Advances in Applied Sociology
2012. Vol.2, No.3, 167-178
Published Online September 2012 in SciRes (http://www.SciRP.org/journal/aasoci) http://dx.doi.org/10.4236/aasoci.2012.23023
Migration, Sexual Behavior and Perceptions of Risk: Is the Place
of Origin a Factor in HIV Infection?
Sitawa R. Kimuna1, Yanyi K. Djamba2
1Department of Sociology, East Carolina University, Greenville, USA
2Centre for Demographic Research & Department of Sociology, Auburn University at Montgomery,
Montgomery, USA
Received April 29th, 2012; revised June 5th, 2012; accepted June 15th, 2012
Migration is an important process of change, especially for populations in developing countries. Just by
moving to new places, migrants are different from those who do not migrate in terms of socio-demo-
graphic characteristics. This study focuses on migration in Kenya and its interaction with human immu-
nodeficiency virus (HIV) risk. Two main research questions are addressed: To what extend does the sex-
ual behavior of migrants differ from non-migrants? Do migrants know more about HIV risk than non-
migrants? The analysis is based on the 2003 Kenya Demographic and Health Survey data. The results
show that migrants are significantly more likely to report fear of HIV infection than non-migrants. The
perception of risky sexual behavior is significantly correlated with non-use of condoms for migrants than
for non-migrants. Migrants who perceive themselves as being at risk of HIV infection are less likely to
use a condom at their last non-marital sexual encounter. Also, migration is significantly correlated with
multiple sexual partners. There is a remarkable difference in the mean age of migrants and non-migrants;
migrants on average are significantly older and more likely to be married than non-migrants.
Keywords: Internal Migration; Risky Sexual Behaviour; STIs/HIV; Kenya; Africa
Introduction
Migration and risky sexual behaviour that might lead to
sexually transmitted infections (STIs) and human immunodefi-
ciency virus (HIV) infection have been examined separately in
sub-Saharan Africa; however, we are still far from understand-
ing in detail just how and to what extent migration affects the
spread of STIs and HIV. Previous studies of migration and
health tend to concentrate on the urban or receiving areas with
little attention paid to people living in the rural or sending areas.
Research on the dimensions of social and health impacts of
internal migration would illuminate our understanding of the
consequences of migration on risky sexual behaviour, espe-
cially the vulnerability of migrant populations. Studies have
shown that people who are more mobile, or who have recently
changed residence, tend to be at higher risk for STIs/HIV than
those in more stable living arrangements (Arnafi, 1993; Pison
et al., 1993; Brockerhoff and Biddlecom, 1999; Lurie et al.,
2003; Nyanzi et al., 2004). For example, itinerant traders and
long-distance truck drivers have shown to have an increased
risk of contracting STIs and HIV (Wilson et al., 2000).
Therefore, it can be argued that mobility creates conducive
environments for high risk sexual behaviour, which is deter-
mined by the number of recent heterosexual partners and by
non-use of condoms with these partners. Of particular interest is
whether migrants’ previous exposure to urban areas increases
their likelihood of high-risk sexual behaviour in rural areas, for
example, through socialization to less restrictive sexual norms
or acquisition of enabling characteristics (e.g., wealth) in urban
areas (Kimuna and Djamba, 2005). Another factor that has been
cited in the literature is the separation from a regular sexual
partner or spouse (Wilson, 1972; Nyanzi et al., 2004). These
studies indicate that the risk of engaging in risky sexual behav-
iour increases even for married persons when they are away
from their spouses.
Other studies have shown that migration of young, unmarried
adults from rural areas, which presumably are known to har-
bour conservative ideas about sexuality and sexual behaviour
than urban centres, have been regarded as partly responsible for
high STIs/HIV observed in urban centres (Brockerhoff and
Biddlecom, 1999). Others have noted the rise in the number of
HIV infections in rural areas due to the return migration to rural
communities by those who previously had been living and
working in towns and cities (Topouzis and du Guerney, 1999).
The association between migration and HIV infection has
been echoed in a number of studies (Arnafi, 1993; Lurie et al.,
2003; Nyanzi et al., 2004; Lurie, 2006). Also, the spread of
infectious diseases can be exacerbated by inadequate structural
arrangements that contain the diseases within an environment,
especially in rural areas (Anarfi, 2005). Travel between places
with different health risk profiles places people in environments
where their health is subject to new influences and impacts.
This is because migrants’ behaviour away from home often
differs from one that is maintained while at home and the social
norms that guide and control their behaviour in stable home
communities are often missing.
For example, in a study of itinerant traders in Uganda,
Nyanzi et al. (2004) found that the urban infrastructure and
availability of social and economic opportunities in these areas
not only acted as magnates for labour migration but also pro-
vided an environment that led to risky sexual behaviour with
new partners. Nyanzi and colleagues found that most of the

S. R. KIMUNA, Y. K. DJAMBA
itinerant traders were “single” in town and ‘married’ in their
rural homes (Nyanzi et al., 2004: p. 244). Their single status
allowed them to have a myriad of sexual networks, which in-
cluded various types of partners: commercial sex workers,
one-night stand women, semi-permanent sex partners (or what
in sexual behaviour literature is known as “second office”),
lovers, and legal wives. When these migrants move without
their partners, and probably because of loneliness or as a result
of social pressure, they may engage in behaviour that may place
them at risk of contracting STIs/HIV.
Kenyan Context
Historically, Kenya has had a circular pattern of population
movement since independence and the ghost syndrome wit-
nessed in most Kenyan cities during major holidays attests to
this pattern. Further, Kenya’s population is still predominantly
rural with 80 percent of the population living in the countryside
(CBS et al., 2004). Hence, the widespread assumption that
wives of migrants remain in rural areas to take care of the rural
household has been used to explain the circular pattern of mi-
gration in Kenya. The high mobility of Kenyan population is
due to the fact that young people graduating from rural schools
move to urban areas to seek employment. These young people
return home periodically depending on how far the urban areas
are from their places of origin. Elsewhere, this pattern has also
been linked to the spread of STIs/HIV in rural areas (Lurie et
al., 1997). Furthermore, migrants from rural areas to urban
areas in search of better economic opportunity may be at risk
when they arrive at their destination. Particularly, when sepa-
rated from the social controls exercised by families, communi-
ties, and wider social norms. The situation can be exacerbated
by rural-urban migrants experiencing emotional instability on
exposure to the urban environment, which can lead to potential
high risk sexual relationships.
Although Kenya’s rates of new HIV infections show a de-
cline (UNAIDS, 2006), its prevalence rate of 7 percent among
adults is still alarming (CBS et al., 2004). The situation is even
more critical in urban areas where 10 percent of residents are
HIV positive compared to 6 percent in rural areas. This would
mean that migrants moving to urban centres have a greater risk
for contracting STIs/HIV as well as non-migrants in rural areas,
who have a sexual network of migrants.
This study focuses on migration in Kenya and its interaction
with HIV risk. Two main research questions are addressed: To
what extend does the sexual behavior of migrants differ from
non-migrants? Do migrants know more about HIV risk than
non-migrants? The paper aims to make a contribution to the
evidence of the association of migration and risky sexual be-
haviour by systematically examining the impact of migration
status and migration stream on both the perceived and actual
risks of HIV infection among both men and women. Our main
focus is on the extent to which internal migration effects a
change in the environment and contributes to the spread of
STIs/HIV. Internal migration is defined within this paper as the
movement of people from one place of residence to another in
search of economic opportunities for a period of time within the
same national territory. Studies have shown that this type of
migration process exists in most sub-Saharan African countries.
