or rural area. Generally,
HIV/AIDS awareness is higher in urban than in rural
areas. However, the difference was more pronounced in
1988 than in 2001.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
115
Figure 10: Percent Distribution of Women Aged 15-49
who had Ever Heard about HIV/AIDS by Residence: 1988
vs 2001
100
-p
£
0 )
O
£
( D
60
40
20
Urban Rural
Residence
1988
2001
It is quite possible that in the near future this
difference may disappear altogether, judging from its
shrinkage between 1988 and 2001.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
116
Having established differentials in knowledge by
the above factors, this thesis goes further to
establish awareness regarding prevention methods,
again in 1988 and 2001. Specifically, analysis
focuses on condom use, limiting the number of sexual
partners, and non-use of "dirty" needles as means
through which transmission of the AIDS virus can be
contained. Again, the scope of analysis is limited by
the availability of comparable variables for the two
time periods.
Figure 11 shows that between 1988 and 2001
there was a massive increase in awareness regarding
the effectiveness of the condom as an effective
prevention method against HIV/AIDS. While awareness
remained highest among 15 to 29 years olds for both
time periods, the difference between them constitutes
nothing less than a revolution. However, this is not
surprising because public campaigns promoting condom
use to prevent HIV/AIDS have been vigorous over the
years.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
117
Figure 11: Percent Distribution of Women aged 15-49
who had Ever heard of HIV/AIDS and Believed Condom Use
is an Effective Prevention Method: 1988 vs 2001
15-19 20-24 25-29 30-34 35-39 40-44 45-49
BCondom 1988
■Condom 2001
A glimpse at Figure 12 could suggest that
Botswana women aged between 15 and 4 9 years were more
aware and apprehensive of the danger posed having
multiple sexual partners in 1988 than in 2001.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
118
However, this is quite unlikely given the amount of
knowledge that has been accumulated between these two
time periods concerning this particular risk factor.
The above figure most likely depicts the clarity of
public campaign messages that has occurred with the
passage of time: "limiting" the number of partners
does not necessarily imply having one partner, and it
would seem this difference was not clearly understood
in 1988, whereas in 2001 it was.
Figure 13 clearly demonstrates this fact. By
2001, people understood the difference between
limiting the number of partners and having no
relationships outside the current one, and were aware
that the latter was what it took to avoid contracting
HIV/AIDS.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
119
Figure 12: Percentage of Women Aged 15-4 9 who had ever
Heard about HIV/AIDS and Believed "Limiting partners"
is an Effective Prevention Method by Age: 1988 vs 2001
90
80
70
60
-p
9 . 50
CD
O
S-l
CD
O j
40
30
20
10
0
_
J 1
15-19 20-24 25-29 30-34 35-39 40-44 45-49
Age
ILimit Partners
1988
ILimit Partners
2001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
120
Figure 13: Percentage of Women Aged 15-49 who had Ever
Heard about HIV/AIDS and believed that "Limiting
Partners is an Effective Prevention Method" versus
Both Partners Have No Other Partners" by Age: 2001
u 20
15-19 20-24 25-29 30-34 35-39 40-44 45-49
A9e □Limit Partners 2001
IBoth Partners Have
No Other Partner
2001
Figure 14 shows that awareness regarding risk of
HIV infection due to use of "dirty" needles was higher
in 1988 compared to 2001, across all age groups.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
121
Indeed this is a surprise finding, given the fact one
would expect the reverse. However, "dirty" needles
are associated with intravenous drug use, which is not
common in Botswana.
Figure 14: Percent Distribution of Women Aged 15-49
who had Ever Heard about HIV/AIDS and Believed
Avoiding "Dirty" Needles is an Effective Prevention
Method by Age: 1988 vs 2001
20
18
16
14
12
10
15-19 20-24 25-29 30-34 35-39 40-44 45-4
Age
□Avoid "Dirty
Needles 1988
■Avoid "Dirty
Needles 2001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
122
The above outcome is therefore probably a
reflection of the fact that in 1988 public messages
were very general, covering all possible transmission
routes, considering the fact that HIV/AIDS was a new
public health problem, but by 2001 messages were
focused on transmission channels common in Botswana.
This difference may also reflect lack of clarity in
the term "dirty" needles, and that the public was more
aware of this problem in 2001 as awareness grew and
messages became clearer relative to 1988.
4.2 Attitudes towards People Living With HIV/AIDS
This section traces attitudinal changes between
1988 and 2001, regarding people living with HIV/AIDS.
Figure 15 below shows a vast change in attitudes
during the thirteen years, occurring across all age
groups.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
123
Whereas in 1988 the modal percentage of women aged 15-
49 who felt that a person with HIV/AIDS should
continue going to school or continue teaching was less
than 30%, being the age group 45-49, by 2001 this
proportion had increased to as high as to 70%, being
among 25-29 year olds. In fact, this proportion did
not fall below 50% for any age group in 2001.
Figure 16 focuses on attitudinal change by
residential status, whether urban or rural.
Consistent with the picture already shown above, there
is indication for more tolerance of people living with
HIV/AIDS (PLWAs) in 2001 than 1988, and this is true
for both urban and rural areas. However, it appears
there is a bigger difference in tolerance for PLWAS
between urban and rural areas in 2001 than in 1988 .
In other words, in 1988 tolerance for PLWAs was almost
equal between the two types of residence, despite the
fact that it was relatively low, while in 2001 the
difference is visibly pronounced, in spite of greater
levels of tolerance.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
124
This suggests the existence of an "urban bias" in the
concentration of messages seeking to promote
acceptance for PLWAs.
Figure 15: Percent Distribution of Women Aged 15-4 9
who believed that a person with HIV/AIDS should
continue going to school/teach by Age: 1988 vs 2001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
125
Figure 16: Percent Distribution of Women Aged 15-49
who believed that a person with HIV/AIDS should
continue to go to school/teach by Residence: 1988 vs
2001
90
80
70
60 A
+j
G 50
0 )
o
o 40
Oj
30
20
10
0
Urban Rural
Residence
H 1988
2001
With regard to education (Figure 17) there is a
clear indication that attitudes towards people living
with HIV/AIDS have improved between 1988 and 2001.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
126
Perhaps most interesting is the fact that tolerance
increases with education, and this pattern gained more
prominence during the period under investigation,
becoming more discernible in 2001 relative to 1988,
especially between people with incomplete primary and
those with complete primary education.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
127
Figure 17: Percent Distribution of Women Aged 15-49
who believed that a person with HIV/AIDS should
continue to go to school/teach, by Educational
Attainment: 1988 vs 2001 2001
80
70 -
60 -
50 -
-p
c
( U
O 40 -
M
( D
CU
30 -
20 -
0 -f- -----— i —
Incomplete
Primary
Educati
Complete Secondary
Primary or Higher
onal Attainment
■ 1988
■ 2001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
128
4.3 Sexual Behavior/Practices
As indicated earlier, concern has been expressed
in various quarters over persistence of unsafe sexual
behavior in Botswana despite high HIV/AIDS awareness
levels. This section looks into trends in sexual
practices, starting with a general overview of use of
modern contraception before zeroing in on condom use
as a protective measure against HIV/AIDS. It has
already been noted that Botswana's Family Planning
programme has long propagated modern contraception as
a means to limit fertility before the advent of
HIV/AIDS. It is against this background that analysis
in this section is approached.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
129
Figure 18: Percent Distribution of Women Aged 15-49
years Currently Using Modern Contraception: 1984,
1988, and 1996
60
50
40
-p
a
a >
o 30
M
o
Pj
20
10
0
15- 20- 25- 30- 35- 40- 45-
19 24 29 34 39 44 49
□ 1984
Age
■ 1988
□ 1996
Figure 18 shows a steady increase in the
proportion of women using modern contraception between
1984 and 1996. It can also be seen from this figure
that the pattern of use by age has been consistent,
with low use rates at lower and higher ends of the age
spectrum and high use rates for middle ages, notably
between age 20 and 44.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
130
Figure 19 takes a deeper look into contraceptive
use, focusing on four types of contraceptives to see
how their use has developed over the reference period.
Specifically, the pill, Intrauterine device (IUD), the
injection, and condom are brought under the spotlight.
Selection for these methods is based on the fact that
they were found to be consistently more popular than
other methods not covered here.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
131
Figure 19: Percent Distribution of Women Aged 15-49
years by Current Use of Specific Contraception Method;
1984, 1988, and 1996
o 10
1984 1988 1996
□ Pill
■ IUD
□ Inj ection
□Condom
It is clear from the above figure that the pill
remains the most popular contraceptive method amongst
women aged 15-49 years in Botswana. However, between
1988 and 1996, no growth was recorded for this
contraceptive as it stagnated at 17.7%.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
132
This is important because of the fact that between
1984 and 1988, a much shorter space of time, use rate
for the pill jumped from 8.5% to 17.7%. Similarly,
use rates for IUD dropped between 1988 and 1996, after
a notable increase between 1984 and 1988.
Interestingly, during the same period, condom use
grew substantially from 1% in 1984 to 11.4% in 1996.
As can be seen, the condom comes only second to the
pill in 1996, and is the only method, save for the
pill, to ever cross the 10% use rate mark. The fact
that most of the growth in condom use occurred between
1988 and 1996 suggests a link between adoption and the
incidence of HIV/AIDS, given the wide promotion that
this method was given following the outbreak of this
disease.
It is also worth noting that during the period
under investigation use rates for the injection
maintained a steady growth.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
133
It is possible that women were aware of the risks
associated with condoms, and subsequently chose to
reinforce it with another method if only to bar
pregnancy thus prevent vertical transmission of
HIV/AIDS.
Figure 20 focuses on the proportion of women
currently using the condom by age, comparing 1988 and
1996. Over this 8-year period, the proportion of
women using the condom has increased considerably,
across all ages. For example, while only 1.9% of
women aged 15-19 were currently using the condom in
1988; by 1996 this proportion had jumped to 12.4%.
The proportion currently using the condom peaks at 20-
24 before dropping steeply as age increases.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
134
Figure 20: Women Aged 15-49 Currently Using the Condom
by Age, 1988 versus 1996
■ 1988
■ 1996
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
135
CHAPTER 5
Determinants of HIV/AIDS Preventive
Health Behavior
This chapter looks at perceptions as determinants
of condom use, HIV testing, and having one sexual
partner as medically prescribed preventive measures
against HIV/AIDS infection. The logistic regression
technique is employed to assess the odds of adopting
these measures in the framework of the Health Belief
Model. It takes the form:
Ln [p/ (1-p) } = b0 + biXi+ b2X2+...+bkX]c (Pampel,
2000)
where p = the probability that the dependent variable,
in this case condom use, takes the value 1 as opposed
to 0
1-p = the probability that the dependent variable
takes the value 0
b0 is the intercept
i], b2 ,-. bfr are the regression coefficients of the k
independent variables Xi, X2-X]i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
137
The result, ln[p/(l-p)] represents the log odds
of using a condom, getting an HIV test, or having one
sexual partner for an individual possessing
characteristic Xj_, net of other characteristics in the
model. The exponent of the log odds yields the odds
of adopting the preventive measure of reference.
Consideration was made of the fact that
perceptions regarding preventive behaviors of interest
to this study may be inter-related, more so because
the sample used was drawn at the household level, with
all individuals aged 10 to 64 qualifying for an