This study does not present data or sero-prevalence.
We attempt to answer the following questions by examining
the influence of migration status and migration stream on risky
sexual behaviour of men and women by place of residence:
How do migrants and non-migrants perceive their risk of infec-
tion? Do migrants and non-migrants have the same risk in
terms of condom use and multiple sexual partners?
Although this research does not focus on gender differences,
it recognizes that migration and sexuality are gender-specific
phenomena. For example, a study in Ethiopia and South Africa
showed that, although men and women expressed the same
intentions for moving, their actual migration experiences were
significantly different (Djamba, 2003).
Also, underlying the relationship between migration and
risky sexual behaviour are clear gender differences, which other
studies have noted and argued that the difference in having
multiple sexual partners is more prevalent among men (Kimuna
& Djamba, 2005) than among women (Brockerhoff & Biddle-
com, 1999). This is partly because in most societies, men have
the cultural prerogative to initiate and negotiate sexual rela-
tionships and women do not. In addition, women are more
likely to apply an economic rationale to sexual risk taking such
as “survival sex” (Barnett & Whiteside, 2006; for a detailed
discussion of “survival sex,” see Leclerc-Madlala, 2003). There-
fore, we conducted the analysis of the role of migration in risky
sexual behaviour separately for men and women.
Data and Method
Data
This study is based on data from the 2003 Kenya Demo-
graphic and Health Survey (KDHS). The 2003 KDHS is a na-
tionally representative sample of adult Kenyans. Data were
collected on health, sexual behaviour, marital status, and
household characteristics. The survey instruments were based
on model questionnaires developed by the MEASURE DHS+
program. The survey was implemented by the Kenya Central
Bureau of Statistics (CBS) in collaboration with other local
agencies. The survey response rates were high (94% for women
and 86% for men), with a total of 8195 women and 3574 men
successfully interviewed. More details about survey design and
methodology were published elsewhere (CBS et al., 2004).
Variables
This study focuses on the extent to which migration effects a
change in the environment and contributes to risky sexual be-
haviour. We consider three dependent variables that measure
risky sexual behaviour: 1) perceived risk of HIV; 2) condom
use at last non-marital sexual intercourse; and 3) having multi-
ple sex partners during the last 12 months. Because of the ques-
tionable sexual behaviour data based on self-reports, perceived
risk of HIV infection has been identified as a preferred proxy
variable to looking at HIV vulnerability (Smith & Watkins,
2005) from the respondent’s point of view. Condom use is
known to reduce the risk of sexually transmitted diseases in-
cluding HIV/AIDS. Number of sexual partners is known to be
positively associated with the risk of contracting STIs/HIV
(Ntozi & Ahimbisibwe, 1999; Zulu et al., 2002).
Migration is the main explanatory variable. It was defined
using information from the question on the duration of resi-
dence at the current place of residence. Respondents were asked,
“How long have you been living continuously in (name of the
current place of residence)?” There was a choice of three an-
swers: 1) “number of years,” for those who had lived elsewhere;
168

S. R. KIMUNA, Y. K. DJAMBA
2) “always,” for those who never moved, and 3) “visitor”. After
examining the frequency distribution of the answers, we re-
tained the first two categories because the visitor category was
negligible (1% in the male sample and 2% in the female sam-
ple). The first category included in this analysis represents mi-
grants or those who have moved. The second category includes
non-migrants or, those who have always lived at their current
place of interview.
We also used the information on previous place of residence
for migrants to construct migration streams. The information
was obtained from the responses to the question, “Just before
you moved here, did you live in Nairobi, Mombasa, in another
city or town, or in the countryside?” This question was asked
only of those who moved. Thus, we have not only the respon-
dents’ migration status, but also their migration stream status
measured by their last move.
We included two other correlates of sexual risk in our analy-
sis: age at first sexual intercourse and knowing someone who
has or died of AIDS. Age at first intercourse was dichotomized
dividing respondents into two categories: 1) those who had sex
before age 15 and 2) those who had sex at age 15 or later. The
second category also included respondents who had not yet had
sex (16% of all male respondents and 18% of all female re-
spondents were in that category).
Sexual behaviour at a young age poses multiple risks includ-
ing prolonged exposure to risk of being infected with STIs and
HIV/AIDS. Furthermore, young girls are less likely to know
safe sex methods and more likely to be sexually disadvantaged
due to myriad reasons including lack of bargaining power.
Longfield et al. (2004) have argued that young women who
have sex with older men are unlikely to insist on condom use.
Previous research has produced conflicting results on the effect
of knowing someone who has or died of AIDS on behaviour
change; some found positive and significant behaviour changes
(Palekar et al., 2007) and others denoted no significant associa-
tion (Camlin & Chimbwete, 2003). We re-examined this asso-
ciation separately for men and women.
Socio-demographic variables included in the analysis are:
age, marital status, education, ethnicity, and religion. Age was
broken down into three categories: 15 - 19, 20 - 29, and 30+. In
a country where life expectancy remains relatively low, this age
classification is important in order to examine potential differ-
ences in risky sexual behaviours between teenagers, young
adults, and older adults. Marital status was categorized into
currently married and not currently married. Education variable
had two categories: less than high school and high school or
higher.
Analytical Appro ach es
Our analyses were conducted in two steps. First, we used
descriptive statistics to examine the characteristics of the study
samples and to explore binary differences between migrants
and non-migrants. Second, we conducted multivariate analyses
to examine the impact of migration on risky sexual behaviour,
net of the effects of other variables. We dichotomized the de-
pendent variables with “1” (or yes) indicating that the respon-
dent engaged in risky sexual behaviour (e.g., considered
him/herself to be at risk of being infected with HIV, did not use
a condom at the last non-marital sexual intercourse, or had sex
with multiple sexual partners), and “0” otherwise.
Because each of our three dependent variables was dichoto-
mized, we fitted adjusted logistic regression models in the form
of:


01122 nn
logRSB1-RSB = β+ βX+ βX+ ... + βX

where RSB represents the probability that the respondent en-
gaged in risky sexual behavior; X1, X2, ···, Xn are the inde-
pendent variables, and β0, β1, β2, ···, βn are the regression coef-
ficients from the fitted models.
Our analyses focused on the complexity of the migration
patterns in Kenya as well as the vulnerability of migrants and
their networks (those with whom they interact). Because mi-
grants are a heterogeneous group, it was important to examine
how their sexual behaviours, which may lead to infection, var-
ied between migrants and non-migrants, separately in urban and
rural areas. These variables and other important correlates of
risky sexual behaviour and migration are shown in Table 1.
Table 1.
List of variables used in the analysis of migration and risky sexual
behaviour in Kenya, KDHS 2003.
Variables Categories
Risky se x ual behaviour
0 = No
Perceives chance of getting AIDS
1 = Yes
0 = No
Used condom at last non-marital
sex 1 = Yes
0 = One or none
Had multiple sexual partners in
last 12 months 1 = Two or more
Migration status and migration
stream by residence
0 = Non-migrant
1 = Urban-urban migrant Urban residence
2 = Rural urban
0 = Non-migrant
1 = Urban-rural migrant
Rural residence
2 = Rural-rural
Socio-demographi c and other
variables
1 = 15 - 19 years
2 = 20 - 29 Age
3 = 30 or older
0 = Not currently married
Marital status 1 = Currently married
0 = Less than High School
Education 1 = High School or higher
0 = Other ethnic group
Ethnicity 1 = Kikuyu
0 = Other religious affiliation
Religion 1 = Muslim
0 = At age 15 or later
Age at first sex 1 = Before age 15
0 = No
Knows someone who has or died
of AIDS 1 = Yes

S. R. KIMUNA, Y. K. DJAMBA
170
Some Limitations
The data set has some limitations. DHS questionnaires were
primarily not designed to collect data on sexual behavior.