interview. This posed the possibility that individual
perceptions may be a product of social interaction,
especially within households.
Consequently, a test for clustering was carried
out to see if this was the case. Specifically,
hierarchical cluster analysis was carried out, and a
cluster membership variable introduced into the full
model to see if there would be any changes with
respect to the significance pattern of the variables,
especially perception ones.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
138
There was no major difference between results of
the regression with the cluster membership variable
and those without it, leading to a conclusion that
clustering is not a problem. For confirmation,
results of the full models with and without clustering
are presented here. In light of this finding,
analysis is based on simple logistic regression
without clustering.
Table 2 below presents a summary of descriptive
statistics for the variables used in this analysis.
It will be observed that for some variables very few
cases were obtained, a situation that could
potentially bias findings. To avoid this problem,
data was weighted to mirror the relative
characteristics of the population by taking the
average of the weighting factor, and then multiplying
the product with individual weights. The result is a
weighted sample reflecting the population on the
distribution of characteristics of interest.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
139
Table 2: Summary of Descriptive Statistics for Variables Used in
the Regression Analyses8
Variable Question Asked Distribution of Cases (or
Mean, where Applicable)
Condom Use Did you use a condom
the first time you had
sex with this (most
recent) partner?
Yes = 1296
No(Reference) = 787
Ever Tested
for HIV
I don't want to know
the results but have
you ever tested to see
if you have HIV?
Yes = 543
No (Reference) = 1866
Number of
Sexual
partners over
last 12months
In the last 12 months
with how many people
overall have you had
sex?
Mean = 0.98
Minimum = 0
Maximum = 21
Susceptibility
(applies
across all
Models)
Do you think your
partner has other
partners?
Yes = 1152
No(Reference) = 933
Severity
(applies
across all
Models)
Do you personally know
anyone who has or who
has died from
HIV/AIDS?
Yes = 846
No(Reference) = 1529
Cost of Condom
Use
Do you think women
should always be
allowed to buy
condoms?
Yes = 2058
No(Reference) = 391
Cost HIV Test If you chose to be
tested for HIV and
were told that you had
HIV, would you tell
anyone the results?
Yes = 1818
No (Reference) = 593
8 All cell entries should add up to 2466; any discrepancy is due
to missing values
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
140
Table 2: Summary of Descriptive Statistics for Variables Used in
the Regression Analyses (Continued)
Variable Question Asked Distribution of Cases (or
Mean, where Applicable)
Cost of
Reducing
Number of
Sexual
partners
In the last 12 months
have you exchanged or
received money for sex?
Yes = 28
No(Reference) = 2089
Benefits of
Condom Use
Can people reduce their
chances of HIV
infection by using
condoms?
Yes = 1897
No (Reference) = 526
Benefits of
Getting HIV
Test
Can a person who looks
healthy be infected
with HIV/AIDS?
Yes = 2006
No(Reference) = 230
Benefits of
Having One
Sexual Partner
Can people reduce their
chances of getting
HIV/AIDS by having only
one partner?
Yes = 2106
No (Reference) = 304
Self-Efficacy Is there anything a
person can do to reduce
their chances of
becoming infected with
HIV?
Yes = 2223
No(Reference) = 188
Age Mean = 30.6 years
(Analysis Restricted to 15
to 4 9 year-olds)
Sex Male(Reference) = 1054
Female = 1412
Marital Status Married (Reference) = 447
Living Together = 549
Divorced/Separated/Widowed
= 81
Never Married = 1388
Residence Urban (Reference) = 748
Rural = 1748
Socioeconomic
Status (SES)
How many years of
education in total did
you complete?
Low SES (0 to 7 = 745
Mid SES (8 to 12) = 1176
High SES(Reference) = 215
Mean = 9.1 years
Minimum = 0
Maximum = 25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
141
Table 2: Summary of Descriptive Statistics for Variables Used in
the Regression Analyses (Continued)
Variable Question Asked Distribution of Cases (or
Mean, where Applicable)
Age Difference
Between
Partners (in
years)
Derived by subtracting
respondent's age from
partner's age
0 to 5 (Reference = 1291)
6 to 10 years = 523
More than 10 years = 271
Mean = 1.4
Minimum = -4 8
Maximum = 67
Availability
of Information
In the past 4 weeks
have you heard or seen
any information about
HIV/AIDS?
Yes (Reference) = 1812
No = 600
Discussion of
HIV
During the past 4 weeks
have you discussed
HIV/AIDS with anyone?
Yes (Reference) = 1041
No = 1370
Partner's
Place of
Residence
Where does your partner
live?
With Respondent (Ref=1283)
Elsewhere = 803
Knows Where to
Obtain HIV
Test
Do you know a place
where you can go to get
an HIV test?
Yes (Reference) = 1773
No = 634
5.1 Condom Use
The HBM informs us that health behavior is an
outcome influenced by perceptions of the individual
towards prescribed preventive measures. In the
context of HIV/AIDS, adoption of recommended
preventive measures such as the condom will therefore
be influenced by people's perceptions of the disease
as well as the preventive methods.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
142
The objective of this section is to establish
perceptions about HIV/AIDS and condom use and how they
predict actual usage, in light of the critical role
that condoms play in the prevention of HIV/AIDS
infection. A number of factors are controlled for,
namely age of the individual, sex, age difference
between partners, marital status, socioeconomic
status, place of residence, and access to information
about HIV/AIDS.
Marital status should however be approached with
caution. Ideally, the analysis had wished to use
marital status at the time of the particular sexual
encounter as referenced by the dependent variable, ie
the first encounter. However, BAIS simply collected
data about current marital status at the time of
interview and did not relate it to the first sexual
encounter between partners. It is nevertheless
assumed that for most individuals this took place
during their current marital status.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
143
One interaction term is also included in this
analysis, sex versus age difference and place of
residence versus access to information. The other
two, sex versus marital status and place of residence
versus access to information, did not pass the set
2
test, which involved observing R with and without the
interaction term, and then subtracting the value for
the log likelihood of the former from the latter.
This is a chi square test with the degrees of freedom
equal to the number of variables corresponding to that
interaction term.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
144
Table 3: Logistic Regression Model for the Odds of Using a
Condom: Different Scenarios under the Health Belief Model
Variable Model 1 Model 2
beta Odds beta Odds
Intercept
***
-0.781
(0.246)
0.465
(0 .301)
Susceptibility
(Suspects partner
has others = Ref)
Doesn't Suspect
Partner
* *
0.316
(0.107)
1.372
0.275*
(0.123)
1.316
Severity
(Knows Someone
with/died from
HIV/AIDS = Ref)
Doesn't Know
Anyone
0. 059
(0.111)
1.061 0.141
(0.127)
1.152
Condom Use Cost
(Thinks Women
should always be
allowed to buy
condoms = Ref)
Women should not
always be allowed
to buy condoms
* * *
0. 603
(0.150)
1.828
* * *
0.753
(0.176)
2.124
Benefits of Condom
Use
(Using Condom
Reduces Chances
of HIV infection
= Ref)
Condom does not
reduce chances of
HIV infection
★ * *
0. 624
(0.139)
1. 867
0.353*
(0.162)
1.424
Self-Efficacy
(People can do
something to
reduce chances of
HIV infection=Ref)
People cannot do
anything
0.262
(0.208)
1.299 0.361
(0.241)
1.434
*p < 0.05 **p < o.oi ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3: Logistic Regression Model for the Odds of Using a
Condom: Different Scenarios under the Health Belief Model
(Continued)
Variable Model 1 Model 2
beta Odds beta Odds
Age
(20-30=Ref)
15-19
31-39
40-49
***
1.375
(0.352)
***
-1.450
(0.139)
***
-2.373
(0.178)
3. 954
0.235
0.093
Sex
(Male = Ref)
Female
***
-0.551
(0.128)
0.576
Partners' Age
Difference
(0-5 = Ref)
6-10 years
> 10 years
■ k *
-0.393
(0.142)
-0.427*
(0.185)
0.675
0.652
2
R
-2 LOG Likelihood
0.034
2044.039
0.024
1649.483
*p < 0.05 **p < 0.01 ***p < 0.001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
146
Table 3: Logistic Regression Model for the Odds of Using a
Condom: Different Scenarios under the Health Belief Model
(Continued)
Variable Model 3 Model 4
beta Odds
Ratio
beta Odds
Ratio
Intercept -0.337
(0.307)
-0.261
(0.390)
Susceptibility
(Suspects partner
has others = Ref)
Doesn't Suspect
Partner
0.233
(0 .125)
1.263 0.029
(0.135)
1.029
Severity
(Knows Someone
with/died from
HIV/AIDS = Ref)
Doesn't Know
Anyone
0.151
(0.128)
1.163 0.165
(0.137)
1.179
Condom Use Cost
(Thinks Women
should always be
allowed to buy
condoms = Ref)
Women should not
always be allowed
to buy condoms
***
0.776
(0.176)
2.173
***
0. 643
(0 .188)
1. 902
Benefits of Condom
Use
(Using Condom
Reduces Chances
of HIV infection
= Ref)
Condom does not
reduce chances of
HIV infection
0.352*
(0.163)
1.422 0.222
(0.174)
1.249
Self-Efficacy
(People can do
something to
reduce chances of
HIV infection=Ref)
People cannot do
anything
0.368
(0.241)
1.445 0.395
(0.256)
1.484
*p < 0.05 **p < 0.01 ***p < 0.001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
147
Table 3: Logistic Regression Model for the Odds of Using a
Condom: Different Scenarios under the Health Belief Model
(Continued)
Variable Model 3 Model 4
beta Odds beta Odds
Age(20-30=Ref)
15-19
***
1.365
(0.352)
3. 915
* *
1.127
(0.356)
3.087
31-39
★ * ★
-1.492
(0.140)
0.225
***
-0.920
(0.159)
0.398
40-49 ★ ★ ★
-2.495
(0.140)
0.083 ***
-1.179
(0.211)
0.169
Sex(Male = Ref)
Female -0.297
(0.163)
0.846 -0.304
(0.171)
0.738
Partners' Age Difference
(0-5 = Ref)
6-10 years -0.168
(0.235)
1.484 -0.075
(0.250)
0.927
> 10 years 0.395
(0.317)
0.703 0.123
(0.344)
1.130
Female*6-10 yrs
Female* > 10 yrs
-0.352
(0.296)
***
-1.236
(0.389)
0.703
0.291
-0.301
(0.314)
-0.651
(0.414)
0.740
0.522
Marital Status
(Married=Ref)
Living Together
★ ★ ★
1.226
(0.189)
* ★ ★
2.396
(0.461)
* * *
1.807
(0.188)
3.408
Divorced/Sep/Widowed
Never Married
10.978
6. 094
SES (Years of Education)
(>13 years= Ref)
0-7 years
★ ★ ★
-1. 080
(0.234)
0.340
8-12 years -0.472
(0.231)
0. 624
2
R
0.245 0.301
-2 LOG Likelihood
1639.087 1513.023
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
148
*p < 0.05 **p < 0.01 ***p < 0.001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3: Logistic Regression Model for the Odds of Using a
Condom: Different Scenarios under the Health Belief Model
(Continued)
Variable Model 5
beta Odds
Intercept -0.100
(0.399)
Susceptibility
(Suspects partner
has others = Ref)
Doesn't Suspect
Partner
0.038
(0.135)
1.039
Severity
(Knows Someone
with/died from
HIV/AIDS = Ref)
Doesn't Know
Anyone
0.133
(0.138)
1.142
Condom Use Cost
(Thinks Women
should always be
allowed to buy
condoms = Ref)
Women should not
always be allowed
to buy condoms
* ★ ★
0. 668
(0.189)
1.950
Benefits of Condom Use
(Using Condom
Reduces Chances
of HIV infection
= Ref)
Condom does not
reduce chances of
HIV infection
0.208
(0.175)
1.231
Self-Efficacy
(People can do
something to
reduce chances of
HIV infection=Ref)
People cannot do
anything
0.368
(0.256)
1.445
*p < 0.05 **p < 0.01 ***p < 0.001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3: Logistic Regression Model for the Odds of Using a
Condom: Different Scenarios under the Health Belief Model
(Continued)
Variable Model 5
beta Odds Ratio
Age(20-30=Ref)
15-19
***
1.164
(0.357)
3.203
31-39
***
-0.935
(0.159)
0.392
40-49 ***
-1.776
(0.212)
0.169
Sex(Male = Ref)
Female -0.276
(0.171)
0.759
Partners' Age Difference
(0-5 = Ref)
6-10 years -0.082
(0.251)
0. 922
> 10 years 0.142
(0.346)
1.153
Female*6-10 yrs -0.301
(0.316)
0.740
Female* > 10 yrs -0.648
(0.415)
0.523
Marital Status
(Married=Ref)
Living Together
Divorced/Sep/Widowed
* * *
1.270
(0.191)
***
2.486
3.562
12.013
Never Married
(0.466)
1.876
(0.191)
6.525
SES (Years of Education)
(>13 years= Ref)
0-7 years
***
-1.017
(0.236)
0.362
8-12 years -0.462
(0.232)
0.630
*p < 0.05 **p < o.oi ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3: Logistic Regression Model for the Odds of Using a
Condom: Different Scenarios under the Health Belief Model
(Continued)
Variable Model 5
beta Odds
Place of Residence
(Urban= Ref)
Rural
-0.246
(0.142)
0.782
Information Access
(Seen/Heard Info
on HIV last 4
weeks = Ref)
Didn't See/Hear 0.258
(0.152)
0.773
2
R
-2 LOG Likelihood
0.303
1506.688
*p < 0.05 **p < o.oi ***p < 0.001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
152
Table 4: Logistic Regression Model for the Odds of Using a Condom
Accounting for Clustering
Variable beta Odds
Intercept -0.256 (0.506)
Susceptibility 0.017 (0.136) 1.017
Severity 0.128 (0.138) 1.136
Cost of Using a Condom
0.715 (0.198)
2.043
Benefits of Using a Condom 0.222 (0.176) 1.248