Rather, they have traditionally sought information on fertility,
contra-ception, infant and child health, and mortality. Questions
on sexual behavior have been introduced only recently, and
responses to such questions may suffer from reporting prob-
lems. Data on risky sexual behaviors collected through popula-
tion surveys have been questioned for many years because they
are based on self-reported behavior (Aral & Peterman 1996;
Fishbein & Pequegnat, 2000; Cleland et al., 2004).
Questions about sexual behavior are sensitive, and men may
distort their reports to correspond to socially desirable re-
sponses (Bessinger, Akwara, & Halperin, 2003; Mensch et al.,
2001). Although researchers must rely on people’s honesty in
reporting their sexual activity, there are indications that people
don’t always do this. Sometimes people may not recall exactly
what age they were when they had their first sexual intercourse,
or the number of sexual partners they have had over a given
period. Perhaps, that is why the 2003 KDHS lacks data on the
number of lifetime sexual partners or number of sexual partners
during the last five to 10 years.
Further, other common critiques are that commercial sex
workers and clients with numerous liaisons are likely to report
rounded numbers of recent partners; men may inflate (and
women underreport). Nonetheless, the DHS questionnaires
have been tested in numerous countries and are adapted to each
country-specific setting, thus reducing the sensitivity of the
questions. Also, we tried to reduce some of the bias by di-
chotomizing the number of sexual partners (0 = did not have
multiple sexual partners; 1 = had multiple sex partners).
Results
Descriptive Analysis
The study sample consists of 3537 men and 8195 women, but
the actual totals are smaller for some variables due to missing val-
ues (see Table 2). The majority of men and women considered
Table 2.
Percent distribution of respondents by selected variables, KDHS 2003.
Men Women
Variables N % N %
Perceive d risk of getting AIDS
No risk at all 1243 35.2 3014 37.5
At risk 2293 64.8 5016 62.5
Condom used at last non-marital sex
Yes 366 47.4 240 25.5
No 406 52.6 703 74.5
Had multiple s ex partners i n l ast 12 months
Yes 402 11.3 128 1.6
No 3171 88.7 8040 98.4
Migration status
Non-migrant 1603 45.3 2230 27.9
Migrant 1934 54.7 5757 72.1
Type of residence
Urban 1150 32.1 2751 33.6
Rural 2428 67.9 5444 66.4
Age
15 - 19 829 23.2 1820 22.2
20 - 29 1188 33.2 3110 37.9
30+ 1561 43.6 3265 39.8
Marital status
Not currently married 1723 48.2 3319 40.5
Currently married 1855 51.8 4876 59.5
Education
Less than High School 2219 62.0 5639 68.8
High School or higher 1359 38.0 2556 31.2
Ethnicity
Other ethnic groups 2733 76.4 6218 75.9
Kikuyu 845 23.6 1977 24.1
Religion
Other religious affiliation 3196 89.3 7160 87.4
Muslim 381 10.6 1025 12.5
Age at first sex
Before age 15 833 23.4 1229 15.7
At age 15 or later 2726 76.6 6598 84.3
Knows someone who has or died of AIDS
Yes 2652 75.1 5902 73.6
No 881 24.9 2118 26.4
Total 3578 100.0 8195 100.0
Note: Total numbers of cases may not add up to the sample size for some variables, due to missing data. Likewise, total percent may not add up to
100 due to rounding.

S. R. KIMUNA, Y. K. DJAMBA
themselves to be at risk of getting HIV (65% of men and 63% of
women). Yet, among those who had sex outside of marriage, more
than half did not use condoms at their last non-marital intercourse
(53% of men and 75% of women). On the other hand, more men
(11%) than women (2%) reported having had sexual intercourse
with multiple partners within the 12 months prior to the survey.
About 54 percent of men and 70 percent of women were mi-
grants. The majority of respondents lived in rural areas (68% of
men and 66% of women). At least 40 percent of men and
women in the sample were 30 years or older. About 52 percent
of men and 60 percent of women in the sample were married at
the time of the survey. Educational levels were very low.
Thirty-eight percent of male respondents had a secondary edu-
cation, compared to 31 percent among females. Nearly one
fourth of men and women were of Kikuyu ethnic group; the rest
were from other ethnic groups. In terms of religious affiliations,
little more than one-in-ten respondents were Muslim.
Most respondents did not have their first sexual intercourse
before age 15. Only about 23 percent of men and 16 percent of
women reported having had their first sexual intercourse before
age 15. The rest had their first sexual intercourse either at 15 or
later, including those who were still virgins. About three-
fourths of respondents knew someone who had HIV or died of
AIDS.
Main Effect of Mig ration
The data in Table 3 show means and percentages of migrants
and non-migrants across a series of variables. Figures in paren-
theses are number of cases. Of the three dependent variables
measuring risky sexual behaviour, only perceived chance of
being infected by HIV was significantly associated with migra-
tion across sex and place of residence. Compared to non-mi-
grants, migrants were significantly more likely to report that they
feared the risk of HIV infection, regardless of sex and place of
residence. Hopefully this perception translates into behavioural
change. Having had multiple sex partners was only marginally
significant (p < .10) for men living in the rural areas.
Table 3.
Mean and percent of sexual behaviour variables and other characteristics by place of residence and migration status, KDHS 20031.
Women
Urban Rural
Characteristics Non-migrants Migrants Non-migrants Migrants
Perceives chance of getting AIDS 57.4 (268) 63.8 (1387)** 52.0 (885) 67.1 (2337)***
Used condom at last non-marital sex2 36.9 (31) 31.7 (104) 17.6 (46) 19.9 (46)
Had multiple sex partners in last 12 months 1.9 (9) 1.6 (36) 1.1 (20) 1.5 (54)
Previously lived in urban area - 51.6 (1111) - 22.8 (797)
Previously lived in rural area - 48.4 (1044) - 77.2 (2705)
Mean age (years) 26.4 (480) 28.0 (2195)*** 24.3 (1750) 30.3 (3562)***
Currently married 36.9 (177) 55.9 (1228)*** 34.1 (596) 77.6 (2765) ***
Attained at least High School 51.7 (248) 47.8 (1049) 21.1 (370) 22.9 (816)
Kikuyu ethnic group 23.5 (113) 27.8 (610)+ 19.3 (338) 24.1 (857) ***
Muslim 33.0 (158) 12.2 (267)*** 15.9 (278) 8.5 (304)***
Had sex before age 15 11.8 (54) 12.5 (263) 15.1 (254) 18.5 (626)**
Knows someone who has or died of AIDS 69.3 (323) 75.4 (1636)** 67.2 (1138) 76.0 (2651)***
Number of cases 480 2195 1750 3562
Men
Urban Rural
Characteristics Non-migrants Migrants Non-migrants Migrants
Perceives chance of getting AIDS 58.1 (133) 67.6 (608)** 61.6 (836) 68.0 (687)***
Used condom at last non-marital sex2 69.2 (45) 58.1 (125) 41.2 (113) 37.9 (78)
Had multiple sex partners in last 12 months 11.7 (27) 15.1 (136) 8.9 (122) 11.0 (113)+
Previously lived in urban area - 56.9 (512) - 40.4 (412)
Previously lived in rural area - 43.1 (388) - 59.6 (609)
Mean age (years) 27.2 (231) 30.4 (906)*** 27.1 (1372) 31.9 (1028)***
Currently married 39.4 (91) 57.7 (523)*** 43.6 (598) 60.6 (623)***
Attained at least High School 55.0 (127) 57.2 (518) 25.9 (356) 33.5 (344)***
Kikuyu ethnic group 22.9 (53) 23.7 (215) 21.0 (288) 27.4 (282) ***
Muslim 27.8 (64) 11.8 (107)*** 7.4 (101) 10.0 (103)*
Had sex before age 15 18.4 (42) 22.3 (201) 23.6 (323) 24.9 (255)
Knows someone who has or died of AIDS 74.4 (169) 82.3 (742)** 70.9 (961) 74.3 (749)+
Number of cases 231 906 1372 1028
Notes: 1Figures in parentheses are number of cases in the cell. 2Among respondents whose last sexual activity was with non-marital partners. Level of
significance: ***p < .001; **p < .01; *p < .05; +p < .10.

S. R. KIMUNA, Y. K. DJAMBA
The difference in the mean age of migrants and non-migrants
was remarkable and statistically significant for both men and
women in urban and rural areas. That is, on average migrants
were older than non-migrants. In addition, migrants were sig-
nificantly more likely to be married than were non-migrants. In
urban areas, almost 56 percent of migrant women were married
compared to 37 percent of non-migrant women. In the rural
areas, the difference was even greater; almost 78 percent of
migrant women were married compared to 34 percent for
non-migrants. Similar to women, there were also greater dif-
ferences among men; almost 58 percent of urban migrant men
were married compared to 39 percent of urban non-migrant
men. In the rural areas, almost 61 percent of migrant men were
married compared to 44 percent of non-migrant men.
The association between migration and education was only
significant among men in rural areas, where migrants had
higher education than non-migrants. On the other hand, re-
spondents of the Kikuyu ethnic group were significantly more
likely to be migrants, except for Kikuyu men in urban areas. As
for religion, there were striking differences by sex. Among
women, migrants were significantly less likely to be Muslim as
compared to non-migrants, regardless of place of residence.
Among men, only those migrants living in urban areas were
significantly less likely to be Muslims as compared to
non-migrants in the same areas. In contrast, migrant men in the
rural areas were significantly more likely to be Muslims than
their non-migrant counterparts.
Migrants were more likely to have had sex before age 15
than non-migrants. This difference in having had sex before age
15 was observed across gender, but it was only statistically
significant for women in rural areas. In rural areas, female mi-
grants were significantly more likely to start sexual intercourse
before age 15 (18.5%) compared to non-migrants (15.1%).
On the whole, in both urban and rural areas, male and female
migrants were significantly more likely to know someone suf-
fering from HIV or died of AIDS. Among women living in
urban areas at the time of the survey, 75 percent knew someone
who had HIV or died of AIDS compared to 69 percent of
non-migrant women; the same pattern was observed in rural
areas with 76 percent for migrants versus 67 percent for non-
migrants. This difference was also observed among men. Urban
migrant men were more likely to know someone suffering from
HIV or died of AIDS (82.3%) than were non-migrant men
(74.4%). The difference between rural migrants and non-mi-
grants was 3.4 percent.
Multivariate Analyses
We examined how our independent variables and correlates
of risky sexual behaviour were associated with each of the three
dependent variables. The analyses were conducted using the
logistic regression equations in which we predicted the likeli-
hood of occurrence of the event (for each of the dependent
variables) based on a set of predictors, separately for men and
women and by their current place of residence. Data in Tables
4-6 are adjusted odds ratios (and 95% confident intervals given
in parentheses). The adjusted odds ratios present the net effect
of each variable on the dependent variable, controlling for the
effects of other variables.
Respondent’s Perceived Risk of HIV/AIDS Infection
Data in Table 4 show the association between migration
status (and last migration stream) and socio-demographic char-
acteristics on the perceived risk of being infected with
HIV/AIDS. The reported figures are odds ratios (OR), with
their confidence intervals (CI) given in parentheses at different
probability levels (P). An OR greater than 1 means greater risk,
whereas an OR of less than 1 means lower risk.
Compared to non-migrant women, migrant women in both
rural and urban settings perceived themselves to be at greater
risk of being infected; but the association was only statistically
significant in rural areas. Compared to non-migrant women in
rural areas, urban-rural and rural-rural migrant women were
significantly more likely to perceive themselves at higher risk
of contracting HIV (OR = 1.19 at p ≤ .10 and OR = 1.18 at p
≤ .05, respectively). There was no significant effect of migra-
tion on perceived HIV/AIDS risk among men.
In contrast, having sexual intercourse with multiple partners
in the last 12 months was associated with perceived risk of
HIV/AIDS infection for both men and women. Nonetheless, the
sex difference was greater in urban areas (OR = .04 at p ≤ .01
and OR = .49 at p ≤ .001 for women and men respectively) than
rural areas (OR = .48 at p ≤ .05 and OR = .40 at p ≤ .001 for
women and men respectively).
All other variables in the model were significantly asso-
ciated with women’s perceived risk of HIV/AIDS infection
regardless of their current place of residence. Women data in
Table 4 show that the perceived risk of HIV/AIDS infection
increases with the woman’s age. The odds ratios of such risk
among women in urban areas were .46 (at p ≤ .001) for those
age 15 - 19 and .75 (at p ≤ .01) for those age 20 - 29, as
compared to women 30 years and older. A similar pattern
was observed among women living rural areas (OR = .54 at
p ≤ .001 and OR = .89 at p ≤ .10 for women ages 15 - 19 and
20 - 29 respectively). In contrast, the impact of age was
statistically significant only among young urban resident
men (OR = .65 at p ≤ .01) as compared to men 30 years and
older.
Another interesting finding from this study is that unmarried
women perceived themselves at lower risk of HIV/AIDS as
compared to married women. Data in Table 4 shows that in
both urban and rural areas, unmarried women were signifi-
cantly less likely to consider themselves at risk of being in-
fected with HIV/AIDS. Compared to their married counterparts,
unmarried women’s odds of perceived risk of HIV/AIDS were
significantly lower: OR = .75 (p ≤ .001) and OR = .59 (p ≤ .001)
respectively in urban and rural areas. Marriage did not have a
significant impact on men’s perceived risk of HIV/AIDS. These
results indicate that marriage is not a protective status against
HIV/AIDS risk for women in Kenya.
The results on the impact of education were also statistically
significant. Compared to those with high school and higher,
women with less than high school education perceived them-
selves at lower risk of HIV/AIDS infection. However, the im-
pact was more substantial in urban areas (OR = .69 at p ≤ .001)
than rural places (OR = .86 at p ≤ .10). Among men, the impact
of education was also positively associated with perceived risk
of HIV/AIDS infection. Hence, compared to men with high
school education and higher, those with less than high school
education considered themselves at lower risk of infection, but
more so in urban areas (OR = .57 at p ≤ .001) than in rural areas
(OR = .79 at p ≤ .05).