Self-Efficacy 0.478 (0.296) 0.296
Age(20-30 = Reference)
15-19
31-40
41-49
★ * *
1.156 (0. 357)
* * *
-0.928 (0. 160)
•k ★ ★
-1.781 (0.212)
3.177
0.395
0.169
Sex(Male = Reference)
Female -0.292 (0.171) 0.747
Partners' Age Difference (0-5 = Ref)
6-10 years
More than 10 years
-0.091 (0.252)
0.141 (0.346)
0. 913
1.152
Female*6-10 yrs
Female* > 10 yrs
-0.292 (0.316)
-0.659 (0.416)
0.747
0.517
Marital Status (Married = Reference)
Living Together
Divorced/Separated/Widowed
Never Married
* * *
1.255 (0.192)
★ ic ★
2.456 (0.465)
★ ★ ★
1.866 (0.192)
3.507
11.663
6. 461
Socioeconomic Status(High SES = Ref)
Low SES
Mid SES
•k k •k
-1.028 (0.237)
*
-.460 (0.232)
0.358
0. 631
Place of Residence
(Urban = Reference)
Rural -0.275 (0.193) 0.759
Information Access
(Seen/Heard HIV Info last 4 wks=Ref)
Didn't See/Hear
-0.661* (0.293)
0 . 516
Rural*Seen/Heard Info on HIV/AIDS 0.558 (0.339) 1.747
Cluster Membership 0.001 (0.001) 1.001
2
R
-2 LOG Likelihood
0.305
1502.805
*p < 0.05 **p < 0.01 ***p < 0.001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
153
Table 3 presents results of the logistic
regression for condom use. The first model includes
only perception variables, ie, susceptibility,
severity, costs, benefits, and self-efficacy.
Subsequent models introduce related intervening
variables step by step, but not stepwise as the latter
leaves the order of variable introduction completely
to the computer.
Out of the five perception variables in Model 1,
costs and benefits of use are the most significant
perceptions associated with actual condom use,
followed by susceptibility. The odds of condom use
are nearly twice among people who perceive no costs of
use relative to those who do, and 1.867 times among
people who perceive benefits of use relative to those
who do not. Both these results are significant at
p=0.001. Similarly, people who perceive themselves to
be at some threat of HIV infection have significantly
higher odds of using the condom than those who do not,
and this is significant at p=0.01.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
154
Interestingly, and contrary to hypotheses postulated
earlier, severity and self-efficacy perceptions do not
translate into any significant action in terms of
condom use. However, hypotheses with respect to
perceptions of susceptibility to HIV, costs and
benefits of condom use are confirmed.
Model 2 introduces the first set of intervening
variables, age, sex and age difference between
partners. Our hypothesis expected that younger people
would be more likely to use condoms than relatively
elderly ones. This hypothesis is confirmed. Taking
20 to 30 year olds as the reference, it can be seen
that condom use is progressively inversely related to
age. For example, the odds of condom use among 15 to
19 year olds are more than 4 times those among 20 to
30 year olds, and this result is significant at
p=0.001. On the contrary, 31 to 39 year olds only
have less than one quarter the odds of using a condom
relative to the 20-30 age group. For people aged 41
to 49, the odds are even lower (under 10%).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
155
The odds of condom use are significantly lower
among females relative to males, and this result is
significant at p=0.001. Women have only slightly over
57% odds of using a condom compared to men.
Age difference between partners also has an
influence on condom use - the higher the difference
between partners, the less likely that a condom will
be used during the first sexual encounter. For
example, for people whose age difference ranges
between 6 and 10 years, the odds of using a condom are
only 68%, while for those whose age difference is more
than years the odds are even lower (62%), relative to
couples whose age difference is 5 years or less.
It is important also to note that while variables
introduced in Model 2 are all highly significant,
perception variables introduced in Model 1 maintain
their significance, though with a slight reduction in
significance for perceptions of susceptibility and
benefits. Perceptions of costs associated with condom
use are consistently significant at p=0.001.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
156
This means that benefits, susceptibility and cost
perceptions have a significant influence on condom use
decisions even when age, sex and age difference
between partners are held constant.
In Model 3 an interaction effect between sex and
age difference is introduced into the regression.
This results in a dramatic, though expected, turn of
events - both the sex and age difference variables
lose significance. Even the interaction between sex
and age difference of 6 to 10 years is not
significant. However, interaction between sex and an
age difference of more than 10 years is significant at
p=0.001, with the odds of condom use being only 52%
relative to those for males involved in relationships
of the same age difference.
Model 3 confirms the gender-based dimension of
condom use - that it is not sex or the age difference
between partners that determines condom use; rather it
is the interplay between these two variables that
matters.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
157
Specifically, being female and having a partner who is
10 or more years older significantly reduces the odds
that a condom will be used during the first sexual
encounter.
It should be noted also that the perceptions of
susceptibility to HIV and benefits of condom use have
lost significance in Model 3. This suggests that once
the sex/age difference dynamic comes into play, condom
use is no longer determined by whether or not one
feels susceptible to HIV/infection; neither is it
influenced by perceived benefits associated with
usage.
Model 4 introduces marital status and
socioeconomic status of the respondent into the
equation. As can be seen, the odds of using a condom
are significantly higher for all marital statuses
(p=0.001) relative to the married category.
Divorced/separated/widowed people have the highest
odds of using a condom relative to married people,
almost 11 times, followed by never married individuals
with more than six times the odds.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
158
Living together couples, though with more than three
times the odds of condom use relative to married ones,
are the closest to them.
At this stage perceptions of costs associated
with condom use are still significant determinants of
condom use. Sex and age difference has been
completely eliminated out from the picture, together
with their interactions, which suggests that marital
relations tend to supercede these two factors with
respect to decisions about condom use.
As hypothesized, low SES is associated with lower
odds of condom use, only 34% relative to the high
status group (p=0.001). The odds of condom use among
medium SES respondents are also significantly lower
(62%) relative to high SES (p=0.05).
In Model 5, place of residence and access to
information are introduced into the regression. Both
factors have no bearing on condom use decisions on the
first sexual encounter.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
159
Of all perception variables, only the perception of
cost of condom use is still significant in Model 5,
and that it remains so at p=0.001.
Having assessed various perceptions and how they
relate to condom use controlling for the specified
factors, interest is shifted to probing interactions
between perception and some of the control variables.
Specifically, age and socioeconomic status are
assessed for possible interaction with the various
perception variables.
5.1.1 Perceptions, Age and Condom Use
Table 5 seeks to check for interaction between
age and perception variables. The sample is divided
into four age groups, 15-19, 20-30, 31-39, and 40-49.
Logistic regression is then carried out for each of
these groups, predicting the odds of condom use within
each age group using only measures of perception as
independent variables.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
160
The slopes obtained from this procedure are then
compared between the youngest and oldest age groups to
see if they differ significantly, and if they do the
null hypothesis is rejected in favor of the
alternative. A t-test using following formula
(Mcnemar, 1962) is used to test hypotheses of no
difference between the slopes:
b (40-49) _ b (1 5 -1 9 )
"\/(Seb(15-49) +Seb(40-49) )
where b(40_49) is the slope of the perception variable
for age group 40 to 4 9
b(i5_i9) is the slope of the perception variable for the
age group 15 to 19
seb ( 40-49) the standard error of the slope for
perception variable for the age group 40 to 4 9
seb (14-19) the standard error of the slope of
perception variable for the age group 15 to 19.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
161
The test statistic t has n-2 degrees of freedom, and
because the sample size is very large (n= 2466), the
critical value for t is obtained from df = 00.
Table 5: Age-Specific Logistic Regression Model for the Odds of
Using a Condom: Different Scenarios under the Health Belief Model
Age Perception b
seb
Odds Ratio
15 to 19
Intercept
Susceptibility
Severity
Costs
Benefits
Self-Efficacy
0.791
1.238
0.466
-0.231
0. 942
0.512
0.107
0.654
0.679
0.800
0.891
1.044
3.448
1.594
0.793
2.564
1.669
20 to 30
Intercept
Susceptibility
Severity
Costs
-0.123
0.168
-0.100
**
0. 688
0.375
0.178
0.184
0.236
1.182
0. 905
1.990
Benefits
**
0.564
0.219 1.759
Self-Efficacy 0. 480 0.298 1.617
Intercept
***
-1.542
0.426
31 to 39
Susceptibility
Severity
Costs
0.040
0.251
* *
0.773
0.184
0.192
0.280
1.040
1.285
2.167
Benefits
Self-Efficacy
0.266
0.450
0.244
0.378
1.305
1.569
Intercept
***
-4.236
0.808
Susceptibility
Severity
0.435
0.532*
0.266
0.271
1.546
1.703
40 to 49 Costs
★ *
1.581
0.527 4.858
Benefits
Self-Efficacy
0.541
0. 829
0.341
0.639
1.718
2.290
*p < 0.05 **p < 0.01 ***p < 0.001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
162
5.1.1.1 Age and Perceived Susceptibility to HIV/AIDS
Using results from Table 5, the following
hypotheses are tested.
H0: The perception of susceptibility to HIV/AIDS is the
same for young and old people
Hx: Young people perceive themselves to be more
susceptible to HIV/AIDS than older people.
1.238-0.435
V(0.654)2 +0.266)2
t = 1.137
The value of t is less than the critical value
for the one-tail t( ( = 0.05) (1.645), leading us to
retain the null hypothesis and conclude that the
perception of susceptibility to HIV/AIDS does not
differ significantly between old and young people.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
163
5.1.1.2 Age and Perceived Severity of HIV/AIDS
With respect to perceived severity of HIV/AIDS,
the following hypotheses are tested.
H0: Older people's perception of the severity of
HIV/AIDS is the same as that for young people
Hx: Older people's perception of the severity of
HIV/AIDS is greater than that for young people.
0.466-0.532
V(0.679)2 +(0.271)2
= -0.090
and from the tables, the associated critical value is
associated with a one-tailed t-test is 1.645. We
therefore retain the null hypothesis and conclude that
there is no difference in the perception of HIV/AIDS
severity between older and younger people.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
164
5.1.1.3 Age and Perceived Costs of Condom Use
The same test is applied with regard to perceived
costs associated with condom use between young and old
people. In this case, the hypotheses are as follows:
H0: Old people's perception of the costs of condom use
is the same as those for young people
H-l : Old people perceive greater costs of condom use
relative to young people.
Using results from Table 5 above,
1.581 + 0.231
V(0.527)2 +(0.800)2
t = 1.891
Given that the critical value of the t is 1.645,
we reject the null hypothesis and conclude that
perception of costs associated with condom differ
significantly between old and young people, with the
former attaching greater costs to condom use.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
165
5.1.1.4 Age and Perceived Benefits of Condom Use
The t test is repeated for benefits associated
with condom use, again using results presented in
Table 5.
H0: The perception of benefits of condom use is the
same between old and young people
Hx: Young people perceive greater benefits of condom
use than older people.
0.942-0.541
V(0.891)2 +(0.341)2
t = 0.420
The value of this test statistic is less than the
critical value of the test; therefore we conclude that
perceptions of susceptibility do not differ
significantly between young and old people.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
166
5.1.1.5 Age and Perceived Self-Efficacy towards Condom
Use
The following hypotheses are tested:
H0: The perception of self-efficacy in condom use does
not differ between old and young people
H-^ Younger people perceive greater self-efficacy in
condom use than older people. Results from Table 5
yield
0.512-0.829
V(0.639)2 +(1.044)2
t = -0.259