The influences of ethnicity were all significant at p ≤ .001 for
women. Compared to women in the Kikuyu ethnic group,
172

S. R. KIMUNA, Y. K. DJAMBA
Table 4.
Adjusted odds ratios of perceiving a risk of getting HIV/AIDS by place of residence and by sex, KDHS 2003.
Models
Urban Rural
Variables
Women Men Women Men
Non-migrant 1.00 1.00 1.00 1.00
Migrant
Urban-urban 1.03 (.81 - 1.31) 1.28 (.90 - 1.84) - -
Rural-urban 1.08 (.84 - 1.39) .97 (.67 - 1.42) - -
Urban-rural - - 1.19 (.98 - 1.45)+ 1.25 (.96 - 1.63)
Rural-rural - - 1.18 (1.01 - 1.37)* 1.11 (.89 - 1.39)
Had multiple sex partners in last 12 months
No .04 (.01 - .30)** .49 (.33 - .75)*** .48 (.26 - .90)* .40 (.28 - .56)***
Yes 1.00 1.00 1.00 1.00
Age
15 - 19 .46 (.36 - .60)*** .88 (.53 - 1.45) .54 (.45 - .65)*** .65 (.47 - .90)**
20 - 29 .75 (.61 - .91)** 1.24 (.88 - 1.76) .89 (.77 - 1.02)+ 1.20 (.93 - 1.55)
30+ 1.00 1.00 1.00 1.00
Marital status
Not currently married .72 (.60 - .87)*** 1.00 (.69 - 1.43) .59 (.50 - .69)*** 1.03 (.79 - 1.35)
Currently married 1.00 1.00 1.00 1.00
Education
Less than High School .69 (.57 - .83)*** .57 (.43 - .74)*** .86 (.74 - 1.00)+ .79 (.64 - .97)*
High School or higher 1.00 1.00 1.00 1.00
Ethnicity
Other ethnic groups 1.49 (1.22 - 1.81)*** .95 (.69 - 1.32) 1.29 (1.11 - 1.50)*** .38 (.30 - .49)***
Kikuyu 1.00 1.00 1.00 1.00
Religion
Other religious affiliation 2.42 (1.86 - 3.14)*** 3.09 (2.09 - 4.56)*** 3.06 (2.5 - 3.8)*** 3.49 (2.45 - 4.96)***
Muslim 1.00 1.00 1.00 1.00
Age at first sexual intercourse
At age 15 or later .67 (.51 - .89)** .63 (.45 - .88)** .76 (.64 - .90)*** 1.00 (.81 - 1.24)
Before age 15 1.00 1.00 1.00 1.00
Knows someone who has or died of AIDS
No .83 (.67 - 1.01)+ .73 (.52 - 1.02)+ .82 (.71 - .95)** .75 (.61 - .91)**
Yes 1.00 1.00 1.00 1.00
Model X2 219.09*** 117.86*** 438.45*** 261.54***
Number of cases 2467 1110 4879 2345
***p < .001; **p < .01; *p < .05; +p < .10.
women in other ethnic groups were more likely to perceive
themselves at risk of HIV/AIDS infection (OR = 1.49 and 1.29
in urban and rural areas, respectively). In contrast, being a Ki-
kuyu was associated with lower perception of risk of
HIV/AIDS for men; but the effect was only statistically sig-
nificant in rural areas (OR for non-Kikuyu men as compared to
Kikuyu men was .38 at p ≤ .001).
Religion had a similar effect on the perceived risk of
HIV/AIDS for both men and women, regardless of their place
of residence. Compared to Muslim, men and women of other
religious faiths and non-believers were about three times more
likely to perceive themselves at risk of HIV/AIDS infection (at
p ≤ .001).
The timing of first intercourse is significant factor of per-
ceived risk of HIV/AIDS, especially among women. Compared
to their counterparts who had their first sexual intercourse be-
fore age 15, Kenyan women who had their sexual debut at age
15 or later were significantly less likely to perceive themselves
at risk of HIV/AIDS infection (OR = .67 at p ≤ .01 and OR
= .76 at p ≤ .001, respectively in urban and rural areas). Among
men, only those in urban areas had significant odds of perceiv-
ing themselves at low risk of HIV/AIDS, if they had first sexual
experience at age 15 or later (OR = .63 at p ≤ .01).
For both men and women, not knowing someone who has
HIV or died of AIDS was associated with lower perception of
HIV/AIDS risk. However, such risk was lower for men than for
women both in urban areas (OR = .73 and .83 respectively for
men and women both at p ≤ .10) and rural setting (OR = 75
and .82 respectively for men and women both at p ≤ .01).
Correlates of Condom Use at Last Non-marital
Sexual Intercourse
Results presented in Table 5 focused on condom use at last

S. R. KIMUNA, Y. K. DJAMBA
Table 5.
Adjusted odds ratios of using condom at last non-marital sexual intercourse by place of residence and by sex, KDHS 2003.
Models
Urban Rural
Variables Women Men Women Men
Non-migrant 1.00 1.00 1.00 1.00
Migrant
Urban-urban 1.15 (0.63 - 2.07) 0.58 (0.29 - 1.17)
Rural-urban 0.70 (0.38 - 1.27) 0.58 (0.29 - 1.19)
Urban-rural - - 1.34 (0.70 - 2.59) 1.02 (0.58 - 1.81)
Rural-rural - - 1.30 (0.71 - 2.35) 0.97 (0.61 - 1.53)
Perceived risk of getting HIV
No 1.03 (0.60 - 1.77) 2.09 (1.06 - 4.12)* .54 (.32 - .92)* .56 (.35 - .91)*
Yes 1.00 1.00 1.00 1.00
Age
15 - 19 .77 (.38 - 1.56) .63 (.24 - 1.66) 1.73 (.85 - 3.56) 2.50 (1.22 - 5.12)**
20 - 29 1.09 (.64 - 1.84) .70 (.30 - 1.62) 1.91 (.96 - 3.82)+ 3.55 (1.85 - 6.83)***
30+ 1.00 1.00 1.00 1.00
Marital status
Not currently married .68 (.14 - 3.24) 1.38 (.43 - 4.38) 4.75 (.60 - 37.47) 1.29 (.49 - 3.40)
Currently married 1.00 1.00 1.00 1.00
Education
Less than High School .59 (.36 - .96)* .46 (.27 - .77)** .51 (.30 - .86)* .72 (.47 - 1.10)
High School or higher 1.00 1.00 1.00 1.00
Ethnicity
Other ethnic groups .92 (.56 - 1.51) 1.15 (.62 - 2.14) 1.11 (.63 - 1.94) 1.73 (1.07 - 2.82)*
Kikuyu 1.00 1.00 1.00 1.00
Religion
Other religious affiliation 1.20 (.49 - 2.98) .85 (.35 - 2.07) 2.28 (.27 - 19.17) 4.98 (1.05 - 23.57)*
Muslim 1.00 1.00 1.00 1.00
Age at first sexual intercourse
At age 15 or later 1.10 (.58 - 2.11) 2.05 (1.16 - 3.62)** .68 (.38 - 1.22) 1.10 (.73 - 1.65)
Before age 15 1.00 1.00 1.00 1.00
Knows someone who has or died of AIDS
No .54 (.30 - .97)+ .92 (.43 - 1.95) .42 (.20 - .87)* .63 (.39 - 1.02)+
Yes 1.00 1.00 1.00 1.00
Model X2 21.627*** 26.493** 31.104*** 39.190***
Number of cases 390 280 468 477
***p < .001; **p < .01; *p < .05; +p < .10.
sexual intercourse for those whose last intercourse was with a
non-marital sex partner. These data show that there was no
significant difference per migration status. Similarly, we found
no significant effect of current marital status on condom use at
last non-marital sexual intercourse. In contrast, the remaining
variables have some significant association with the dependent
variable.
One surprising finding was that, for both men and women in
urban areas, those who perceived themselves at risk of HIV
infection were less likely to use a condom at their last
non-marital sexual intercourse than those who did not consider
themselves at such risk. However, the association was statisti-
cally significant only for men (OR = 2.09 at p ≤ .05). On the
other hand, perceived risk of getting HIV was positively and
significantly associated with the use of condoms at last
non-marital sexual encounter for both men and women in rural
areas. Compared to rural residents who perceived themselves at
risk of HIV/AIDS, those who did not consider themselves at
risk were significantly less likely to have used a condom at the
last non-marital sexual intercourse (OR = .56 and .54 respec-
tively men and women both at p ≤ .05).
Overall, education was associated with condom use. Those
who had at least a high school education were more likely to
have used a condom at their last non-marital sexual intercourse
than respondents with less than high school education. Among
women, the magnitude of impact was similar in both urban and
rural areas (OR = 59 and .51 respectively, and both at p ≤ .05).
For men, the association was statistically significant only in
urban areas (OR = .46 at p ≤ .01). In addition, respondent’s age
was significant only for rural residents. However, the relationship
was that of an inverted U-shape with those aged 20 - 29 more
likely to have used a condom at their last non-marital sexual in-
tercourse than younger and older respondents. However, the im-
pact of age on condom use was only statistically significant
among rural residents; among women those aged 20 - 29 were
about two times (OR = 1.91 at p ≤ .10) more likely to have used a
condom at their last non-marital sexual intercourse as compared
to women aged 30 and older. Among men in rural areas, all
174

S. R. KIMUNA, Y. K. DJAMBA
those below age 30 were significantly more likely to have used
a condom at their last non-marital sexual experience as com-
pared to the older men (OR = 2.50 at p ≤ .01 and OR = 3.55 at
p ≤ .001 for age-groups 15 - 19 and 20 - 29 respectively).
Ethnicity and religion were significantly associated with
condom use only for men in rural areas. Compared to men from
the Kikuyu ethnic group, men from other ethnic groups were
nearly two times (OR = 1.73 at p ≤ .05) more likely to have
used a condom at their last non-marital sexual intercourse. As
for religion, non-Muslims were about five times (OR = 4.98 at
p ≤ .05) more likely to have used a condom at their last
non-marital sexual intercourse as compared to Muslim men.
On the other hand, age at first sex had significant impact on
condom use only for men in urban areas. Urban men who de-
layed the onset of sexual activity were significantly more likely
to use a condom at their last non-marital sexual intercourse than
those who had sex before age 15 (OR = 2.05 at p ≤ .01).
Finally, knowing someone who had HIV or died of AIDS
affected condom use. Results in Table 5 show that regardless
of place of residence and sex, respondents who were acquainted
with people infected with HIV were more likely to use con-
doms at their last non-marital sexual intercourse than those who
had no knowledge of people with HIV. In urban areas, such
impact is only significant for women. Compared to those who
don’t know anybody who has HIV or died of AIDS, those who
do, had a significantly low odds (OR = .54 at p ≤ .10) of having
used a condom at their last non-marital sexual relation. In rural
areas, the effect was significant for both sexes, although the
magnitude of the impact was stronger for women (OR = .42 at
p ≤ .05) than men (OR = .63 at p ≤ .10).
Table 6.
Adjusted odds ratios of having had sexual intercourse with more than one partner in the last 12 months by place of residence and by sex,
KDHS 2003.
Models
Urban Rural
Variables Women Men Women Men
Non-migrant 1.00 1.00 1.00 1.00
Migrant
Urban-urban 1.73 (.72 - 4.14) 1.31 (.80 - 2.13) - -
Rural-urban 1.33 (.66 - 2.71) 1.08 (.64 - 1.84) - -
Urban-rural - - .76 (.40 - 1.46) 1.39 (.95 - 2.03)+
Rural-rural - - 1.51 (.80 - 2.85) 1.13 (.81 - 1.58)
Perceived risk of getting HIV
No 25.69 (3.50 - 188.46)*** 2.04 (1.34 - 3.10)* 2.12 (1.13 - 3.98)* 2.48 (1.76 - 3.49)***
Yes 1.00 1.00 1.00 1.00
Age
15 - 19 .95 (.37 - 2.43) .45 (.22 - .94)* .75 (.35 - 1.61) .55 (.33 - .91)*
20 - 29 1.13 (.55 - 2.29) 1.13 (.73 - 1.75) 1.01 (.59 - 1.75) 1.10 (.76 - 1.58)
30+ 1.00 1.00 1.00 1.00
Marital status
Not currently married 2.51 (1.28 - 4.93)** 1.40 (.90 - 2.17) 1.53 (.84 - 2.78) 1.35 (.93 - 1.97)
Currently married 1.00 1.00 1.00 1.00
Education
Less than High School 1.82 (.92 - 3.58)+ 1.54 (1.07 - 2.20)* 4.54 (1.61 - 12.79)** 1.60 (1.14 - 2.23)**
High School or higher 1.00 1.00 1.00 1.00
Ethnicity
Other ethnic groups 1.34 (.63 - 2.87) 1.44 (.93 - 2.23) 1.87 (.87 - 4.02) 3.20 (2.05 - 4.98)***
Kikuyu 1.00 1.00 1.00 1.00
Religion
Other religious affiliation 3.50 (.79 - 15.44)+ .86 (.50 - 1.49) 2.23 (.68 - 7.33) 1.09 (.60 - 1.98)
Muslim 1.00 1.00 1.00 1.00
Age at first sexual intercourse
At age 15 or later .65 (.29 - 1.43) .85 (.57 - 1.27) .42 (.43 - .70)*** .57 (.43 - .77)***
Before age 15 1.00 1.00 1.00 1.00
Knows someone who has or died of AIDS
No .67 (.29 - 1.56) .48 (.27 - .83)** .67 (.36 - 1.25) .89 (.64 - 1.23)
Yes 1.00 1.00 1.00 1.00
Model X2 48.877*** 44.738** 48.887*** 99.983***
Number of cases 2467 1110 4879 2345
***p < .001; **p < .01; *p < .05; +p < .10.

S. R. KIMUNA, Y. K. DJAMBA
Sexual Experience with Multiple Partners
The results in Table 6 show the likelihood that a respondent
would have had sex with multiple partners in the 12 months
before the survey. The independent variables included in these
analyses are the perceived risk of being infected with HIV,
socio-demographic characteristics, age at first sexual inter-
course, and knowledge of someone living with HIV or who
died of AIDS.
Table 6 results show that migration was a significant corre-
late of multiple sexual partners only among men living in rural
areas. Among rural resident men, only urban-rural migrants
were significantly more likely to have had sex with multiple
partners than non-migrants (OR = 1.39 at p ≤ .10). In contrast,
we found that respondents who did not consider themselves to
be at risk of being infected with HIV were significantly more
likely to have had sexual intercourse with multiple partners.
This pattern was consistent for men and women in both rural
and urban areas; but the most outstanding case was found
among urban resident women with an odds ratio value of 22.69
(p ≤ .001).