Again we retain the null hypothesis and conclude
that there is no significant difference in perceptions
of self-efficacy between the young and old.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
167
5.1.1.6 Summary
The afore-going analysis sought to check if there
is any interaction between perception variables and
age in the prediction of condom use. Specifically,
the intention was to assess if the observed
diminishing of significance in perception variables
accompanying the addition of control variables (Model
6, Table 3) could be partially explained by
interaction between age and perceptions of HIV/AIDS.
Results of the above analysis do not support the
hypothesis of interaction between age and all
perception variables except costs of condom use.
Perceived costs associated with condom use
increasingly become an important predictor of actual
condom use as the age of an individual increases, with
slopes of -0.231 for 15-19 year-olds (not
significant), 0.688 for 20-30 year-olds,
(significant), 0.773 for 31-39 year-olds (significant)
and 1.581 for 40-49 year-olds (significant).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
168
Also, the interaction between age and cost of condom
use is significant as shown by the test involving the
difference in the slopes for age groups 15-19 and 40-
49.
Overall, it appears costs associated with condom
use are a critical predictor of actual usage,
depending on the age of the individual. While young
people do not seem to perceive much costs related to
using condoms, older people perceive a lot more costs
to using condoms.
5.1.2 Perceptions, Socioeconomic Status and Condom Use
Table 6 seeks to check for interaction between
SES and perception variables. The sample is divided
into three SES groups, low, mid, and high. Logistic
regression is then carried out for each of these
groups, predicting the odds of condom use within each
SES group using only measures of perception as
independent variables. Like was done with age and
condom use, Mcnemaar's t-test is used.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
169
Table 6: SES-specific Logistic Regression Model for the Odds of
Using a Condom: Different Scenarios under the Health Belief Model
Socioeconomic
Status
Perception b
seb
Odds
Ratio
Intercept
Susceptibility
1.480
* *
0.466
0.373
0.174 1.594
Low SES
Severity 0.017 0.186 1.017
Costs
**
0. 689
0.220 1. 991
Benefits 0.297 0.214 1.345
Self-Efficacy 0.358 0.329 1.431
Intercept
Susceptibility
0.333
0.195
0.406
0.166 1.215
Mid-SES
Severity -0.207 0.169 0.813
Costs 0.054 0.272 1.056
Benefits
0.531*
0.232 1.701
Self-Efficacy 0.443 0.310 1.557
Intercept
Susceptibility
0.158
0.575
1.258
0.361 1.776
High SES
Severity
0.762*
0.347 2.143
Costs 0. 698 0.640 2.009
Benefits 0. 647 0.445 1.909
Self-Efficacy -1.208 1.126 0.299
*p < 0.05 **p < o.oi ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
170
5.1.2.1 SES, Perceived Susceptibility to HIV/AIDS and
Condom Use
Testing for difference in susceptibility between
high and low SES, the hypotheses are stated as
follows:
H0: There is no difference in perceptions of
susceptibility to HIV/AIDS between High and low SES
groups
Hi: High SES people perceive themselves to be more
susceptible to HIV/AIDS than low SES people.
0.575-0.466
V(0.361)2 +(0.174)2
= 0.273
The value of the test statistic t is less than
the critical value 1.645, which leads to the retention
of the null hypothesis. The conclusion therefore is
that the observed difference in the betas between
susceptibility among high and low SES groups is due to
chance.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
171
5.1.2.2 SES, Perceived Severity of HIV/AIDS and Condom
Use
With respect to perceived severity of HIV/AIDS,
the following hypotheses are stated:
H0: There is no difference in perceptions of severity
between high and low SES people
Hi: High SES people perceive HIV/AIDS to be more severe
than low SES people.
0.762-0.017
V(0.347)2 +(0.186)2
= 1.891
The test statistic t is greater than the critical
value of 1.645, leading us to reject the null
hypothesis in favor of the alternative. We therefore
conclude that perceptions of HIV/AIDS severity are
stronger among high than low SES people.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
172
5 .1.2.3 SES, Perceived Costs of Condom Use and Condom
Use
With respect to this perception, the hypotheses
are stated as follows:
H0: There is no difference in perceived costs of condom
use between high and low SES people
Hi: Low SES people perceive greater costs to condom use
than high SES people.
0.698-0.689
^(0.640)2 +(0.220)2
t = 0.013
Given that the critical value of t with 00 degrees
of freedom is 1.645, we fail to reject the null
hypothesis and conclude that the observed difference
in the slopes of condom use between low and high SES
groups in Table 6 is due to chance.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
173
5.1.2.4 SES, Perceived Benefits of Condom Use and
Condom Use
Checking for interaction between perceptions of the
benefits of condom use and SES, we state the following
hypotheses:
H0: There is no difference in perceptions of benefits
associated with condom use between high and low SES
people
Hi: High SES people perceive greater benefits of condom
use than low SES people.
0.647-0.297
■y/(0.445)2 +(0.214)2
= 0.709
The test results suggest that perceptions
associated with condom use benefits do not vary
significantly by socioeconomic status; we therefore
retain the null hypothesis.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
174
While this thesis had wished to test for
differences in perceptions of self-efficacy between
high and low SES groups this could not be done because
the slope for high SES was negative, suggesting lower
odds of condom use among those who perceived greater
self-efficacy relative to those who did not. This is
probably due to the small number of people in the high
SES group, as evidenced by the relatively high
standard error (1.126) associated with this result.
5.1.2.1 Summary
The most significant result observed here is the
difference in perceptions of severity between high and
low SES groups. It appears the deadliness of HIV/AIDS
is better understood and appreciated more among high
SES than low SES people. All other perception
variables have no significant interaction with
socioeconomic status.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
175
5.1.3 Perceptions, Gender and Condom Use
Table 7: Sex-specific Logistic Regression Model for the Odds of
Using a Condom: Different Scenarios under the Health Belief Model
Sex Perception b seb Odds
Ratio
Intercept
★ ★ ★
-1.341
0.351
Susceptibility
•k Jc ★
0.669
0.155 1. 953
Males
Severity 0.197 0.163 1.218
Costs
* * *
0.751
0.203 2.119
Benefits
0.637
0.191 1.891
Self-Efficacy 0. 494 0.299 1. 638
Intercept
★ ★ *
-1.176
0.276
Females
Susceptibility
Severity
0.123
0.091
0.129
0.134
1.131
1.096
Costs
* * *
0. 656
0.183 1.927
Benefits
0.851
0.167 2.341
Self-Efficacy 0.209 0.235 1.232
*p < 0.05 **p < 0.01 ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
176
Table 7 investigates interaction in condom use
between sex and perception variables. Perceptions of
condom use cost and benefits seem to be significantly
associated with condom use for both sexes. However,
the odds of condom use appear to be significantly
higher among males who perceived themselves
susceptible to HIV/AIDS relative to those who do not,
than is the case with females.
The following five subsections carry out t tests
to assess the difference between slopes for each
perception, comparing the two sexes.
5.1.3.1 Gender, Perceived Susceptibility to HIV/AIDS
and Condom Use
This subsection tests the hypothesis of no
difference in condom use between males and females
based on the perception of susceptibility to HIV/AIDS.
The slopes used are obtained from Table 6.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
177
H0: There is no difference in condom use between males
and females who perceive themselves to be susceptible
to HIV/AIDS
H-l : Males who perceive themselves to be susceptible to
HIV/AIDS are more likely to use condoms than females
who perceive the same
0.669-0.123
V(0.155)2 +(0.129)2
2.703
The critical value of this test is 1.645. We
therefore reject the null hypothesis in favor of the
alternative and conclude that perceived susceptibility
is significantly with the odds of condom among males
than among females. Put differently, males who
perceive themselves vulnerable to HIV/AIDS have
significantly greater odds of using condoms relative
to males who do not; on the other hand, females who
perceive themselves vulnerable to HIV/AIDS do not have
significantly higher odds of condom use relative to
females who do not.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
178
5.1.3.2 Gender, Perceived Severity to HIV/AIDS and
Condom Use
Here the test seeks to establish whether or not
there is a difference in condom use between males and
females based on perceived severity of HIV/AIDS.
Again the slopes used are obtained from Table 7.
H0: There is no difference in condom use between males
and females who perceive HIV/AIDS as a severe disease
Hx: Males who perceive HIV/AIDS as a severe disease
are more likely to use condoms than females who
perceive the same
0.197-0.091
V(0.163)2 +(0.134)2
0.500
The test statistic is less than the critical value, so
we retain the null hypothesis and conclude that the
perception of severity equally leads to adoption of
condom use among both males and females.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
179
5.1.3.3 Gender, Perceived Cost of Condom Use and
Condom Use
This subsection looks at the difference in condom
use between males and females based on their perceived
costs associated with condom use.
H0: There is no difference in condom use between males
and females who perceive no cost to condom use
H-l : Males who perceive no cost to condom use are more
likely to use condoms than females who perceive the
same
0.751-0.656
V(0.203)2 + (0.183)2
0.349
Since t is less than the critical value we retain the
null hypothesis and conclude that the odds of condom
use are equal among both and females who perceive no
costs to condom use relative to males and females who
do.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
180
5.1.3.4 Gender, Perceived Benefits of Condom Use and
Condom Use
This subsection looks at the difference in condom
use between males and females based on their perceived
benefits associated with condom use.
Hc: There is no difference in condom use between males
and females who perceive benefits to condom use
H-l : Females who perceive benefits to condom use are
more likely to use condoms than males who perceive the
same
0.851-0.637
V(0.167)2 +(0.191)2
0.846
The test statistic t is less than 1.645, leading us to
retain the null hypothesis and conclude that females
and males who perceive some benefits to condom use
have equal odds of using condoms, relative to females
and males who do not have the same perception.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
181
5.1.3.5 Gender, Perceived Self-Efficacy towards Condom
Use and Condom Use
This subsection looks at the difference in condom
use between males and females based on their perceived
self-efficacy associated with condom use.
H0: There is no difference in condom use between males
and females who perceive some self-efficacy towards
condom use
H-l : Males who perceive some self-efficacy towards
condom use are more likely to use condoms than females
who perceive the same
Q.494-0.209
V(0.299)2 +(0.235)2
0.752
We retain the null hypothesis since t is less than
1.645 and conclude that there is no difference in
condom use between males and females who share the
same perception towards self-efficacy towards condom
use.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
182
5.1.3.5 Summary
This section has brought to light one more
differential with regard to perception variables and
how they influence condom use - the perception of
susceptibility to HIV/AIDS leads to condom use only if
it is the male who feels vulnerable to the disease.
For women, even if they feel susceptible to HIV/AIDS,
this does not translate into adoption of preventive
behavior, specifically condom use.
The above finding seems to confirm the gender
disparities in sexual decision making; the perceived
risk of infection for women is not an important factor
in the prevention process, suggesting that women are
not able to influence the adoption of safety measures
even though they feel at risk. It is men who need to
feel the risk for this to happen.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
183
5.2 HIV Testing
Logistic regression was used to assess dynamics
behind this phenomenon, and again cluster analysis was
found not to be necessary for this analysis. Table 9
may be compared to Model 4 in Table 8 for confirmation
of this fact.
Again the first model in Table 8 presents the
results of the regression of HIV testing only on
perception variables as specified under the Health