The age category showed teenagers to be somewhat different
from the rest of the respondents in terms of their risk of having
sex with multiple partners. Still, the effect was significant only
for men for whom the likelihood of having had sex with multi-
ple partners was significantly lower among teenagers compared
to men aged 30 years and older (OR = .45 and .55, both at p
≤ .10, respectively in urban and rural areas) . Also, we found
that marriage had only some protective effect among women in
urban areas. For urban women, those who were married had
lower risk of having had sex with multiple partners (OR = 2.51
at p ≤ .01).
Interestingly, for both men and women in rural and urban
areas, education was significantly associated with lower risk of
having had sex with more than one partner in the last 12
months. However, the magnitude of the effect was not the same
across models. The strongest impact was found among rural
women with less than high school education. For such women,
the likelihood of having had sex with multiple partners in the
12 months before the survey was nearly five times higher (OR
= 4.54 at p ≤ .01) compared to rural women with high school
education or higher. The lowest, yet significant effect was ob-
served among urban resident men with less than high school
education (OR = 1.54 at p ≤ .05), in comparison to those with
high school education or higher.
The odds ratios for ethnicity show that the Kikuyu were less
likely to have had sex with multiple partners, compared to other
ethnic groups. Nonetheless, this ethnic effect was statistically
significant only among men in rural areas (OR = 3.20 at p
≤ .001). Religion had only a marginally significant effect on the
number of sexual partners among urban. For these women,
being of Muslim religion greatly reduced the risk of sexual
activity with multiple partners women (OR = 3.50 at p ≤ .10).
As one would expect, having had first sexual intercourse at a
younger age was associated with having had sexual experience
with multiple partners even in this case where the timeframe
was one year. However, this association was only significant in
rural areas where both men (OR = .57 at p ≤ .001) and women
(OR = .42 at p ≤ .001).who initiated sex at younger ages were
more likely to have had more sexual partners in the last 12
months than those who delayed their debut of sexual inter-
course. Ironically, for men in urban areas, knowing at least one
person living with HIV or who died of AIDS significantly in-
creased the likelihood of having had sex with multiple partners.
This study set out to examine internal migration in Kenya
and its effect on risky sexual behaviour as measured by the
perceived risk of getting HIV, use of condom at last non-mari-
tal sexual intercourse, and sexual intercourse with multiple
partners. We argued that because migration exposes people to
different cultural norms and values, migrants would be more
likely than non-migrants to engage in risky sexual behaviour.
However, our findings show only a marginal impact of urban-
rural migration among rural residents. Overall, more respon-
dents considered themselves to be at risk of being infected with
HIV (little more than 60% of men and women in the sample).
Such risk was consistently and statistically significantly higher
among migrants than non-migrants, regardless of sex and place
of residence.
However, once socio-demographic characteristics and other
correlates of sexual behaviour were introduced in the regression
models, migration effect became statistically significant only
for migrant women in rural areas. Even for such women, only
those who moved from urban to rural areas were significantly
more likely to perceive themselves at risk of HIV infection,
compared to non-migrants.
This may suggest that migrant women in rural areas are rela-
tively more attractive in the local sexual network market hence
their heightened perceived risk of HIV infection. As noted in
Brokerhoff and Biddlecom (1999) study, there could be a
number of explanations for these findings. First, rural migrant
women may be moving away from kin, which may remove the
behavioural constraints imposed by family and second, they
may be moving to rural road sites and trading centres, where
the forces of migration may further aggravate gender related
inequalities by increasing women’s economic and physical
exposure to casual or transactional sex as a survival strategy
(Anarfi, 1993).
As it has been noted elsewhere (Kimuna & Djamba, 2005;
Hattori & Dodoo, 2007), we also found that marriage does not
reduce the risk of having HIV in Kenya. This is probably be-
cause married people have more resources and other qualities
that are attractive in mate selection process. Likewise, our
findings showed that education does not reduce the risk of be-
ing infected with HIV in this sample.
Despite the fact that most respondents reported fear of being
infected with HIV, the majority still did not use condoms at
their last non-marital sexual activity (53% for men and 75% for
women). Moreover, the results from multivariate analyses in-
dicate that there is no significant difference in condom use at
last non-marital sexual activity by migration status. This may
indicate either the presence of strong cultural norms against
condom use, or that there is no significant difference in the
knowledge and availability of condoms between urban and
rural areas.
The lack of migration effect on condom use in this analysis
may also suggest that contraceptive knowledge and practices
are well diffused across Kenya. It may also indicate a more
circular nature of migration in which those involved don’t stay
longer in place of destination. Another important finding is that
marital status does not affect the chance of using condoms in
non-marital relations and this was regardless of sex and place of
residence.
These results suggest that high risk sexual behaviour such as
non-condom use with multiple sexual partners is very complex
176

S. R. KIMUNA, Y. K. DJAMBA
and it needs to be approached from a socio-cultural standpoint.
Moore and Oppong’s (2007) study noted the futility of advo-
cating for people to use condoms, especially those in marriages.
Their research found that people in steady relationships such as
marriage did not believe in using condoms with their spouses,
even when they knew of their sero-positive HIV status. The fact
that condom use is not related to marital status even when mar-
ried people engage in extramarital sex is alarming. Prevention
programs should be multifaceted and take the environment in
which people live and local realities into consideration.
The migration effect was also non-significant in the models
of the likelihood of having multiple sexual partners, except
among men in rural areas. For the latter, we found that net of
other control variables urban-rural male migrants were signifi-
cantly more likely to have had sex with multiple partners than
non-migrants. With the exception of unmarried women in urban
areas who were significantly more likely to have had sex with
multiple partners, we did not find a significant effect of mar-
riage among men or women in rural areas. Again, this weak and
generally lack of strong effect of marriage on risky sexual be-
haviour shows that marriage per se is not a protective factor in
the combat against HIV/AIDS.
One encouraging finding is that those who consider them-
selves to be at risk of being infected with HIV tend to limit the
number of sexual partners. This result supports the popular
view that knowledge is power. Yet, this was not translated into
condom use at last non-marital sexual intercourse for urban
residents.
Education emerged as an important factor associated with
low sexual risk. Although more educated respondents tended to
see themselves at higher risk of being infected with HIV, they
were significantly more likely to use condoms at last non-
marital sexual activity and to report lower risk of having had
sexual intercourse with multiple partners. This finding supports
the idea that formal education is a strong empowerment vari-
able that leads to more responsible sexual behaviour. Likewise,
knowing someone who had HIV or died of AIDS seems to raise
awareness about HIV risk and to lead to better likelihood of
condom use in an extra-marital relationship.
We also found that delaying the onset of sexual intercourse
helps to lower risky sexual behaviour. Respondents who had
their first sexual intercourse at age 15 or later were significantly
less likely to see themselves as being at risk of HIV infection.
Such respondents were also more likely to use condoms at their
last non-marital sexual activity and were less likely to have had
sex with multiple sexual partners in the 12 months before the
survey. This result suggests that delayed sexual activity may
result in more knowledge and safer sexual practices. Other
cultural factors, such as religion and ethnicity also played some
role in risky sexual behaviour, although their effects were not
all consistent across sex and place of residence.
Overall, this study suggests that there is little or no signifi-
cant difference in sexual behaviours between migrants and
non-migrants. This is probably because migration in Kenya is
ongoing and quite complex. First, Kenya is a country of inter-
nal migration as compared to other nations in the region (Choi,
2003). As shown in this study, more than half of the respon-
dents in the sample were migrants. Second, Kenya is character-
ised by unique ethnic groups that have distinctive cultural
norms that have specific effects on sexual behaviours. As a
result, the findings from this study must be used with caution
when comparing to other regions of the world.
Nonetheless, the lack of strong effects of migration on the
sexual risk and sexual behaviour variables examined here sug-
gests that efforts to reduce risky sexual behaviour should target
all segments of the population, regardless of their migration
status and place of residence.
Acknowledgements
This paper was presented at the International Organization of
Social Sciences and Behavioural Research, the spring 2012
conference. Thanks also to Tawanda Blackmon and Erin Brown,
respectively research assistant and program assistant at the
Center for Demographic Research at Auburn University at
Montgomery, for helping with the preparation of the statistical
tables. The authors are grateful to the anonymous reviewers for
their comments and suggestions.
REFERENCES
Anarfi, J. K. (1993). Sexuality, migration and AIDS in Ghana: Asocio-
behavioural study. Health Transition Review, 3, 45-67.
Aral, S. O., & Peterman, T. A. (1996). Measuring outcomes of behav-
ioral interventions for STD/HIV prevention. International Journal of
STD and AIDS, 7, 30-38. doi:10.1258/0956462961917753
Barnett, T., & Whiteside, A. (2006). AIDS in the twenty-first Century:
Disease and globalization (2nd ed.). New York, NY: Palgrave Mac-
millan.
Bessinger, R., Akwara, P., & Halperin, D. (2003). Sexual behavior,
HIV and fertility trends: A comparative analysis of six countries.
USAID measure evaluation project. Chapel Hill, NC: University of
North Carolina.
Brockerhoff, M., & Biddlecom, A. E. (1999). Migration, sexual behav-
iour and the risk of HIV in Kenya. International Migration Review,
33, 833-856. doi:10.2307/2547354
Camlin, C. S., & Chimbwete, C. E. (2003). Does knowing someone
with AIDS affect condom use? An analysis from South Africa. Edu-
cation and Prevention , 15, 231-244.
doi:10.1521/aeap.15.4.231.23831
Central Bureau of Statistics (CBS) [Kenya], Ministry of Health (MOH)
[Kenya], ORC Macro (2004). Kenya demographic and health survey
2003, CBS, MOH, and ORC Macro, Calverton, Maryland.
Cleland, J., Boerma, J. T., Caraël, M., & Weir, S. S. (2004). Monitoring
sexual behaviour in general populations: A synthesis of lessons of
the past decade. Sexually Transmitted Infections, 80, ii1-ii7.
doi:10.1136/sti.2004.013151
Choi, S. Y. P. (2003) Report of the INDEPTH Migration and Urbani-
sation Workshop (21-24 January 2003). Gauteng: Centre for Health
Policy, Witswaterand.
Djamba, Y. K. (2003). Gender differences in motivations and intentions
to move: Ethiopia and South Africa compared. Genus, LIX, 93-111.
Fishbein, M., & Pequegnat, W. (2000). Evaluating AIDS prevention
interventions using behavioral and biological outcome measures.
Sexually Transmitted Diseases, 27, 101-110.
doi:10.1097/00007435-200002000-00008
Hattori, M. K., & Dodoo, F. (2007). Cohabitation, marriage, and “sex-
ual monogamy” in Nairobi’s slums. Social Science & Medicine, 64,
1067-1078. doi:10.1016/j.socscimed.2006.10.005
Kimuna, S. R., & Djamba, Y. K. (2005). Wealth and extra-marital sex
among men in Zambia. International Family Planning Perspectives,
31, 83-89.
Leclerc-Madlala, S. (2003). Transactional sex and the pursuit of mod-
ernity. Social Dynamics, 29, 213-233.
doi:10.1080/02533950308628681
Longfield, K., Glick, A., Waithaka, M., & Berman, J. (2004). Rela-
tionships between older men and younger women: Implications for
STIs/HIV in Kenya. Studies in Family Planning, 35, 125-134.
doi:10.1111/j.1728-4465.2004.00014.x
Lurie, M. N. (2006). The epidemiology of migration and HIV/AIDS in

S. R. KIMUNA, Y. K. DJAMBA
178
South Africa. Journal of Ethnic and Migration Studies, 32 , 649-666.
doi:10.1080/13691830600610056
Lurie, M., Williams, B., Zuma, K., Mkaya-Mwamburi, D., Garnett, G.,
Sturm, A. W., Sweat, M. D., Gittelsohn, J., & Karim, S. S. (2003).
The impact of migration on HIV transmission in South Africa: A
study of migrant and non-migrant men and their partners. Sexually
Transmitted Diseases, 30, 149-156.
doi:10.1097/00007435-200302000-00011
Mensch, B., Clark, W. H., Lloyd, C. B., & Erulkar, A. S. (2001). Pre-
marital sex, schoolgirl pregnancy, and school quality in rural Kenya.
Studies in Family Planning, 32, 285-301.
doi:10.1111/j.1728-4465.2001.00285.x
Moore, A.R., & Oppong, J. (2007). Sexual risk behaviouramong people
living with HIV/AIDS in Togo. Social Science & Medicine, 64,
1057-1066. doi:10.1016/j.socscimed.2006.10.004
Ntozi, J. P. M., & Ahimbisibwe, F. (1999). Some factors in the decline
of AIDS in Uganda. In J. C. Caldwell, J. P. M. Ntozi, & I. O. Oru-
buloye (Eds.), The continuing HIV/AIDS epidemic in Africa (pp. 93-
107). Canberra: Health Transition Centre, Australian National Uni-
versity.
Nyanzi, S., Nyanzi, B., Kalina, B., & Pool, R. (2004). Mobility, sexual
networks and exchange among bodaboda men in southwest Uganda.
Culture, Health & Sexuality, 6, 239-254.
doi:10.1080/13691050310001658208
Palekar, R., Pettifor, A., Behets, F., & MacPhail, C. (2008). Association
between knowing someone who died of AIDS and behaviour change
among South African youth. A IDS and Behaviour, 12, 903-912.
Pison, G., Le Guenno, B., Lagarde, E., Enel, C., & Seek, G. (1993).
Seasonal migration: Risk factor for HIV in rural Senegal. Journal of
Acquired Immune Deficiency Syndromes, 6, 196-200.
Smith, K. P., & Watkins, S. C. (2005). Perceptions of risk and strate-
gies of prevention: Responses to HIV/AIDS in rural Malawi. Social
Science & Medicine, 60, 649-660.
doi:10.1016/j.socscimed.2004.06.009
Topouzis, D. J., & Guerney, D. U. (1999). Sustainable agriculture:
Rural development and vulnerability to the AIDS epidemic, UN-
AIDS Best Practice Collection. Geneva: UNAIDS & FAO.
UNAIDS (2006). 2006 Report on the Global AIDS Epidemic. Geneva:
UNAIDS.
Wilson, D. (2000). Corridors of hope in Southern Africa: HIV preven-
tion needs and opportunities in four border towns. Arlington: Family
Health International.
Wilson, F. (1972). Migration labour in South Africa, Johannesburg,
South Africa.
Zulu, E. M., Dodoo, F. N., & Ezeh, A. C. (2002). Sexual risk taking in
the slums of Nairobi, Kenya, 1993-98. Population Studies, 56, 311-
323. doi:10.1080/00324720215933