Belief Model.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
184
Table 8: Logistic Regression Model for the Odds of Getting an
HIV/AIDS Test: Different Scenarios under the Health Belief Model
Variable Model 1 Model 2
beta Odds
Ratio
beta Odds
Ratio
Intercept
* * *
-2.465
(0.393)
***
-2.352
(0.408)
0.095
Susceptibility
(Suspects partner
has others = Ref)
Doesn't Suspect
Partner
-0.071
(0.123)
0.932 0.034
(0.124)
1.030
Severity
(Knows Someone
with/died from
HIV/AIDS = Ref)
Doesn't Know
Anyone)
***
0.681
(0.123)
1. 977
★ ★ ★
0. 687
(0.124)
1.989
Cost of Testing
for HIV
(Would Disclose
Test Result = Ref)
Wouldn't Disclose
Test Results
**
0.533
(0.165)
1.704
★ *
0.512
(0.166)
1.668
Benefits of
Testing for HIV
(Someone who looks
healthy can be
HIV+ = Ref)
Someone who looks
healthy cannot be
HIV+
0.089
(0.213)
1.093 0.095
(0.213)
1.100
Self-Efficacy
(People can do
something to
reduce chances of
HIV infection=Ref)
People cannot do
anything
0.546
(0.331)
1.395 0.535
(0.332)
1.708
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
185
Table 8: Logistic Regression Model for the Odds of Getting an
HIV/AIDS Test: Different Scenarios under the Health Belief Model
(Continued)
Variable Model 1 Model 2
beta Odds
Ratio
beta Odds
Ratio
Age
(20-30=Ref)
15-19
31-39
40-49
-0.148
(0.221)
0.100
(0.150)
0.212
(0.177)
0.863
1.105
1.236
Sex
(Male = Ref)
Female
-0.282*
(0.124)
0.754
2
R
0.033 0.038
-2 LOG Likelihood
1622.156 1614.339
*p < 0.05 **p < 0.01 ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
186
Table 8: Logistic Regression Model for the Odds of Getting an
HIV/AIDS Test: Different Scenarios under the Health Belief Model
Variable Model 3 Model 4
beta Odds
Ratio
beta Odds
Ratio
Intercept
* * *
-1.913
(0.464)
***
-1.377
(0.479)
Susceptibility
(Suspects partner
has others = Ref)
Doesn't Suspect
Partner
0.001
(0.129)
1.000 0.006
(0.131)
1.006
Severity
(Knows Someone
with/died from
HIV/AIDS = Ref)
Doesn't Know
Anyone)
***
0. 662
(0.125)
1. 939
***
0.579
(0.128)
1.785
Cost of Testing
for HIV
(Would Disclose
Test Result = Ref)
Wouldn't Disclose
Test Results
* *
0.496
(0.167)
1. 643
* *
0.450
(0.170)
1.568
Benefits of
Testing for HIV
(Someone who looks
healthy can be
HIV+ = Ref)
Someone who looks
healthy cannot be
HIV+
0.061
(0.214)
1.063 -0.026
(0.218)
0. 974
Self-Efficacy
(People can do
something to
reduce chances of
HIV infection=Ref)
People cannot do
anything
0.530
(0.333)
1. 699 0.430
(0.338)
1.538
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
187
Table 8: Logistic Regression Model for the Odds of Getting an
HIV/AIDS Test: Different Scenarios under the Health Belief Model
(Continued)
Variable Model 3 Model 4
beta Odds
Ratio
beta Odds
Ratio
Age(20-30=Ref)
15-19
31-39
40-49
-0.105
(0.225)
0.019
(0.168)
0.048
(0.213)
0. 901
1.019
1.050
-0.037
(0.229)
0.016
(0.171)
0.017
(0.216)
1.016
1.017
Sex(Male = Ref)
Female
-0.280*
(0.126)
0.756 -0.244
(0.128)
0.784
Marital Status
(Married=Ref)
Living Together
Divorced/Sep/Widowed
Never Married
-0.214
(0.194)
0.456
(0.441)
-0.220
(0.188)
0.807
1.577
0.802
-0.160
(0.196)
0.520
(0.445)
-0.140
(0.191)
0.852
1.682
0.870
SES (Years of
Education)
(>13 years= Ref)
0-7 years
8-12 years
-0.240
(0.206)
-0.225
(0.199)
0.787
0.799
0.055
(0.214)
-0.150
(0.210)
1.057
0.861
Place of Residence
(Urban= Ref)
Rural
★ ★ ★
-0.532
(0.129)
0.587
Information Access
(Seen/Heard Info
on HIV last 4
weeks = Ref)
Didn't See/Hear
* * *
-0.871
(0.180)
0.419
2
R
-2 LOG Likelihood
0.041
1608.916
0.069
1564.176
*p < 0.05 **p < 0.01 ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
188
Table 9: Logistic Regression Model for the Odds of Getting an
HIV/AIDS Test Accounting for Clustering
Variable beta Odds Ratio
Intercept
-1.222* (0.532)
Susceptibility 0.009 (0.131) 1.009
Severity
0.577*** (0.128)
1.780
Cost of Testing for HIV
0.436* (0.171)
1.547
Benefits of Testing for HIV -0.053 (0.222) 0. 949
Self-Efficacy 0.369 (0.350) 1.446
Age
(20-30 = Reference)
15-19
31-40
41-49
-0.031 (0.229)
0.018 (0.171)
0.018 (0.216)
0.970
1.018
1.018
Sex(Male = Reference)
Female -0.242 (0.128) 0.785
Marital Status
(Married = Reference)
Living Together
Divorced/Separated/Widowed
Never Married
-0.154 (0.197)
-0.525 (0.446)
-0.132 (0.191)
0. 857
1.690
0.877
Socioeconomic Status
(High SES = Reference)
Low SES
Mid SES
0.066 (0.215)
-0.144 (0.201)
1.068
0. 866
Place of Residence
(Urban = Reference)
Rural
-0.597*** (0.162)
0.551
Knows a Place to Obtain Test
(Yes = Reference)
No
-0.871*** (0.180)
0.419
Cluster Membership 0.001 (0.001) 1.000
2
R
-2 LOG Likelihood
0.069
1563.731
*p < 0.05 **p < 0.01 ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
189
Of the perception variables in Model 1 in Table
8, only severity and cost of HIV testing are
significantly associated with going for a test (both
at p=0.001). Relative to people who did not perceive
HIV/AIDS to be a severe disease, the odds of getting
an HIV test for those who did were almost double.
Similarly, the odds of getting a test among people who
perceived less costs were more than one and half times
higher than those who perceived the costs of testing
to be high.
Perceptions of Susceptibility to HIV/AIDS,
benefits associated with getting a test, and self-
efficacy of testing were not of any significance in
the decisions regarding whether or not to get tested
for HIV.
Introducing age and sex in Model 2 does not
change the picture in any major way. Still, the odds
of getting a test associated with the perception
variables, and their level of significance, are not
affected.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
190
The only thing to note is that women have less odds of
getting a test compared to men (p=0.05), regardless of
perception and age.
In Model 3 marital status and socioeconomic are
introduced into the analysis, with no visible impact
on the results reported in Model 2.
The only model that yields interesting results is
Model 4, which introduces place of residence as well
as whether or not one knows a facility where a test
could be obtained. The odds for getting a test are
significantly lower (59%) for rural residents
(p=0.001) than they are for urban dwellers. Further,
knowing a place where an HIV test can be obtained
makes a big difference in whether one ends up getting
the test or not. The odds of getting a test for
people who did not know any such facility were only
42% compared to those who knew it (p=0.001).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
191
The possibility that availability and
accessibility of testing centers could depend on place
of residence was taken into account by introducing an
interaction variable between these two variables into
the regression model. However, this interaction term
did not make a significant contribution to the
explanatory power of the model, and was thus
disqualified.
The most interesting thing about the HIV test
model is that perception variables have not only
remained significant but have done so at very high
levels throughout the different stages of the
regression analysis. Specifically, the significance
of the severity perception of HIV/AIDS variable
remained unshaken throughout the analysis, while that
for cost of getting tested only dropped slightly from
p=0.001 for the first model to p=0.008 in the last
model. Perceptions therefore appear to have a strong
bearing on getting an HIV/AIDS test.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
192
5.3 Number of Sexual Partners
The logistic regression in this case was used to
estimate the odds of having one sexual partner for the
different perceptions specified under HBM, controlling
for an array of factors, namely age, sex,
socioeconomic status, place of residence, partner's
place of residence, and openness of one's community
with regard to discussion of HIV/AIDS. As was the
case with condom use and HIV testing, clustering was
found not to affect the results of the regression
analysis. A comparison between Model 4 in Table 10
and Table 11 confirms this.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
193
Table 10: Logistic Regression Model for the Odds of Having One
Sexual Partner: Different Scenarios under the Health Belief Model
Variable Model 1 Model 2
beta Odds
Ratio
beta Odds
Ratio
Intercept
* * *
3.244
(0.778)
* * *
3.153
(0.939)
Susceptibility
(Thinks Partner has
others = Reference)
Doesn't think Partner
has Others
-0.325
(0.291)
0.723 -0.265
(0.300)
0.768
Severity
(Knows someone with/died
from AIDS = Reference)
Doesn't Know anyone
with/died from AIDS
-0.241
(0.287)
0.785 -0.343
(0.295)
0.710
Cost of Having One
Partner
(Did not Exchange or
Receive Gifts/Money for
Sex = Reference)
Exchanged Gifts/Money
for Sex
***
-2.244
(0.690)
0.106
***
-2.331
(0.719)
0.097
Benefits-Having 1 Partner
(Having 1 Partner Does
Not Reduce Chances of
HIV Infection = Ref)
Having 1 Partner Reduces
Chances of HIV Infection
0.878*
(0.343)
2.406
0.875*
(0.350)
2.399
Self-Efficacy
(I Cannot Do Anything to
Reduce Chances of HIV
Infection = Ref)
I Can Something to
Reduce Chances of HIV
Infection
-0.428
(0.738)
0.652 -0.369
(0.744)
0. 692
*p < 0.05 **p < 0.01 ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
194
Table 10: Logistic Regression Model for the Odds of Multiple
Sexual Relationships: Different Scenarios under the Health Belief
Model (Continued)
Variable Model 1 Model 2
beta Odds beta Odds
Ratio
Age
(20-30=Ref)
15-19 0.340
(0.497)
1.405
31-39 0. 631
(0.420)
1.880
40-49 0.985
(0.563)
2. 677
Sex
(Male = Ref)
Female
***
1.002
2.724
Socioeconomic Status/Educ
(0 .297)
(>13 years = Ref)
0-7 years
0.095
0. 910
8-12 years
(0.558)
-0.337
0.714
Place of Residence
(0.517)
(Urban= Ref)
Rural
-0.683*
(0.336)
0.505
2
R
0.011 0.026
-2 LOG Likelihood
435.273 412.965
*p < 0.05 **p < o.oi ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
195
Table 10: Logistic Regression Model for the Odds of Multiple
Sexual Relationships: Different Scenarios under the Health Belief
Model (Continued)
Variable Model 3 Model 4
beta Odds beta Odds
Ratio
Intercept
* * *
3.023
(0.949)
20.55
0
***
2.942
(0.954)
18.953
Susceptibility
(Thinks Partner has
others = Reference)
Doesn't think Partner
has Others
-0.292
(0.302)
0.747 -0.306
(0.303)
0.736
Severity
(Knows someone with/died
from AIDS = Reference)
Doesn't Know anyone
with/died from AIDS
-0.337
(0.295)
0.714 -0.320
(0.296)
0.726
Cost of Having One
Partner
(Did not Exchange or
Receive Gifts/Money for
Sex = Reference)
Exchanged Gifts/Money
for Sex
***
-2.382
(0.728)
0.092
★ *
-2.321
(0.732)
0.098
Benefits-Having 1 Partner
(Having 1 Partner Does
Not Reduce Chances of
HIV Infection = Ref)
Having 1 Partner Reduces
Chances of HIV Infection
0.862*
(0.350)
2.367
0.877*
(0.351)
2.404
Self-Efficacy
(I Cannot Do Anything to
Reduce Chances of HIV
Infection = Ref)
I Can Something to
Reduce Chances of HIV
Infection
-0.357
(0.745)
0.700 -0.347
(0.746)
0.707
*p < 0.05 **p < o.oi ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
196
Table 10: Logistic Regression Model for the Odds of Multiple
Sexual Relationships: Different Scenarios under the Health Belief
Model (Continued)
Variable Model 3 Model 4
beta Odds beta Odds
Ratio
Age
(20-30=Ref)
15-19 0.360
(0.498)
1.433 0.349
(0.498)
1.418
31-39 0. 625
(0.419)
1.869 0. 647
(0.421)
1.910
40-49 1.019
(0.564)
2.770 1.051
(0.566)
2.859
Sex
(Male = Ref)
Female
***
0.979
2.660
***
0. 979
2.661
Socioeconomic Status/Educ
(0.298) (0.298)
(>13 years = Ref)
0-7 years
-0.028
0.972
-0.106
0.900
8-12 years
(0.563)
-0.294
0.745
(0.576)
-0.334
0.716
Place of Residence
(0.521) (0.517)
(Urban= Ref)
Rural
Partner's Place of
-0.689
(0.336)
0.336
-0.709*
(0.337)
0.492
Residence
(Same Locality/Village etc
= Reference)
Elsewhere
0.298
(0.308)
1.348
0.315
(0.309)
1.371
Discussion of HIV/AIDS
Issues
(Discussed HIV within last
4 weeks = Reference) 1.225
Did Not Discuss HIV last 4
weeks
0.203
(0.304)
2
R
0.027 0.027
-2 LOG Likelihood
412 . 001 411.557
*p < 0.05 **p < o.oi ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
197
Table 11: Logistic Regression Model for the Odds of Multiple
Sexual Relationships Taking Clustering into Account
Variable beta Odds Ratio
Intercept
***
2 . 628
(1.097)
13.840
Susceptibility
(Thinks Partner has
others = Reference)
Doesn't think Partner
has Others
-0.304 (0.303) 0.738
Severity
(Knows someone with/died
from AIDS = Reference)
Doesn't Know anyone
with/died from AIDS
-0.318 (0.297) 0.727
Cost of Having One
Partner
(Did not Exchange or
Receive Gifts/Money for
Sex = Reference)
Exchanged Gifts/Money
for Sex
★ **
-2.380
(0.738)
0.093
Benefits-Having 1 Partner
(Having 1 Partner Does
Not Reduce Chances of
HIV Infection = Ref)
Having 1 Partner Reduces
Chances of HIV Infection
0.907* (0.356)
2 .478
Self-Efficacy
(I Cannot Do Anything to
Reduce Chances of HIV
Infection = Ref)
I Can Something to
Reduce Chances of HIV
Infection
-0.234
(0.772)
0.791
*p < 0.05 **p < o.oi ***p < o.ooi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
198
Table 11: Logistic Regression Model for the Odds of Multiple
Sexual Relationships Taking Clustering into Account (Continued)
Variable beta Odds Ratio
Age
(20-30=Ref)
15-19 0.347 (0.498) 1.415
31-39 0. 658 (0.421) 1. 931
40-49 1.053 (0.565) 2.865
Sex
(Male = Ref)
Female
0.980*** (0.298)
2.663
Socioeconomic Status/Educ
(>13 years = Ref)
0-7 years
8-12 years
-0.126 (0.576)
0.882
Place of Residence
(Urban= Ref)
Rural
-0.542 (0.443)
0.709
Partner's Place of Residence
(Same Locality/Village etc =
Reference)
Elsewhere
0.315 (0.309)
1.370
Discussion of HIV/AIDS Issues
(Discussed HIV within last 4
weeks = Reference)
Did Not Discuss HIV last 4
Weeks
0.213 (0.305)
1.237
Cluster Membership
0.001 (0.002)
1.001
*p < 0.05 **p < o.oi ***p < 0.001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
199
Model 1 in Table 10 reflects the relationship
between perception variables and odds of having one
sexual partner. Only the perceptions of costs and
benefits associated with having one partner are
significant in decisions on the number of sexual
partners one has. The odds of having one partner for
an individual who perceives this to be costly are only
slightly over 10% relative to those who did not
(p=0.001). On the other hand, the odds of having one
partner for people who perceived benefits in terms of
reduced chances of HIV infection from having one
partner were more than twice those for people who did
not perceive such benefits (p=0.05).
Perceptions of susceptibility to HIV infection,
severity of HIV/AIDS, and self-efficacy with respect
to limiting the number of partners do not have any
significant association with the number of sexual
partners one has.
In Model 2 personal characteristics of age, sex,
socioeconomic status, and place of residence are
introduced.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
200
The odds of having one partner are significantly
higher for women, almost thrice, than men (p=0.001).
This confirms the prediction made earlier that men
would be more likely to have more than one sexual
partner, relative to women. It therefore appears the
argument that Setswana culture tends to turn a blind
eye to promiscuity among men is valid.
Place of residence is also significantly, though
barely, associated with the number of sexual partners,
with rural dwellers showing about half the odds of
having one partner relative to urban residents. This
finding contradicts the hypothesis postulated here,
that rural residents would be more likely to have one
partner than urban dwellers. The reason advanced here
was that rural areas tend to be more morally
reinforcing than towns and cities, therefore
minimizing socially deviant behavior such as multiple
relationships.
Age and socioeconomic status of the individual
have no significant association with the number of
partners one has.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
201
It is also clear that variables added in Model 2 have
not reduced the significance of perception variables
of cost and benefits associated with having one sexual
partner.
Model 3 introduces place of residence of the
respondent's sexual partner into the regression. As
seen in the review of literature, widespread concern
has been expressed over the tendency of some
employers, chief among them the government, to
separate couples by posting their employees to places
far away from their significant others. This then
leads to weakening of family ties and establishment of
new sexual relationships. As cab seen from this
model, this factor however does not have a significant
bearing on the number of sexual partners that one has.
The last variable, whether one ever discussed
issues of HIV/AIDS with anyone in the last 4 weeks
prior to the interview date, is introduced in Model 4.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
202
The logic behind this variable is that discussion of
HIV among the community in general contributes to
awareness creation, which might trigger behavioral
change such monogamous relationships.
Interestingly, even at this stage perceptions of
costs and benefits associated with number of sexual
partners still maintain their level of significance.
It appears therefore that decisions over having one
sexual partner is mainly a function of perceived costs
of not having multiple relationships, and benefits to
be derived from having one partner.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
203
CHAPTER 6
Discussion, Conclusions and
Recommendations
This chapter summarizes findings presented in
Chapters 4 and 5, discusses their implications
especially for HIV/AIDS policy in Botswana and makes
appropriate recommendations.
6.1 Condom Use
Chapter 4 looked at trends in HIV/AIDS awareness,
knowledge of transmission channels and practices to
prevent HIV infection. What comes out clearly in the
analysis is that awareness regarding HIV/AIDS is very
high in Botswana, and has always been so since the
early days of the epidemic. Perhaps even more
important is the fact that some differentials of
awareness have lost prominence over time, notably age
and place of residence.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
204
While in 1988 there were clear differences in
awareness between the young and old, with younger
women showing higher levels, this gap was no longer as
pronounced in 2001. The same development is observed
for differences between urban and rural residents.
This is probably a result of sustained public
education programmes that the government and other
interested parties have undertaken over the years. It
might also reflect the fact that since HIV/AIDS has
been around for almost two decades, people have
developed considerable knowledge about the disease
over time through other means such as the media,
personal research, social interaction, etc.
Effort was also made to assess awareness in terms
of preventive methods. Awareness of the effectiveness
of the condom as a preventive measure increased
immensely between 1988 and 2001, even though 15 to 29
year olds maintained the highest levels of awareness
in this regard during this period.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
205
An interesting finding was made with respect to
the number of sexual partners. Results presented here
suggest vagueness in the understanding of risk of
infection posed by multiple relationships in the early
days of the HIV/AIDS epidemic, probably emanating from
inaccurate public messages during that time. In 1988
most people believed that limiting the number of
sexual partners reduced the risk of HIV infection.
Indeed this was misleading, and fortunately this seems
to have been recognized by 2001, when a majority of
people specifically cited having one partner who had
no other partners as a means of reducing chances of
infection. Public messages therefore seem to have
grown more accurate over time, as people show clearer
understanding of the risk associated with multiple
relationships.
In addition, it appears public messages have also
narrowed their focus to the local level over time.
For example, in 1988 there was high awareness about
possible transmission of HIV through "dirty" needles;
but by 2001 this transmission was hardly mentioned.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
206
Intravenous drug use is not a problem in Botswana, and
this possibly explains why "dirty" needles have
disappeared from the list of commonly mentioned
transmission channels. Public messages could
therefore be seen, in this light, to have improved not
only in accuracy but also in relevance. The danger to
this however is that as globalization intensifies and
accessibility of all parts of the world improves, what
used to be "foreign" to some places may soon become
local - the challenge therefore remains for public
health programmes to monitor intravenous drug use and
scale their responsiveness accordingly.
Overall, therefore, it appears people do not only
know about the existence of HIV/AIDS, they are also
aware of its prevention channels, in their own
context. This is an encouraging finding in the sense
that the situation seems conducive for intervention;
people know about the disease and how it can be
stopped. The question however remains - why does
Botswana still have high STD prevalence?
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
207
Obviously, this suggests that people are not using
condoms, and are possibly still engaging in multiple
sexual relationships despite their awareness of the
inherent risk.
Chapter 5 investigated this issue further, and
found that condom use is largely a function of
perceptions of cost of use, benefit of use, and
perceived vulnerability to HIV/AIDS. In other words,
awareness in itself is not enough motivation for the
adoption of preventive methods; an individual's
behavior is more an outcome of how they perceive their
own situation with respect to HIV/AIDS.
Perceived costs of condom use are particularly
important as a determinant of actual usage, people who
perceive costs to condom use are significantly less
likely to use it relative to those who do not. Even
when other factors such as age, sex, age difference
between partners, socioeconomic status, marital
status, place of residence, and accessibility of
information are controlled for, this perception
remains significant at very high levels.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
208
There is also a significant interaction between
perceived costs of condom use and age of the
individual. Young people do not seem to attach much
cost to condom use, but as age increases the situation
changes significantly. Also, a test in the difference
between the coefficients of costs for young and old
people turns out significant, confirming the
difference in the perception of cost of condom use
between them.
The above finding makes sense - for young people,
using a condom is part of their reality, they grew up
with the threat of HIV/AIDS around them, together with
information about prevention and safety. On the other
hand, older people have to change their lifestyle;
they were never under so much pressure to use condoms
when they entered the age of sexual maturity, but of
all of a sudden they have to readjust their sexual
ways. Obviously they have been exposed to sex without
the condom, and invariably weigh up its advantages
against those of use.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
209
There is therefore an obvious need for age
specific intervention with regard to condom promotion
as an integral part of the anti-AIDS campaign.
Educational programmes need to devise different
strategies for different age groups. While for the
young generation emphasis need not be eased, messages
especially targeted towards older members of the
society need to be developed. Specifically, issues of
costs associated with using the condom should be
tackled, emphasizing benefits and addressing common
myths about its use.
This analysis also went further and checked for
interaction between socioeconomic status and
perceptions towards HIV/AIDS and condom use. Results
indicate that there is significant difference between
high and low socioeconomic status people's perceptions
of the severity of HIV/AIDS. Low SES people are
significantly less likely to perceive HIV/AIDS as a
severe disease relative to the high SES group.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
210
The Health Belief Model informs us that people who
perceive a disease as severe are more likely to take
preventive measures compared to those who do not.
It has already been seen from Chapter 4 that
differences in awareness in terms of education have
not closed quite as fast as those for age and place of
residence. This probably explains at least partly why
low SES people's perception of HIV/AIDS is that of
relatively less severity.
This discrepancy needs to be addressed urgently.
One way could be for public messages to be simplified
and presented largely in local languages for people to
understand, emphasizing the danger posed by HIV/AIDS.
While this project did not obtain data about the
quality and usefulness of current HIV/AIDS information
available in Botswana, it did notice the large number
of billboards written in English across the whole
country. In a country with about an estimated quarter
of its population neither able to read nor write, such
an approach leaves a considerable number of people
uninformed.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
211
Interaction was also checked between perceptions
and gender. While most of the perception variables do
not seem to differ between men and women with respect
to their influence of condom use, the perception of
susceptibility is significantly associated with condom
use for men, and not for women. The odds of condom
use among men who feel susceptible to HIV/AIDS are
nearly twice compared to those for men who do not.
Women, however, do not seem able to translate their
feeling of susceptibility into behavioral change.
This confirms their weaker position in negotiating
safe sex as alleged in the literature. To them,
vulnerability does not make a difference in whether or
not condoms are used.
This is an important finding in that while it
confirms what was already known, it also demonstrates
the magnitude of the problem. The question that could
be asked at this point is what policy action could be
taken to correct this imbalance?
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
212
There are two options, and these could be seen to be
mutually reinforcing. HIV/AIDS prevention programs
need to lobby society for the empowerment of women,
particularly in issues of sexual decision making.
Condom use should not be seen as men's domain,
especially in the era of the female condom. On the
other hand, men should also be brought to the center
stage of prevention programs regarding condom use;
their acceptance of condoms, especially recognition
and appreciation of women's contribution in sexual
decision making needs to be emphasized.
This imbalance is however a product of century
old power-sharing traditions, and should not be
expected to disappear in the short term.
HIV/AIDS Testing
As already seen, not much information is
available on HIV testing in Botswana. What is known
however is that since its inception, turn out has been
disappointing.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
213
It is therefore critical to understand factors that
hinder people from getting tested, so that appropriate
measures could be put in place in favor of testing.
Specifically, people's perceptions of getting tested
need to be understood in order for policy to intervene
accordingly.
Results of the HIV test analysis indicate that
perceptions of severity and cost of getting an HIV
test are significantly associated with getting tested.
Relative to people who did not perceive much severity
of HIV/AIDS, people who did show significantly greater
odds of getting an HIV test. Similarly, the odds of
getting a test among people who perceived greater
costs to getting a test were significantly less than
for those who did not, presumably due to fear of
social exclusion.
Efforts to persuade people to get tested could
benefit from emphasis of the severity aspect of
HIV/AIDS. Particularly beneficial is the close
relationship between testing and prospects of living
longer, fruitful lives due to availability ARVs.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
214
It is also possible that as more people develop full
blown AIDS and begin to show its signs, severity of
the disease would become more apparent, with the
effect of motivating people to go for a test.
The relationship between perceived costs of
getting tested and obtaining the test is also worth
highlighting. It has already been seen that
acceptance of PLWAs grew considerably in Botswana
between 1988 and 2001, by age, place of residence and
education. Results suggest that fears still exist
about getting tested despite the noted gains with
respect to social acceptance of PLWAs.
Intervention aimed at motivating people to get
tested therefore needs to address the element of fear
of getting tested. While people have been encouraged
to go public about their HIV status, it could also be
helpful for those who do so to go beyond simply
stating their HIV positive status and share their
experiences regarding the reception they received from
family, the workplace, etc.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
215
By so doing, issues of stigma and discrimination would
be brought to the center stage of public discussion
and, in the long run, help foster change towards even
greater acceptance of PLWAs. UNAIDS has already taken
the initiative by adopting an appropriate slogan "Live
and Let Live" as a central theme for its 2002/2003
World AIDS Campaign - this effort could therefore be
undertaken in this spirit.
Of the control variables used in the HIV test
model, only place of residence and knowing a place
where a test could be obtained were significant
determinants of the outcome. The odds for rural
residents to get a test were much less relative to
urban dwellers. Likewise, the odds of getting a test
were a lot higher among people who knew a testing
center compared to those who did not. Promotion of
HIV testing in rural areas as well as provision of
more testing facilities will therefore have to feature
prominently as part of programmes that seek to
encourage people to get tested.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
216
6.2 Number of Sexual Partners
On the number of sexual partners, perception
variables once again maintain a steady level of
significance throughout the regression, particularly
costs and benefits associated with multiple
relationships. Of the intervening variables only sex
and place of residence come out significant.
The number of sexual partners an individual has
is associated with the perceived costs attached to
having more than one partner, and this relationship
remains highly significant throughout the analysis.
People who report deriving some benefits such as money
or other gifts for sex have greater odds of being in
multiple relationships. Such people obviously attach
great cost to staying in one relationship since this
may not give them the material benefits possible with
several relationships.
There is no easy way to circumvent this
situation, since it is driven directly by macro level
processes such as unemployment, poverty, and general
social deprivation.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
217
It is important however for this relationship to be
taken into cognizance, especially within policy making
organs of government. Policies that seek to create
employment opportunities and promote social equity and
eliminate poverty have an added benefit in the sense
that they would indirectly contribute to the
prevention of HIV/AIDS. Such realization would
hopefully add impetus to policy implementation and in
the process weaken people's need for engaging in
multiple sexual relationships in order to make ends
meet.
6.4 Summary
This project has broken new ground for research
seeking to understand HIV preventive behavior and the
HIV/AIDS epidemic in Botswana. Hopefully, more
research of this nature will follow.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
218
In this way social research could contribute more
meaningfully to the solution of the HIV/AIDS problem
instead of limiting itself to description of
prevalence levels and impact of the disease.
There is also a dire need for more data. While
this project benefited immensely from BAIS, the need
for issue-specific and theoretically inclined data
remains stark clear. In the absence of large budget
privately funded social research on HIV/AIDS, the
government should consider smaller-scale, more focused
research to address specific preventive issues such as
condom use, HIV testing, etc. Dynamics fuelling these
phenomena are critical in understanding trends in
HIV/AIDS incidence and prevalence, which could help
position policy strategically for maximum
effectiveness.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
219
Bibliography
1. Andrews, S., Williams, A.B. & Neil, K. (1993).
The Mother-Child Relationship in the HIV-I
Positive Family Image. Journal of Nursing
Scholarship, 25(3), 193-198
2. Antoni, M.H., Schneiderman, N. A., O'Sullivan, L.
M.J., Marks, J., Efantis, J., Skyler, J., &
Fletcher, M.A. (1992) Mothers with AIDS. In P.I.
Ahmed (Ed.), Living and Dying with AIDS
(pp. 47-85). Plenum Press
3. Arimistad, L., & Forehand, R. (1995). For whom
the Bell Tolls: Parenting Decisions and
Challenges Faced by Mothers who are HIV
Seropositive. American Psychological Association
2(3), 239-250
4. Bloom, D.E. & Carliner, G. (1988). The Economic
Impact of AIDS in the United States Science, New
Series, 239, 604-610
5. Bongaarts, J. (1996). Global Trends in AIDS
Mortality. Population and Development Review
22(1), 21-45
6. Caldwell, J.C., Caldwell P., & Quiggin, P.
(1989). The Social Context of AIDS in sub-Saharan
Africa. Population and Development Review 15,
185-234
7. Central Statistics Organization-Republic of
Botswana. (2002). Botswana AIDS Impact Survey
2001. Gaborone, Botswana.
8. Central Statistics Organization - Republic of
Botswana. (2001). Botswana Demographic Survey
1998. Gaborone, Botswana.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
220
9. Central Statistics Organization - Republic of
Botswana. (1999). Botswana Family Health Survey
III (1996). Gaborone, Botswana.
10. Central Statistics Organization - Republic of
Botswana (1999); Botswana Family Health Survey
III (1996). Gaborone, Botswana
11. Chen, P. Y., & Popovich, P. M. (2002)
Correlation: Parametric and Nonparametric
Measures. Sage Publications, Inc.
12. Chirimuuta, R.C. & Chirimuuta, R. J. (1987).
AIDS, Africa and Racism. Bretby House,
United Kingdom.
13. DeMaris, A. (1995). A Tutorial in Logistic
Regression. Journal of Marriage and the Family
57, 956-968
14. Fee, E. & Krieger, N. (1994). Thinking and
Rethinking AIDS: Implications for Health
Policy. In N. Krieger and M. Margo (Eds.) AIDS:
The Politics of Survival (pp. 227-277)
15. Hair Jr, J. F., Anderson, R. E., Tatham R. L.,
Black W. C. (1995) . Multivariate Data Analysis,
with Readings (4th ed.), Prentice Hall
16. Harris, D.M., & Guten S. (1979). Health
Protective Behavior: An Exploratory Study.
Journal of Health and Social Behavior 20,
(17-29)
17. Harris, M. B. (1995). Basic Statistics for
Behavioral Science Research, Allyn and Bacon.
Needham Heights, Mass. 02194
18. Hombs, M. E. (1992). AIDS Crisis in
America: A Reference Handbook ABC-CLIO
19. Idele-Akwara, P. (2002). The Social Context of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
221
Perception of AIDS Risk and Sexual Behavior in
Kenya. Paper presented at the Annual Population
of America in Atlanta, GA.
20. Kayongo-Male, D., & Onyango, P. (1994). The
Sociology of the African Family, London:
Longman.
21. Langlie, J. K. (1977). Social Networks, Health
Beliefs, and Preventive Health Behavior, Journal
of Health and Social Behavior, Vol 18(3)
244-260.
22. Laumann, E. 0., Gagnon, J. H., Michales, S.,
Michael R. T., & Schumm L. P. (1993). Monitoring
AIDS and Other rare Population Events: A Network
Approach. Journal of Health and Social Behavior,
34(1) 7-22.
23. Mann, J., Tarantola D.J.M., & Netter T.W. (Eds.)
(1992). AIDS in the World: A Global Report,
Havard University Press.
24. May, R.M. & Anderson, R.M. (1988). The
Transmission Dynamics of Human Immunodeficiency
Virus (HIV). Philosophical Transaction of the
Royal Society of London - Series B: Biological
Sciences 321(1207) 565-607.
25. Mernard, S. (2002). Applied Logistic Regression
(2nd ed.) Sage Publications Inc. Thousand Oaks
CA 91320
26. Ministry of Labor and Home and Home Affairs -
Republic of Botswana. (2001). Report of the
First National Conference on Gender and
HIV/AIDS (21st to 23rd June 2001, Boipuso Hall,
Gaborone). Gaborone, Botswana.
27. Pampel, F. C. (2 000) . Logistic Regression - A
Primer. Sage Publications Inc. Thousand Oaks,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
222
CA 91320
28. Pratt, L. The Relationship of SES to Health.
American Journal of Public Health 61, 281-291
29. Sharma, S. (1996). Applied Multivariate
Techniques. John Wiley and Sons, Inc.
30. Stegling, C. (2000). Current Challenges of
HIV/AIDS in Botswana. University of Botswana,
Gaborone, Botswana.
31. Suchman, E. (1964). Sociomedical Variations
Among Ethnic Groups. American Journal of
Sociology 70:319-331
32. Suchman, E. Social patterns of illness and
medical Care. in E.G. Jaco (Ed.), Patients,
Physicians, and Illness (pp. 262-79). New York:
Free Press
33. The World Bank. (1999). Confronting AIDS: Public
Priorities in a Global Epidemic. Oxford
University Press.
34. Joint United Nations Programme on HIV/AIDS.;
(2002) . Report on the Global HIV/AIDS Epidemic
2002. UNAIDS.
35. United Nations Organization. (1989) .Levels and
Trends of Contraceptive Use as Assessed in
1988. United Nations: New York.
36. United Nations Development Programme
& Government of Botswana. (2000). Socio-economic
Impact of HIV/AIDS: An Impact Assessment of
HIV/AIDS on Current and Future Population
Characteristics and Demographics in Botswana.
UNDP.
37. United Nations Development Programme
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
223
& Government of Botswana. (2000). Botswana Human
Development Report 2000 - Towards an AIDS-Free
Generation, UNDP.
38.Vanlandingham, M.J., Suprasert, S.,
Grandjean,N., & Sittitrai W. (1995). Two Views of
Risky Sexual Practices Among Northern Thai Males:
The Health Belief Model and The Theory of
Reasoned Action. Journal of Health and Social
Behavior, 36,(2), 195-212
39. Van Ryzin, J. (1977). Classification and
Clustering, Proceedings of an Advanced Seminar
Conducted by the Mathematics Research Center -
The University of Wisconsin, May 3-5, 1976,
Academic Press.
40. Walker, Sharon E. (1998). Women with AIDS and
Their Children. Garland Publishing Inc.
39. Webb, D. (1997). HIV and AIDS in Africa, Pluto
Press.
40. Ybarra, S. (1991). Women and AIDS: Implications
for Counseling Journal of Counseling and
Development 69, 285-287
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Factors affecting child survival in Pakistan

PDF

Changing patterns of interracial marriages of Asian Indians in California

PDF

Assimilation and ethnicity: Adaptation patterns and ethnic identity of Armenian -Americans in central California

PDF

Socioeconomic determinants of infant mortality in Botswana: 1978-1988

PDF

Heart disease among middle -aged and elderly persons in the United States: Trends and a multistate model

PDF

Domestic violence in Korean immigrant families

PDF

"Relative strangers": An ethnography of grandmothers raising grandchildren in the Los Angeles County welfare system

PDF

Interbirth interval analysis of fertility: The case of Kenya in 1993

PDF

A cultural -ecological analysis of educational aspirations and satisfaction of educational opportunities of Asian -Americans and Hispanics: Understanding the interaction of race and class, paren...

PDF

Cross -cultural examination of mental health measures in dementia caregivers: Assessment of the Center for Epidemiologic Studies -Depression Scale and the Zarit Burden Inventory

PDF

Dissonance in the population -environment movement over the politics of immigration: Shifting paradigms of discourse vis -a -vis individual rights and societal goals

PDF

Dementia caregiving and ethnicity: African American caregivers and the sociocultural stress and coping model

PDF

Family functioning, intergenerational conflict, and psychological symptomology of Asian American college students

PDF

Effects of a brief safer sex intervention on provider counseling behaviors, attitudes, and the provider -patient relationship

PDF

Effects of cultural values and behaviors in sociocultural stress and coping among Korean-American and Korean caregivers

PDF

Filial expectations and social exchange patterns among older Taiwanese parents and their adult children

PDF

Social Class And Attitudes Toward Sexually Oriented Materials

PDF

Constructing the Zhonghua minzu: The frontier and national questions in early 20th century China

PDF

Gender role conflict, personality, and help -seeking in adult men

PDF

A daily diary approach to compare the accuracy of depressed and nondepressed participants' estimation of positive and negative mood: A test of the depressive realism hypothesis

Asset Metadata

Creator Fidzani, Boga (author)

Core Title HIV/AIDS preventive behavior in Botswana: Trends and determinants at the turn of the 21st century

Contributor Digitized by ProQuest (provenance)

School Graduate School

Degree Doctor of Philosophy

Degree Program Sociology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,sociology, demography,Sociology, general

Language English

Advisor James, Angela (committee chair), Ransford, Edward (committee member), Silverstein, Merril (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-635054

Unique identifier UC11335001

Identifier 3116698.pdf (filename),usctheses-c16-635054 (legacy record id)

Legacy Identifier 3116698.pdf

Dmrecord 635054

Document Type Dissertation

Rights Fidzani, Boga

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA