DHS WORKING PAPERS

Fertility Transition and Its Determinants

in Kenya: 2003-2008/9

Andrew Kyalo Mutuku

2013 No. 82

DEMOGRAPHIC
AND
February 2013
HEALTH
This document was produced for review by the United States Agency for
International Development. SURVEYS
 Fertility Transition and Its Determinants in Kenya: 2003-2008/9

Andrew Kyalo Mutuku

ICF International
Calverton, Maryland, USA

February 2013

Corresponding author: Andrew Kyalo Mutuku, Population Studies and Research Institute, University of Nairobi,
Nairobi. Kenya; Email: [email protected]
ACKNOWLEDGEMENTS

The author is grateful to the staff at ICF Macro and particularly to Dr. Thomas Pullum
and Wenjuan Wang for their timely suggestions and feedback. These were very instrumental in
the writing and finalization of this paper. The study was supported through a fellowship from the
United States Agency for International Development (USAID) through the DHS Fellows
Program at ICF International. Thank you to Bryant Robey for editing and to Yuan Cheng for
formatting the document.

The DHS Working Papers series is an unreviewed prepublication series of papers reporting on research in
progress that is based on Demographic and Health Surveys (DHS) data. This research is carried out with
support provided by the United States Agency for International Development (USAID) through the
MEASURE DHS project (#GPO-C-00-08-00008-00). The views expressed are those of the authors and
do not necessarily reflect the views of USAID or the United States Government.

MEASURE DHS assists countries worldwide in the collection and use of data to monitor and evaluate
population, health, and nutrition programs. Additional information about the MEASURE DHS project can
be obtained by contacting MEASURE DHS, ICF International, 11785 Beltsville Drive, Suite 300,
Calverton, MD 20705 (telephone: 301-572-0200; fax: 301-572-0999; e-mail: [email protected];
internet: www.measuredhs.com).
ABSTRACT

Kenya began to experience rapid fertility decline in the late 1980s. However, this decline
in fertility stalled in the late 1990s. This study examined factors influencing transition to third
and fourth births in Kenya for the period between 2003 and 2008/9, in order to provide a better
understanding of the current fertility transition in Kenya with respect to birth spacing. Data was
obtained from the Kenya Demographic and Health Surveys (KDHS) of 2003 and 2008/9.
Logistic regression model was the main method of data analysis. The dependent variable was
whether or not a woman had made a transition from second to third birth and from third to fourth
birth in the five years preceding each survey. The main independent variables were: education
level, place of residence (urban-rural), region of the country, wealth index, marital status,
religion, ever-use of contraception, child mortality, and mother’s age at the start of the birth
interval.

The results showed that for the five-year period before KDHS 2008/9 the odds of
transition to a third birth were lower for women with some education compared with women
with no education. Odds were lower for women in Central, Eastern, and North Eastern provinces
compared with women in Western province, but higher for women of the Protestant religion
compared with Catholic women. The odds of transition to a third birth were lower for women
from middle, richer, and richest households compared with women from poorest households, and
lower for women who had ever used a modern method of contraception compared with women
who had never used contraception.

For the five year-period before KDHS 2008/9, the odds of transition from third to a
fourth birth were significantly influenced by education level, wealth index, age at the start of the
interval, ever-use of contraception, and desire for an additional child. The results for the five
year-period before KDHS 2003 showed that transition to fourth birth was significantly associated
with region, wealth index, age at the start of the interval, ever-use of contraception, and desire
for another child.

This paper has two main policy implications. First, there is a need to improve family
planning services across the country, which would address the issues of unwanted fertility and
unmet need for family planning and hence lead to a decline in fertility. Second, there is also a
need to improve the socioeconomic conditions in the country. Higher levels of wealth and
education are strongly associated with lower fertility.
INTRODUCTION AND BACKGROUND

Prior to the 1970s, fertility in Kenya had been rising steadily. But since the mid-1980s,
Kenya has experienced perhaps one of the most remarkable fertility transitions in history,
contrary to earlier expert predictions (Blacker 2002, Brass and Jolly 1993, Population Council
1998, Robinson 1992). The total fertility rate (TFR) declined from 8.1 children per woman in
1977/78 to 6.7 in 1989 and 5.4 in 1993. The TFR further declined to 4.7 children per woman in
1998.

Results from the 2003 Kenya Demographic and Health Survey (KDHS), however,
showed that infant and child mortality indicators continued to worsen in the 1990s, while
women’s preferences for lower fertility stalled, and contraceptive prevalence also stagnated
(CBS et al. 2004). In 2008/9, the TFR was 4.6 children per woman, nearly the same as in 1998.

Three perspectives have emerged as dominant explanations for the observed fertility
declines in Kenya. These were changes in proximate determinants of fertility and specifically
increases in contraceptive prevalence (Blacker 2002, Brass and Jolly 1993, Cross et al. 1991,
Ekisa and Hinde 2005, Ezeh and Dodo 2001, Njogu and Martín 1991, Population Council 1998,
Sibanda 1999), changes in attitudes regarding large family size due to high costs associated with
raising many children (Robinson 1992), and changes in cultural norms supporting high fertility
due to modernization (Watkins 2000).

A study by Westoff and Cross (2006) attributed the reversal of fertility decline in Kenya
to a stall in contraceptive prevalence and also to an increase in the number of women desiring
more children. The desire for more children was seen as a response to increased child mortality
arising from the HIV/AIDS pandemic. This finding was corroborated in a study by Magadi and
Agwanda (2010), who found that HIV/AIDS could have been responsible for the increased
desire for more children, as women sought to replace children who had died.

This study examined dynamics of fertility transition in Kenya between 2003 and 2008/9
from the perspective of birth intervals. A parity-specific approach to studying fertility is useful
for analyzing the family-building process. This is because when people think about having
children, they think in terms of whether or not and when to have a first or a subsequent birth.
This approach indicates changes in fertility due to changes in timing of reproduction, spacing of

1
births, or in the changing reproductive patterns at low or high parities. Few studies have
examined the dynamics of the Kenyan fertility transition using the birth-interval approach.

The study contributes knowledge on fertility transition in Kenya by identifying factors

influencing transition from second to third birth and from third to fourth birth in Kenya between
2003 and 2008/9. These transitions were selected because they may be indicative of change in
family size preferences in Kenya. The two time periods are also significant in Kenya’s fertility
transition. The five-year period before the 2003 KDHS witnessed a period of fertility stall, while
the five-year period before the 2008/9 KDHS witnessed a slight decline in fertility.

Knowledge gained in study will also be useful for policymakers, programme managers,
and researchers, and for developing appropriate programmes to enhance socioeconomic
development in the country. Information will help enable the government to manage population
growth better and to match population growth with development goals, including family
planning and maternal and child health programmes.

This study sought to answer the following questions:

i) What role have socioeconomic, cultural, and demographic factors played in

influencing transition from second to third births and from third to fourth births?

ii) Has child mortality played any role in these transitions?

2
LITERATURE REVIEW

The literature shows strong associations between fertility declines and socioeconomic
factors. Research has shown a close association between women’s education and fertility
(Bledsoe et al. 1999). Education influences fertility through enabling women to acquire new
ideas and values at variance with their traditional roles of childbearing. It also opens new
opportunities for women to participate in the labour force and wage employment (Caldwell and
Caldwell 1987) and to play roles other than those associated with reproduction and child care
(United Nations 1995). Dixon-Mueller (1993) also has argued that women’s education influences
attitudes and knowledge and hence contraceptive behavior.

Studies have also shown that fertility for women in urban areas is lower compared with
women in rural areas (Cohen 1993, Ekisa and Hinde 2005, Westoff 1994, Woldemicael 2005).
Women in urban areas not only are more educated than rural women, on average, and have more
access to job opportunities and better health care but are also better able to embrace new values
and ideas regarding childbearing and rearing (Diamond et al. 1999, Montgomery and Lloyd
1999, Oheneba-Sakyi and Takyi 1997). It has been observed that high costs associated with
urban living are not supportive of large family sizes. Mlewa (2001) established that men in urban
areas were more likely to desire smaller families compared with their rural counterparts.

Region of residence has also been observed to exert a strong influence on the number of
children desired, reflecting differences in language, ethnic origin, and religion as well as
economic development (Westoff 1994). In a study on fertility preferences in Mathare Valley,
Nairobi, Alila (1990) found that people whose places of origin were areas of high fertility
maintained their high fertility even after relocating to urban areas. In a study of fertility transition
in Eritrea, Woldemicael (2005) noted that fertility decline was already underway in Eritrea and
had occurred in urban and rural areas and in every region of the country. Ekisa and Hinde (2005)
noted that in Kenya there were substantial and persistent regional differences in fertility. Fertility
was generally lower in Central province and higher in Coastal and Western areas.

Researchers have also established a strong association between fertility and women’s
work status (Blacker 2002, Hirschman and Young 1998, Wasao 2002, Woldemicael 2005, Zhang
1994). Women who participate in modern wage employment are more likely to use contraception
and hence have lower fertility compared with women who do not work for wages (Shapiro and

3
Tambashe 1997, United Nations 1991). In a study of fertility decline in the Southeast Asian
counties of Thailand, Malaysia, Indonesia, and the Philippines, Hirschman and Young (1998)
attributed the observed decline in fertility to a change in the status of women. The authors noted
that women’s status was enhanced through more participation in wage employment, providing
opportunities other than childbearing. Another study by Rodriguez and Cleland (1981) found that
women’s work status had a large and significant impact on fertility. While employment in the
modern sector may conflict with childbearing and encourage low fertility, work on the farm may
be more compatible with high fertility (Abdalla 1988, Moustafa 1988, United Nations 1995).

Blacker (2002) attributed the substantial differentials in fertility observed in Kenya by

place of residence, region of residence, and educational levels to differences in socioeconomic
development. He observed that Central province, with better development indicators such as
under-five mortality, nutrition, education, and housing amenities, experienced substantial fertility
declines between 1989 and 1998. He also attributed fertility decline in the 1990s to worsening
socioeconomic conditions reflected in indicators such as infant and child mortality, school
enrolment ratios, or GDP per capita. He noted that ideal family size, which is an indicator of
changing attitudes toward fertility, declined between 1984 and 1998. He was of the view that
fertility in Kenya was not going to fall below an average of three births per woman because of
the stabilization of ideal family sizes, the stall in contraceptive prevalence, the trend of fertility
observed in Central region, and also the fertility rate for women with at least a secondary level of
education (Blacker 2002).

Other studies have linked rapid fertility declines experienced in Kenya to increased use of
modern contraception (Blacker et al. 2005, Brass and Jolly 1993, Cross et al. 1991, Ekisa and
Hinde 2005, Robinson 1992, Wasao 2002). In a comparative study of fertility transition in Kenya
and Uganda, Blacker and colleagues (2005) found that over a period of 20 years, fertility
declined more rapidly in Kenya than in Uganda. The rapid decline of fertility observed in Kenya
was attributed to an increase in contraceptive prevalence, while the slower decline of fertility
observed in Uganda was attributed to a decrease in pathological sterility. The study also found
that women in Kenya desired fewer children than women in Uganda and that women in Uganda
also had higher levels of unmet need for family planning. These differences perhaps could be
explained by the aggressive family planning programme in Kenya and the differences in
socioeconomic development between the two countries.

4
 Socio-cultural factors have also played an important role in fertility decline in many
communities (Ahehu 1998, Njogu and Martín 1991, Wasao, 2002). Njogu and Martín (1991)
noted that fertility declined among the Christians and Muslims in Kenya, but increased among
traditional religions. Ayehu (1998) found that a woman with no religious affiliation was about
1.5 times more likely to desire more children compared with a woman of the Catholic faith.
Religion shapes women’s values, norms, and beliefs concerning reproduction and thus affects
their fertility behaviour (Benefo 1995). In some societies, religious schooling provides messages
and teaching about traditional values that are inconsistent with practicing family planning.
Traditional worshippers often preach prolific childbearing as unequivocal evidence of spiritual
and ancestral benevolence. Islam endorses pronatalistic practices, while Christians are more
flexible with regard to reproductive choices, including the use of contraception. Thus, Muslims
and traditionalists exhibit higher fertility compared with Christians (Adegbola 1988).

Polygamous marriages in Africa have generally been associated with high fertility.
Women in areas of widespread practice of polygyny tend to start having sexual intercourse at an
early age and to have high fertility goals (Ezeh 1997). This has often led to short birth intervals.
Women in polygamous unions also tend to have lower socioeconomic status than those in
monogamous unions and are likely to be less educated (Muinde and Mukras 1979). In such
forms of social organization, it is claimed that children are the only tool a woman can use in
laying claim to spousal property and inheritance (Cohen 1993).

Other studies have linked fertility decline to a rise in age at first marriage and first birth
(Blacker 2002, Ekisa and Hinde 2005, Vavrus 2000, Woldemicael 2005, Zhang 1994). A study
by Woldemicael (2005) in Eritrea found that fertility decline was evident across all reproductive
ages and birth orders but was stronger among older mothers and for higher-order births. This
decline in fertility was attributed to prolonged spacing of births, cessation of further
childbearing, and delayed age at marriage. Blacker (2002) attributed fertility decline in Kenya to
increased age at first marriage. Obiero (1999) observed that in fertility decline in the 1980s in
Kenya, women over age 34 contributed to the greatest decline in fertility, even though the
decline occurred across women of all ages. In the early 1990s, however, the rapid decline of
fertility was due to rapid decline in fertility among younger women and a stall in fertility among
women age 35 and above (Obiero 1999).

5
 The crucial role played by mortality decline in the fertility transition is well documented
(Gyimah 2002, Hirschman and Young 1998, Mason 1997). Recent theories explaining fertility
transition need to acknowledge that even classical demographic transition theory held that
mortality decline was a prerequisite for fertility decline. Researchers have attributed the failure
by many countries in sub-Saharan Africa to achieve real fertility transition to the inherently high
levels of child mortality in the region. Gyimah (2002) found that women with prior infant deaths
had more subsequent births than those without experience of infant mortality, suggesting both a
physiological and behavioural response. Hirschman and Young (1998) observed that when
mortality is low the costs of raising many children may lead parents to have fewer children. A
related study by Lloyd and Ivanov (1988) observed that child survival, family planning, and
fertility could be linked through a number of steps in a mortality transition. The study noted that
the stages in this linkage would involve family formation by fate, and family formation by
design, insurance, and replacement. It argued that the evolution of family formation strategies
and mortality declines over time depend on the prevailing socio-cultural environment.

Causes of Stalling Fertility Transitions

According to the classical demographic transition theory, once fertility decline is

underway, it continues until replacement level fertility is attained. This pattern has been
interrupted in a number of developing countries, where fertility seemed to have stalled at levels
above replacement in the late 1990s and early 2000s. Researchers have offered possible
explanations why fertility may stall above replacement level. However, it is not yet clear why
fertility varies or remains constant over the course of time.

Stalling fertility declines have been attributed to slowing of socioeconomic development,

high levels of infant and child mortality, short birth intervals, declines in contraceptive
prevalence, and the impact of HIV/AIDS (Bongaarts 2006, Garenne 2007, Shapiro and
Gebreselassie 2008, Westoff and Cross 2006). Using DHS data from 24 countries with multiple
surveys, Shapiro and Gebreselassie (2008) examined the current status of fertility transition in
sub-Saharan Africa and the extent to which fertility had stalled. They attributed stalling fertility
in some of the countries to faltering in the pace of socioeconomic development, as reflected in
the lagged infant and child mortality rates and lagged growth in GDP per capita. However, in

6
some of the countries slower declines in fertility were associated with higher growth in GDP per
capita, which was contrary to expectations. According to the findings, modern contraceptive use
or ideal family size norms were not significantly related to any changes in fertility. Fertility
transition was also found to be more pronounced in urban areas compared with rural areas.
Moreover, the findings identified education, infant and child mortality, modern contraceptive
use, the percentage of women in union, and place of residence as significantly related to fertility
levels.

A study by Garenne (2007) attributed stalling fertility transitions at the national level to
changes in age at birth, contraceptive use, and socioeconomic conditions. In a related study,
Ojakaa (2008) observed that the patterns and determinants of fertility transition in Kenya could
be explained by increase in age at first marriage and use of contraception. A reversal in the trend
of fertility decline in the five-year period before the 2003 KDHS could be attributed to a rapid
increase in infant and child mortality in the 1990s. He noted that in the timing of transition from
second to third births in Kenya, regional differentials existed and these could be attributable to
varying levels of socioeconomic development and conservative cultural practices with respect to
reproduction. The study found that rural women were more likely to experience second births
compared with their rural counterparts and that transition from first to second and from second to
third births increased with an increase in age at first birth.

Westoff and Cross (2006) in a study of stalled fertility transition in Kenya observed that
fertility stalled throughout the country but the stall was more pronounced among women with no
education. A stall in contraceptive prevalence occurred among women in the young age groups
and those with little education, but contraceptive use increased for women who had been
sexually active in the four weeks prior to the survey. This pattern was also observed in other
countries in Southern and Eastern Africa and was possibly attributed to the high incidence of
HIV/AIDS in those countries. The stall in contraceptive prevalence could perhaps also be
attributed to the observed decline in the proportion of women who want no more children. This
was a marked departure from the steady increase in the proportion of women wanting no more
children since 1997. The findings imply that HIV/AIDS may have played a role in this reversal
of reproductive preferences in Kenya through an increase in child mortality.

7
 Bongaarts (2006) studied the causes of stalling fertility transitions in developing
countries. He noted that evidence in the late 1990s showed that fertility had stalled in mid-
transition in five countries, including Bangladesh, Dominican Republic, Ghana, Kenya, and
Turkey. He observed that the level of stalling varied from 4.7 births per woman in Kenya to 2.5
births per woman in Turkey. The findings attributed stalling fertility to a plateauing in
contraceptive prevalence and demand for contraception and also a stall in women’s desired
number of children. The findings did not reveal any particular pattern in the socioeconomic
determinants of fertility during period of fertility stall. For instance, in Kenya and Ghana fertility
was found to have stalled, while development indicators had not changed. But in Turkey, Peru,
and Columbia socioeconomic development was noted to have continued. He concluded that
stalling fertility was therefore attributed to the stagnant levels of socioeconomic development in
Kenya and Ghana.

In summary, findings on the causes of fertility transitions and apparent stalling remain
inconsistent and controversial. The role of socioeconomic factors in explaining fertility transition
remains inconclusive. Fertility has been observed to decline even in countries with little
socioeconomic development, such as Bangladesh, and stalled in countries with poorer indicators
of socioeconomic development, such Kenya and Ghana. In other countries no relationship was
established between increase in contraceptive use and fertility decline. Moreover, other countries
have experienced a stall in fertility with no significant changes in unmet need and unwanted
fertility.

8
Figure 1. Conceptual Framework to explain fertility transition in Kenya

Background factors
-Education Proximate factors
-Residence -Ever-use of contraception
-Region -Prior child death Birth transition
-Marital status -Desire for more children
-Religion
-Wealth index
-Age at start of interval

Source: Bongaarts (1978)

Objective of the Study

The study seeks to explore the pathways through which background and proximate
factors may influence transition from second to third birth and from third to fourth birth (see
Conceptual Framework, Figure 1). There could be two such pathways: the first seeks to establish
whether background factors affect transition from second to third birth and from third to fourth
birth, independent of proximate factors. The background factors considered in this study are
women’s education, urban-rural residence, region of the country, marital status, religion, wealth
index, and age at the start of the interval. These are a proxy measure for the socioeconomic
conditions. The assumption is that socioeconomic development is a major cause of fertility
decline over time. This is because the high costs and benefits associated with children motivate
parents to have fewer children, while declines in mortality raise survival chances of children, and
that means that parents do not need to have as many births to achieve their desired family size.
This scenario creates demand for family planning as parents seek to maintain their desired
fertility. Consequently, as a society advances in socioeconomic development, the social costs of
birth control are also reduced.

The second pathway examines whether background factors act through proximate factors
to influence transition from second to third birth and from third to fourth birth. The two
proximate factors included in this study are ever-use of contraception and a prior child death.

9
This stems from the assumption that a woman’s background characteristics determine her
socioeconomic status and hence her knowledge of and access to health services. Therefore,
women with more education and those living in urban areas are more likely to access and use
modern contraceptive methods, and this is likely to influence whether or not they transition from
second to third birth and from third to fourth birth.

Thus we hypothesise that more educated women, women living in urban areas, women
from Central province, currently married women, and women from wealthier households are less
likely to transition from second to third births and from third to fourth births. We also
hypothesise that women who have ever used a modern contraceptive method, women who have
never experienced a prior child death, and women who desire another child are less likely to
transition from second to third births and from third to fourth births.

10
DATA AND METHODS

Sources of Data

The data for this study are from the Kenya Demographic and Health Surveys (KDHS)
conducted in 2003 and 2008/9. These were nationally representative surveys of 8,195 and 8,444
women age 15-49, respectively. They collected information on fertility, marriage, sexual activity,
awareness and use of family planning methods, fertility preferences, and breastfeeding.
Additional information was also collected on malaria and use of mosquito nets, domestic
violence, and HIV testing of adults.

The implementation of the survey was successfully undertaken (CBS et al. 2004, KNBS
et al. 2010). For KDHS 2008/9 a total of 9,936 households were selected in the sample and 9,057
households were successfully interviewed, yielding a response rate of 98%. A total of 8,767
women were found to be eligible for interview and 8,444 were interviewed, for a response rate of
96%. In KDHS 2003, 8,561 of the eligible 8,889 households were successfully interviewed,
yielding a response rate of 96%. In the interviewed households, 8,717 eligible women age 15-49
were identified and 8,195 were successively interviewed, for a response rate of 94%.

We examined the transition to a third birth among women who experienced a second
birth, and the transition to a fourth birth among women who had a third birth. For KDHS 2003 a
total of 689 women had experienced a second birth and 531women transitioned to a third birth,
while 421 experienced a fourth birth. For KDHS 2008/9 a total of 789 women had a second birth
and 593 women transitioned to a third birth, while 477 transitioned to a fourth birth.

A complete birth history covering all live births of each woman interviewed was
obtained. For such births, the survival status was ascertained and also the age at death for those
infants /children who had died. Thus it was possible to determine the spacing between any two
live births and also the intervals at which the infant/child deaths occurred. In addition, the survey
obtained respondents’ background and demographic characteristics. Information on sexual
behaviour and family planning practice was also collected.

11
Variables and Their Measurement

Dependent Variable for the Study

The dependent variable in this study is whether or not a woman made the transition from
second to third birth and from third to fourth birth for the five-year period before the survey, for
KDHS 2003 and KDHS 2008/9.

Key Independent Variables for the Study

Socioeconomic and Cultural Factors

Education: Women’s education is highly correlated with fertility. Education increases

the chances of participating in labour force (or employment) for women. More educated women
may reduce their completed fertility because of competing demands of their time by either
limiting the number of births or by having longer birth intervals (spacing births). This study
classified levels of education into three categories: no education, primary education, and
secondary and above level of education. Educational level refers to the number of years of
formal schooling and does not include education obtained in adult education programmes.

Urban-rural residence: This category refers to where the respondent was living at the
time of the survey, classified as either urban or rural. In the analysis, urban area is used as the
reference category.

Region of residence: Kenya has eight administrative provinces, including Nairobi, the
capital city. Seven dummy variables representing each of these regions are created and Western
province is chosen as the reference category.

Marital status: This variable is included to capture the differential exposure status to
sexual intercourse (differences in coital frequency). This variable is classified into: never-
married, currently married (or cohabiting, as married), and formerly married.

Religion: Respondents’ religious affiliation is recoded into: Catholics, Protestants/other

Christians, Muslims, and other religions. This variable is included to capture some of socio-
cultural issues related childbearing.

12
 Demographic factors: The only demographic factor included in this study is the age of
the woman at the start of the birth interval.

Age at birth of child: The age of the woman at the birth of the child opening the interval
is a proxy for fecundity-related differences in childbearing, since the onset of secondary sterility
rises with age. This is a continuous variable and is measured in completed years.

Proximate Factors

Ever-use of contraception: Contraceptive use represents a deliberate effort by a woman

to avoid conception, either using natural or modern methods. This factor is categorized into:
never used, used traditional, and used modern methods.

Survival status of the preceding child (prior child death): This factor is coded as equal
to 1 if a woman ever experienced a death of any prior child and equal to 0 if it didn’t occur.

Desire for more children: This variable is included to capture future fertility intentions.
It is coded as: want no more children, want a child within two years, and want a child after two
years.

Methods of data analysis: The study used descriptive statistics and logistic regression
model as the main methods of data analysis. Descriptive statistics were used at level one to
describe the distribution of births in the five-year period before the 2003 KDHS and the five-year
period before the 2008/9 KDHS. They were also used to describe the distribution of births by
various key background characteristics and also to estimate parity progressions for the five-year
period before each of the two surveys.

Logistic regression was used to identify the socioeconomic, cultural, demographic, child
mortality, and proximate factors associated with transition from second to third births and from
third to fourths birth in Kenya between 2003 and 2008/9. This regression model was appropriate
because the dependent variable was dichotomous or binary. In this study, the dependent variable
was whether or not a woman made the transition from second to third birth and from third to
fourth birth in Kenya for the period 2003 and 2008/9. The impact of predictor variables is
usually explained in terms of odds ratios. Logistic regression applies maximum likelihood
estimation method to estimate parameters. Eight logistic regression models were fitted. These

13
included; four logistic regression models for transition from second to third birth in the period
between 2003 and 2008/9 and four regression models were fitted for transition from third to
fourth birth for the period 2003 and 2008/9.

For each time period, two models were fitted incorporating only the background variables
for the transition from second to third births and from third to fourth births. These models thus
examined the first pathway in our conceptual framework. Moreover, two other models were
fitted for each time period incorporating both the background and proximate variables for the
transition from second to third births and from third to fourth births. These models were used to
examine the second pathway in our conceptual framework. All analyses were done using
STATA 11.

These transitions are critical because they may be indicative of changing family size
preferences. This is important if the goal of reducing fertility from the current average of 4.6
children per women to replacement level is to be realized by the government.

14
DESCRIPTIVE RESULTS

Table 1 shows the distribution of births in the five-year period prior to each of the
surveys. The results show that the distribution of births by various parities for the two time
periods does not indicate significant differences. The results show that there was a decline in the
proportion of first births of 3 percentage points between 2003 and 2008/9. A slight increase in
the proportion of births of parities 2 and 3 was observed for women in 2008/9, while a similar
proportion was noted for births of parity 4 for both time periods. The results also show that the
mean birth order for the two time periods was more or less the same.

Table 1. Distribution of births in the five-year period before KDHS 2003 and before KDHS 2008/9

Birth order
Period 1 2 3 4 5+ N
1999-2003 24.80 19.63 16.05 11.59 28.65 6,102
2003-2008/9 22.38 21.37 16.65 12.04 27.57 5,852
Mean birth order for 1999-2003 3.5
Mean birth order for 2008/9 3.4

Table 2 shows the distribution of births by key background characteristics. There was an
increase in the proportion of births of parities 2 to 4 between 2003 and 2008/9 for women with
no education. The proportion of births of orders 3 to 5 was slightly higher for women with a
primary education in 2008/9 than in 2003, while the proportion of births of order 2 was higher
for women with secondary and above education in 2008/9. There was a marked increase in the
proportion of births of order 2 to women living in urban areas between 2003 and 2008/9. There
was also a decline in the proportion of births of orders 4 and 5 and above between the surveys.
The results do not indicate significant changes in the proportions of births of various orders
occurring to women living in rural areas between the two surveys.

There was a slight increase in the proportions of births of orders 1, 4, and 5 and above
between 2003 and 2008/9 for women living in Western province. In Nairobi the proportion of
births of order 2 increased by seven percentage points between the surveys, and by five and three
percentage points, respectively, for birth orders 4 and 5 and above. Central province registered a

15
decline in the proportions of births of orders 1 and 2, but a marginal increase for births of order
4. Interesting patterns were observed for Coast province, where a decline of two percentage
points was noted for births of order 4, but the proportion of births of all other orders remained
the same. For Eastern province, there was also a decline for births of order 1 and a marginal
increase in births of order 3. Nyanza province registered an increase in the proportion of births of
orders 1 to 4 and a reduction of seven percentage points for births of orders 5 and above. Rift
Valley province registered a marginal increase in the proportion of births of orders 2 to 4 and a
decline for births of order 1. North Eastern province registered a substantial increase in the
proportion of births of orders 2 and 3, but the province also recorded dramatic declines in the
proportion of births of orders 5 and above, from 51% in 2003 to 38% in 2008/9.

The proportion of births of orders 2 to 5 and above increased for women age 15-24, while
for women age 25-34 the proportion of births of order 2 increased slightly but declined
marginally for other birth orders between 2003 and 2008/9. For women age 35 and above, the
proportion of births of order 5 declined but increased for births of orders 2 and 3. The proportion
of births occurring to currently married women increased by one percentage point for birth
orders 2 and 3 and declined by the same margin for birth orders 1 and 5 and above. The
proportion of births to formerly married women increased for birth order 2 but the proportion
remained the same for birth orders 4 and 5 and above. A decline of seven percentage points in
the proportion of births of order 1 was observed for never-married women and a corresponding
increase in the proportion of births of order 2 between 2003 and 2008/9.

16
Table 2. Distribution of births by key background characteristics for KDHS 2003 and KDHS 2008/9

Birth order
Variable 1 2 3 4 5+ N
Education level
No education
KDHS 2008/9 12.77 16.67 15.14 12.53 42.89 763
KDHS 2003 14.44 12.03 13.40 10.45 49.68 938
Primary education
KDHS 2008/9 20.14 20.10 17.26 12.68 29.82 3,713
KDHS 2003 23.45 20.46 16.08 11.91 28.10 3,901
Secondary plus
KDHS 2008/9 33.77 27.39 15.85 10.02 12.96 1,375
KDHS 2003 33.13 22.17 17.93 11.43 14.75 1,263
Residence
Urban
KDHS 2008/9 34.97 30.71 15.32 8.66 10.34 1,074
KDHS 2003 34.22 22.89 16.09 11.06 15.75 1,143
Rural
KDHS 2008/9 19.55 19.26 16.95 12.80 31.44 4,777
KDHS 2003 21.74 18.88 16.04 11.711 31.63 4,959
Region of residence
Western
KDHS 2008/9 19.40 20.04 16.05 11.50 33.01 703
KDHS 2003 21.11 20.09 15.91 11.12 31.76 776
Nairobi
KDHS 2008/9 40.02 29.27 14.54 8.68 7.49 334
KDHS 2003 39.29 22.09 16.00 12.37 10.25 398
Central
KDHS 2008/9 27.40 26.63 17.65 11.69 16.63 466
KDHS 2003 30.88 22.46 17.75 10.94 17.97 652
Coast
KDHS 2008/9 23.56 19.92 16.12 13.50 26.90 495
KDHS 2003 24.34 20.19 16.32 12.47 26.68 510
Eastern
KDHS 2008/9 19.28 20.85 19.85 12.20 27.83 890
KDHS 2003 22.64 19.53 17.21 12.68 27.95 946
Nyanza
KDHS 2008/9 24.43 19.67 15.49 12.73 27.68 1,144
KDHS 2003 21.28 18.39 13.91 10.95 35.46 1000
Rift Valley
KDHS 2008/9 19.45 21.14 16.20 11.96 31.26 1,641
KDHS 2003 22.74 19.30 16.29 11.33 30.35 1,639
North Eastern
KDHS 2008/9 14.05 17.61 17.41 12.80 38.13 178
KDHS 2003 13.10 10.87 13.47 11.89 50.67 181
Cont’d..

17
 Table 2. Cont’d
Birth order
Variable 1 2 3 4 5+ N
Age cohort
15-24 years
KDHS 2008/9 51.95 29.45 12.48 4.43 1.70 1,943
KDHS 2003 55.29 30.15 11.56 2.52 0.48 2,077
25-34 years
KDHS 2008/9 10.29 22.17 22.37 17.87 27.31 2,845
KDHS 2003 10.95 19.37 23.17 18.95 27.56 2, 861
35+ years
KDHS 2008/9 0.69 4.43 8.97 10.35 75.56 1,063
KDHS 2003 0.66 1.50 6.57 9.67 81.60 1,163
Marital status
Currently married
KDHS 2008/9 18.52 21.43 17.67 12.98 29.41 4,933
KDHS 2003 20.11 20.33 16.95 12.22 30.40 5,230
Formerly married
KDHS 2008/9 20.71 21.79 15.83 11.51 30.16 497
KDHS 2003 26.11 16.99 15.16 11.76 29.98 504
Never married
KDHS 2008/9 69.55 20.13 5.68 1.65 2.99 421
KDHS 2003 77.66 13.37 4.55 2.33 2.09 368
Religion
Catholic
KDHS 2008/9 23.16 23.42 16.48 12.20 24.74 1,170
KDHS 2003 25.41 20.81 16.49 11.23 26.06 1.431
Protestant
KDHS 2008/9 22.98 20.96 16.59 12.05 27.42 3,126
KDHS 2003 24.27 19.88 15.74 11.42 28.70 3,925
Muslim
KDHS 2008/9 19.88 19.38 17.53 11.76 31.45 516
KDHS 2003 19.83 15.81 16.76 13.49 34.11 554
Other
KDHS 2008/9 14.21 22.16 16.57 11.64 35.42 239
KDHS 2003 22.43 16.87 17.15 12.21 31.34 192
Source: Primary analysis of KDHS, 2003 and KDHS, 2008/9

18
Parity Progressions for Kenya, 2003 – 2008/9

Table 3 shows parity progressions for Kenya between 2003 and 2008/9. The results show
that the patterns of parity progressions did not change significantly between the two surveys.
Between 2003 and 2008/9 the proportions of women experiencing births of parities 2 to 5
increased by one percentage point. The proportion of women experiencing a tenth birth increased
from 40% to 57% between 2003 and 2008/9.

Table 3. Parity progression in the five-year periods before the surveys, KDHS 2003 and KDHS
2008/9.

Parity
0-1 1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 10
Parity
progression for
KDHS 2003 0.706 0.706 0.691 0.714 0.712 0.674 0.679 0.607 0.589 0.403
Parity
progression for
KDHS 2008/9 0.749 0.724 0.679 0.700 0.701 0.628 0.625 0.605 0.580 0.569

Multivariate Results

Factors Influencing Transition from Second to Third Birth, KDHS 2008/9

Table 4 for model 1 shows factors influencing transition from second to third birth, for all
background factors. (See the Appendix to this paper for Tables 4 to 11.) The results show that
the odds of transition to a third birth were 0.5 times lower for women with a primary education
compared with women with no education. However, the transition to a third birth was 0.3 times
lower for women with secondary and above education compared with women with no education.
Region of residence was significantly related to the odds of transition to a third birth. A woman
in Central province was 65% less likely to experience a third birth compared with a woman in
Western province. The odds of transition to a third birth was higher for women in Nairobi and
Nyanza provinces compared with women in Western province, and lower for women in Eastern,
Coast, Rift Valley, and North Eastern provinces than for women in Western province. These
results were not statistically significant.

19
 Women of the Protestant faith were 50% more likely to experience a third birth compared
with women of the Catholic faith. The odds of transition to a third birth for Muslim and other
women were higher compared with Catholic women, although these results were not statistically
significant. Wealth index was significantly associated with the odds of transition to a third birth.
Women from middle households were 0.5 times less likely to experience a third birth compared
with women from poorest households. The odds of transition to a third birth were lowest for
women from richest households. Women from richest households were 0.3 times less likely to
transition to a third birth compared with women from poorest households.

The results showed that the odds of transition to a third birth were higher for women in
rural areas than for women in urban areas. They were also higher for currently and formerly
married women compared with never-married women. However, the results were not statistically
significant. Age at the start of the interval was associated with reduced odds of transition to a
third birth.

In model 2, when other control variables were introduced, education level remained a
significant predictor of the odds of transition to a third birth. Table 5 shows that women with
primary and secondary and above level of education were 0.5 and 0.3 times, respectively, less
likely to transition to a third birth compared with women with no education. As in model 1,
region of residence was an important factor influencing transition to a third birth. Women in
Central province were 0.3 times less likely to experience a third birth compared with women in
Western province. Women in Eastern and North Eastern province were 0.5 and 0.4 times,
respectively, less likely to experience a third birth compared with women in Western province.
The odds of transition to a third birth were higher for women in Nairobi and Nyanza provinces
and lower for women in Coast and Rift Valley provinces compared with women in Western
province. These results were not statistically significant.

In the presence of other control variables, women of the Protestant faith were 59% more
likely to experience a third birth compared with women of the Catholic faith. The odds of
transition to a third birth were also higher for women of Muslim and other faiths, although these
results were not statistically significant. In model 2 wealth index remained a significant predictor
of transition to a third birth. Women from middle, richer, and richest households were 0.6, 0.5
and 0.4 times, respectively, less likely to experience a third birth compared with women from

20
poorest households. Women who had ever used a modern contraceptive method were 40% less
likely to experience a third birth compared with women who had never used any contraceptive
method. Age at the start of the interval was significantly related to the odds of transition to a
third birth. A one-year increase in age at the start of the interval reduced the odds of transition to
a third birth by 10%. Furthermore, women who desired another child within two years and
women who desired another child after two years were both associated with reduced odds of
experiencing a third birth compared with women who did not want another child. The study did
not establish a significant association between prior child death and transition to a third birth.

Factors Influencing Transition from Third to Fourth Birth, KDHS 2008/9

Table 6 for model 1 gives results for factors influencing transition from third to fourth
birth, for all background factors. The results show that education level was a significant predictor
of transition to a fourth birth. Women with secondary and above level of education were 57%
less likely to transition to a fourth birth compared with women with no education. The odds were
also lower for women with primary education, although these results were not statistically
significant. Region of residence was an important factor influencing the odds of transition to a
fourth birth. Women living in Eastern provinces were 0.5 times less likely to experience a fourth
birth compared with women in Western province. This association was weakly significant at the
10% level. The odds of experiencing a fourth birth were lower for women in Central, Nairobi,
Coast, Nyanza, and North Eastern provinces compared with women in Western province,
although these associations were not statistically significant.

Wealth index was also an important predictor of the odds of experiencing a fourth birth.
Women from middle and richer households were 0.5 and 0.4 times, respectively, less likely to
experience a fourth birth compared with women from poorest households, while women from
richest households were 0.3 times less likely to experience a transition to a fourth birth compared
with women from poorest households. The odds of transition to a fourth birth were higher among
women living in rural areas, currently married women, and women of the Protestant and Muslim
faiths. These results were not statistically significant. Age at the start of the interval was
significantly associated with reduced odds of transition to a fourth birth.

21
 Table 7 shows results of model 2, including all variables. Women with primary and
secondary and above level of education had lower odds of transition to a fourth birth compared
with women with no education. Wealth index was a significant predictor of the odds of transition
to a fourth birth. Women from richer households were 0.5 times less likely to experience a fourth
birth and women from the richest households were 0.3 times less likely compared with women
from the poorest households. Age at the start of the interval was also significantly related to
transition to a fourth birth. A one-year increase in age at the start of the interval was associated
with a lower transition to a fourth birth. Moreover, the odds of transition to a fourth birth were
lower for women who had ever used a modern contraceptive method compared with women who
had never used any method. The odds of transition to a fourth birth were 0.3 and 0.5 times
higher, respectively, for women who wanted another child within two years and women who
wanted another child after two years compared with women who wanted no more children. The
odds of transition to a fourth birth were higher for women living in rural areas compared with
women in urban areas and were also higher for currently married women compared with never-
married women, although these results were not statistically significant.

Factors Influencing Transition from Second to Third Birth, KDHS 2003

Table 8 for model 1 shows factors influencing transition from second to third birth, for all
background factors. Region of residence was an important factor influencing transition to a third
birth. Women in Central and Nyanza provinces were 0.3 and 0.4 times, respectively, less likely
to experience a third birth compared with women in Western province. The odds of transition to
a third birth was lower for women in Nairobi, Coast, Eastern, and Rift valley provinces
compared with women in Western province. These associations were not statistically significant,
however. Currently and formerly married women were nine and seven times, respectively, more
likely to experience a third birth compared with never-married women. Women of the Muslim
faith were three times more likely to transition to a third birth compared with women of the
Catholic faith. The odds of transition to a third birth were higher for women of Protestant and
other faiths compared with Catholic women, although the results were not statistically
significant. Women from poorer, middle, richer, and richest households were 0.2, 0.3, 0.2, and
0.1 times, respectively, less likely to transition to a third birth compared with women from the

22
poorest households. Age at the start of the interval was also significantly associated with reduced
odds of transition to a third birth.

Table 9 for model 2 presents results including both background and proximate variables.
Women in Nyanza province were 0.3 times less likely to experience a third birth compared with
women in Western province. The odds of transition to a third birth were lower for women living
in Nairobi, Central, Coast, Eastern, Rift Valley, and North Eastern provinces, although the
results were not statistically significant. Currently and formerly married women were also
associated with educed odds of transition to a third birth compared with never-married women.
Women from poorer, middle, richer, and richest households were 0.3, 0.3, 0.2, and 0.1 times,
respectively, less likely to experience a third birth compared with women from poorest
households. Age at the start of the interval was also associated with reduced odds of transition to
a third birth.

Furthermore, women who had ever used a modern contraceptive method were 0.4 times
less likely to transition to a third birth compared with women who had never used any method.
Women who wanted another child within two years were 0.3 times less likely to transition to a
third birth compared with women who did not want another child. The odds of transition to a
third birth were higher among women with primary and secondary and above education, among
women in rural areas, and among women of Protestant, Muslim, and other religions, and were
lower for women who experienced a prior child death. However, these associations were not
statistically significant.

Factors Influencing Transition from Third to Fourth Birth, KDHS 2003

Table 10 for model 1 shows results for factors influencing transition from third to fourth
birth, for all background factors. Women in Central province were 0.4 times less likely to
experience a transition to a fourth birth compared with women living in Western province. The
odds of transition to a fourth birth were higher for women in Nairobi and Nyanza provinces,
although the results were not statistically significant. With respect to wealth index, the results
show that women from poorer, middle, richer, and richest households were 0.4, 0.3, 0.4 and 0.3
times, respectively, less likely to experience a fourth birth compared with women from the
poorest households. Age at the start of the interval was also significantly associated with reduced

23
odds of transition to a fourth birth. The risk of transition to a fourth birth was higher for women
with primary level of education, women in rural areas, women of the Muslim and other faiths,
and currently and formerly married women, and were lower for women with secondary and
above level of education. However, these results were not statistically significant.

Table 11 shows results of model 2, which includes both background and proximate
variables. Women in Central province were 0.4 times less likely to experience a fourth birth
compared with women in Western province. Generally, except for women in Nairobi, the odds of
transition to a fourth birth were lower for women in other provinces. These results were not
statistically significant. Women from poorer, middle, and richer households were 0.4 times less
likely to experience a fourth birth, while women from richest households were 0.3 times less
likely to transition to a fourth birth compared with women from the poorest households. Age at
the start of the interval was significantly associated transition to a fourth birth.

A one-year increase in age at the start of the interval was associated with 0.9 times lower
odds of transition to a fourth birth. Women who had ever used a modern contraceptive method
were associated with reduced odds of transition to a fourth birth compared with women who had
never used any method. Women who wanted another child within two years were 0.4 times less
likely to transition to a third birth compared with women who did not want another child at all.
The results also indicate that the odds of transition to a fourth birth were higher for women with
primary and secondary and above education, for women in rural areas, currently married women,
women of Muslim and other faiths, and women who experienced a prior child death. However,
these results were not statistically significant.

24
DISCUSSION

This paper tested a number of hypotheses regarding factors influencing transition from
second to third births and from third to fourth births for Kenya between 1999-2003 and 2003-
2008/9. We hypothesised that more educated women, women living in urban areas, women from
Central province, currently married women, and women from wealthier households were less
likely to transition from second to third births and from third to fourth births. We also
hypothesised that women who had ever used a modern method of contraception and those who
had never experienced a prior child death were less likely to transition from second to third births
and from third to fourth births.

For each of the two surveys studied, KDHS 2003 and KDHS 2008/9, two models were
fitted. The first model included only background variables—education level, urban-rural
residence, region of residence, wealth index, marital status, religion, and age at the start of the
interval. The second model included both background variables and proximate variables. The
results for model 1 for the five-year period before KDHS 2008/9 for transition from second to
third births showed that education level, region of residence, wealth index, religion, and age at
the start of the interval were significantly associated with the odds of transition to a third birth.
Primary and secondary and above levels of education were associated with lower odds of
transition to a third birth. Residence in Central province was also associated with lower odds of
transition to a third birth. The odds of transition to a third birth were higher for women of the
Protestant faith and lower for women from middle, richer, and richest households.

For the full model 2 for 2008/9, the results showed that the odds of transition to a third
birth were lower for women with primary and secondary and above level of education compared
with women with no education, and also lower for women in Central and North Eastern
provinces compared with women in Western province. Odds were higher for women of the
Protestant faith compared with Catholic women. The odds of transition to a third birth were
lower for women from middle, richer, and richest households compared with those from poorest
households. The odds of transition to a third birth were also lower for women who had ever used
a modern contraceptive method compared with women who had never used any method. Age at
the start of the interval and desire for another birth within at least two years were associated with
reduced odds of transition to a third birth.

25
 Concerning transition from second to third birth, the results of model 1 for the five-year
period before KDHS 2003 showed that region of residence, marital status, religion, wealth index,
and age at the start of the interval were significantly associated with the odds of transition to a
third birth. Women in Central province and women in Eastern province had lower odds of
transition to a third birth compared with women in Western province. Currently and formerly
married women had increased odds of transition to a third birth compared with never-married
women, as did women of the Muslim faith compared with women of the Catholic faith. Women
from poorer, middle, richer, and richest households had lower odds of transition to a third birth
compared with women from poorest households.

Model 2 for KDHS 2003, which included both background and proximate variables,
showed that region of residence, marital status, wealth index, age at the start of the interval, ever-
use of contraception, and desire for more children were significant predictors of transition to a
third birth. Residence in Nyanza province was associated with lower odds of transition to a third
birth compared with residence in Western province. Currently and formerly married women had
significantly lower odds of transition to a third birth compared with never-married women.
Women from poorer, middle, richer, and richest households had lower odds of transition to a
third birth compared with women from poorest households, as did women who had ever used a
contraceptive method compared with women who had never used a method. The odds of
experiencing a third birth were lower for women who wanted another child within two years
compared with those who did not want another child at all.

For the five-year period before KDHS 2008/9, the odds of transition from third to a
fourth birth in model 1 were strongly associated with women’s education level, wealth index,
and age at the start of the interval. The odds of transition to a fourth birth were lower for women
with secondary and above education compared with women with no education, and were
significantly lower for women from middle, richer, and richest households compared with
women from poorest households. Odds also decreased with an increase in age at the start of the
interval.

The results for model 1 for the five-year period before KDHS 2003 showed that only
region of residence, wealth index, and age at the start of the interval were significantly associated
with the odds of transition to a fourth birth. Women in Central province were associated with

26
lower odds of transition to a fourth birth compared with women from Western province. The
odds of transition to a fourth birth were lower for women from poorer, middle, richer, and richest
households compared with women from poorest households. Model 2, which included
background and proximate variables, showed that region of residence, wealth index, age at the
start of the interval, ever-use of contraception, and desire for another child were significant
predictors of transition to a fourth birth. Women in Central province were associated with lower
odds of transition to a fourth birth compared with women in Western province. This same pattern
was observed for women from poorer, middle, richer, and richest households compared with
poorest households, women those who had ever used a method of contraception compared with
women who had never used a method, and women who wanted another child within two years
compared with women who wanted no more children. Age at the start of the interval was also
associated with reduced odds of transition to a fourth birth.

Conclusions and Policy Implications

A major conclusion of this analysis is that transitions from second to third births and
from third to fourth births in Kenya were strongly influenced by socioeconomic, cultural, and
proximate factors. This observation supports the view that socioeconomic conditions and
diffusion hypotheses are important in explaining fertility transition in Kenya. Consequently, at
the policy level there is a need to improve family planning services across the country. This will
go a long way in addressing the issues of unmet need for family planning and unwanted fertility
and hence lead to a decline in total fertility. The government has already instituted measures to
reposition family planning in the country. There is also a need to improve the socioeconomic
conditions in the country, recognizing that relative wealth and more education for women are
strongly associated with lower fertility. This means that more efforts should be directed toward
increasing GDP per capita and increasing the proportion of women with secondary and above
education. Education plays a key role in changing reproductive attitudes and behaviour. An
improved GDP per capita will provide more opportunities for employment, better health care,
and alternative investments for families besides having children. These improvements will
ultimately have an impact on desired family size and lead to a decline in fertility rates.

27
REFERENCES

Abdalla, A.A.M. 1988. “Work Experience of Married Women and Desired Fertility in Egypt and
Ghana.” In Studies in African and Asian Demography: CDC Annual Seminar 1987, CDC
Research Monograph Series No. 17, 293-315. Cairo, Egypt: Cairo Demographic Centre.

Adegbola, O. 1988. “Religion and Reproduction in Sub-Saharan Africa.” Paper presented at

African Population Conference, Dakar, Senegal, November 7-12.

Alila, D.D. 1990. “Determinants of Family Size Preferences in Urban Population: A Case Study
of Mathare Valley-4B, Nairobi.” MA Thesis. University of Nairobi, Kenya.

Ayehu, T.H. 1998. “Correlates of Fertility Preference of Kenyan Women: Evidence from the
1993 KDHS.” MS Thesis. University of Nairobi, Kenya.

Benefo, K. 1995. “The Determinants of the Duration of Postpartum Sexual Abstinence in West
Africa: A Multilevel Analysis.” Demography 32(2): 139-59.

Blacker, J. 2002. “Kenya’s Fertility Transition: How Low Will It Go?” Unpublished Manuscript.

Blacker, J., C. Opiyo, M. Jasseh, A. Sloggett, and J. Ssematte-Ssebuliba. 2005. “Fertility in

Kenya and Uganda: A Comparative Study of Trends and Determinants.” Population
Studies 59(3): 355-73.

Bledsoe, C.H., J.B. Casterline, J.A. Johnson-Kuhn, and J.G. Haaga, eds. 1999. Critical
Perspectives on Schooling and Fertility in the Developing World. Washington D.C.,
USA: National Academic Press.

Bongaarts, J. 2006. “The Causes of Stalling Fertility Transitions.” Studies in Family Planning
37(1): 1-16.

Bongaarts, J. 1978. “A Framework for Analyzing the Proximate Determinants of Fertility.”

Population and Development Review 4(1): 105-32.

Brass, W., and C.L. Jolly, eds. 1993. Population Dynamics of Kenya. Washington D.C., USA:
National Academic Press.

Caldwell, J.C., and P. Caldwell. 1987. “The Cultural Context of High Fertility in Sub-Saharan
Africa.” Population and Development Review 13(3): 409-37.

28
Central Bureau of Statistics (CBS) [Kenya], Ministry of Health (MOH) [Kenya], and ORC
Macro. 2004. Kenya Demographic and Health Survey 2003. Calverton, MD, USA: CBS,
MOH and ORC Macro.

Cohen, B. 1993. “Fertility Levels, Differentials and Trends.” In Demographic Change in Sub-
Saharan Africa, edited by K.A. Foote, K.H. Hill, and L.G. Martin, 8-67. Washington
D.C., USA: National Academic Press.

Cross, A.R., W. Obungu, and P. Kizito. 1991. “Evidence of a Transition to Lower Fertility in
Kenya.” International Family Planning Perspectives 17(1): 4-7.

Diamond, I., M. Newby, and S. Varle. 1999. “Female Education and Fertility: Examining the
Links.” In Critical Perspectives on Schooling and Fertility in the Developing World,
edited by C.H. Bledsoe, J.B. Casterline, J.A. Johnson-Kuhn, and J.G. Haaga, 23-48.
Washington D.C., USA: National Academic Press.

Dixon-Mueller, R. 1993. Population Policy and Women’s Rights: Transforming Reproductive

Choice. Westport, CT, USA: Praeger.

Ekisa, L.A., and A. Hinde. 2005. Fertility Transition in Kenya: A Regional Analysis of
Proximate Determinants. Applications and Policy Working Paper A06/03. Southampton,
UK: Southampton Statistical Sciences Research Institute.

Ezeh, A.C. 1997. “Polygyny and Reproductive Behaviour in Sub-Saharan Africa: A Contextual
Analysis.” Demography 34(2): 209-43.

Ezeh, A.C., and F.N. Dodoo. 2001. “Institutional Change and African Fertility Transition: The
Case of Kenya.” Genus LVII(3-4): 135-64.

Garenne, M. 2007. Situations of Fertility Stall in Sub-Saharan Africa. Paper presented at the
Fifth African Population Conference, Arusha, Tanzania, December 10-14.

Gyimah, S.O. 2002. Fertility Response to Childhood Mortality in Sub-Saharan Africa with
Emphasis on Ghana and Kenya. PSC Discussion Papers Series Vol. 16: Iss. 2, Article 1.
London, Canada: Population Studies Centre, University of Western Ontario.

Hirschman, C., and Y. Young. 1998. The Decline of Fertility in Thailand, Malaysia and the
Philippines: 1968-70 to 1988-90. Paper presented at the Conference on Economic

29
 Aspects of Demographic Transitions: The Experience of Asian Pacific Countries, Taipei,
Taiwan, June 19-20.

Kenya National Bureau of Statistics (KNBS) [Kenya], and ICF Macro. 2010. Kenya
Demographic and Health Survey 2008-09. Calverton, MD, USA: KNBS and ICF Macro.

Lloyd, C., and S. Ivanov.1988. “The Effects of Improved Child Survival on Family Planning
Practice and Fertility.” Studies in Family Planning 19(3): 141-161

Magadi, A.K., and A.O. Agwanda. 2010. “Investigating the Association between HIV/AIDS and
Recent Fertility Patterns in Kenya.” Journal of Social Science and Medicine 71: 335-44.

Mason, K.O. 1997. “Explaining Fertility Transitions.” Demography 34: 443-54.

Mlewa, O.K. 2001. “Fertility Preference among Men in Kenya.” MA Thesis. University of
Nairobi, Kenya.

Montgomery, M.R., and C.B. Lloyd. 1999. “Excess Fertility, Unintended Births, and Children’s
Schooling.” In Critical Perspectives on Schooling and Fertility in the Developing World,
edited by C.H. Bledsoe, J.B. Casterline, J.A. Johnson-Kuhn, and J.G. Haaga, 206-266.
Washington D.C., USA: National Academic Press.

Moustafa, E.M. 1988. “Factors Related to Supply and Demand for Children, Upper Egypt
Region.” In Studies in African and Asian Demography: CDC Annual Seminar 1987, CDC
Research Monograph Series No. 17, 111-46. Cairo, Egypt: Cairo Demographic Centre.

Muinde, J.N., and M.S. Mukras. 1979. Some Aspects of Determinants of Fertility in Kenya.
Nairobi, Kenya: Population Studies and Research Institute and Economics Department,
University of Nairobi.

Njogu, W.E., and C.T. Martín. 1991. “Fertility Decline in Kenya: The Role of Timing and
Spacing of Births.” In Proceedings of the Demographic and Health Surveys World
Conference Vol. III, 1883-1901. Columbia, MD, USA: IRD/Macro International.

Obiero, W.O. 1999. “Educational Expansion and Emerging Fertility Transition in Kenya.” PhD
Thesis, University of Michigan.

30
Oheneba-Sakyi, Y., and B.K. Takyi. 1997. “Effects of Couple’s Characteristics on Contraceptive
Use in Sub-Saharan Africa: The Ghanaian Example.” Journal of Biosocial Science 29(1):
33-49.

Ojakaa, D. 2008. “The Fertility Transition in Kenya: Patterns and Determinants.” PhD Thesis.
University of Montreal, Canada.

Population Council. 1998. Fertility Decline in Kenya: Levels, Trends and Differentials. Nairobi,
Kenya: The African Population Policy Research Centre.

Robinson, W.C. 1992. “Kenya Enters the Fertility Transition.” Population Studies 46(3): 445-57.

Rodriguez, G., and W. Cleland. 1981. “The Effects of Socioeconomic Characteristics on Fertility
in 20 Countries.” International Family Planning Perspectives 7(3): 93-101.

Shapiro, D., and T. Gebreselassie. 2008. “Fertility Transition in Sub-Saharan Africa: Falling and
Stalling.” Paper presented at the Population Association of America 2007 Annual
Meeting, New York, NY, March 29-31.

Shapiro, D., and B.O. Tambashe. 1997. “Education, Employment and Fertility in Kinshasa and
Prospects for Change in Reproductive Behaviour.” Population Research and Policy
Review 16(3): 259-87.

Sibanda, A. 1999. “Reproductive Change in Zimbabwe and Kenya: The Role of Proximate
Determinants in Recent Fertility Trends.” Social Biology 46(1-2): 82-99.

United Nations. 1995. Women’s Education and Fertility Behaviour: Recent Evidence from the
DHS. New York, NY, USA: United Nations.

United Nations. 1991. Socioeconomic Development and Fertility Decline: An Application of the
Easterlin Synthesis Approach to Data from the World Fertility Survey. New York, NY,
USA: United Nations.

Vavrus, F. 2000. The Relationship between Schooling and Fertility in Tanzania. Boston, MA,
USA: Harvard School of Public Health.

Wasao, S.W. 2002. “A Comparative Analysis of the Socio-Economic Correlates of Fertility in

Cameroon and the Central African Republic.” In Prospects for Fertility Decline in High

31
 Fertility Countries: Population Bulletin of the United Nations 2002, edited by United
Nations, 129-146. New York, NY, USA: United Nations.

Watkins, S.C. 2000. “Local and Foreign Models of Reproduction in Nyanza Province, Kenya.”
Population and Development Review 26(4): 725-59.

Westoff, C.F. 1994. “Reproductive Preferences and Future Fertility in Developing Countries.” In
the Future Population of the World: What Can We Know Today? edited by W. Lutz, 83-
97. London, UK: Earthscan Publications Ltd.

Westoff, C.F., and A.R. Cross. 2006. The Stall in the Fertility Transition in Kenya. DHS
Analytical Studies No. 9. Calverton, MD, USA: ORC Macro.

Woldemicael, G. 2005. Evidence of Recent Fertility Decline in Eritrea: An Analysis of Trends

and Determinants. MPID Working Paper WP-2005-026. Rostock, Germany: Max Planck
Institute for Demographic Research.

Zhang, J. 1994. “Socioeconomic Determinants of Fertility in Hebei Province, China: An

Application of the Sequential Logit Model.” Economic Development and Cultural
Change 43(1): 67-90.

32
APPENDIX

Table 4. Logistic regression results for transition from second birth to third birth in the five-year
period before the survey: KDHS 2008/9-Model 1: background variables

Variable Exp. (β) S.E. (β) P-value

Educational level
No education (Ref.) - - -
Primary education 0.521 0.167 0.042
Secondary plus 0.315 0.116 0.002
Residence
Urban (Ref.) - - -
Rural 1.426 0.412 0.220
Region of residence
Western (Ref.) - - -
Nairobi 1.432 0.684 0.452
Central 0.349 0.118 0.002
Coast 0.646 0.235 0.229
Eastern 0.565 0.182 0.077
Nyanza 1.036 0.320 0.910
Rift Valley 0.764 0.236 0.384
North Eastern 0.358 0.196 0.060
Marital status
Never married (Ref.) - - -
Currently married 2.062 1.086 0.170
Widowed/Divorced 1.094 0.618 0.873
Religion
Catholic (Ref.) - - -
Protestant 1.499 0.310 0.051
Muslim 1.573 0.588 0.225
Other 1.625 0.865 0.362
Wealth index
Poorest (Ref.) - - -
Poorer 0.772 0.220 0.364
Middle 0.504 0.139 0.013
Richer 0.398 0.112 0.001
Richest 0.345 0.127 0.004

Age at start of interval 0.902 0.019 0.000

33
Table 5. Logistic regression results for transition from second birth to third birth in the five-year
period before the survey: KDHS 2008/9- Model 2: all variables

Variable Exp. (β) S.E. (β) P-value

Educational level
No education (Ref.) - - -
Primary education 0.486 0.166 0.035
Secondary plus 0.290 0.114 0.002
Residence
Urban (Ref.) - - -
Rural 1.484 0.435 0.178
Region of residence
Western (Ref.) - - -
Nairobi 1.406 0.679 0.480
Central 0.326 0.113 0.001
Coast 0.703 0.260 0.341
Eastern 0.514 0.170 0.044
Nyanza 1.076 0.339 0.816
Rift Valley 0.774 0.244 0.417
North Eastern 0.347 0.190 0.054
Marital status
Never married (Ref.) - - -
Currently married 2.203 1.183 0.141
Widowed/Divorced 1.044 0.600 0.941
Religion
Catholic (Ref.) - - -
Protestant 1.593 0.338 0.028
Muslim 1.739 0.662 0.146
Other 1.438 0.769 0.497
Wealth index
Poorest (Ref.) - - -
Poorer 0.852 0.250 0.585
Middle 0.560 0.160 0.146
Richer 0.458 0.135 0.008
Richest 0.407 0.157 0.020
Ever use of contraception
Never used (Ref.) - - -
Used traditional 1.033 0.498 0.946
Used modern 0.600 0.151 0.042
Prior child death
Alive (Ref.) - - -
Dead 1.249 0.292 0.341

Age at start of interval 0.890 0.019 0.000

Desire for more children
Wants no more (Ref.) - - -
Wants within two years 0.539 0.164 0.042
Wants after two years 0.532 0.116 0.004

34
Table 6. Logistic regression results for transition from third birth to four birth in the five-year
period before the survey: KDHS 2008/9-Model 1: background variables

Variable Exp. (β) S.E. (β) P-value

Educational level
No education (Ref.) - - -
Primary education 0.619 0.201 0.140
Secondary plus 0.434 0.168 0.031
Residence
Urban (Ref.) - - -
Rural 1.124 0.389 0.735
Region of residence
Western (Ref.) - - -
Nairobi 0.801 0.459 0.698
Central 0.534 0.210 0.111
Coast 1.033 0.445 0.071
Eastern 0.521 0.188 0.016
Nyanza 0.662 0.227 0.228
Rift Valley 0.893 0.317 0.749
North Eastern 0.766 0.459 0.656
Marital status
Never married (Ref.) - - -
Currently married 1.728 1.339 0.480
Widowed/Divorced 0.857 0.695 0.849
Religion
Catholic (Ref.) - - -
Protestant 1.183 0.293 0.498
Muslim 1.267 0.517 0.562
Other 0.701 0.406 0.540
Wealth index
Poorest (Ref.) - - -
Poorer 0.678 0.209 0.208
Middle 0.448 0.142 0.011
Richer 0.410 0.134 0.007
Richest 0.284 0.121 0.003

Age at start of interval 0.885 0.019 0.000

35
Table 7. Logistic regression results for transition from third to fourth birth in the five-year period
before the survey: KDHS 2008/9- Model 2: all variables

Variable Exp. (β) S.E. (β) P-value

Educational level
No education (Ref.) - - -
Primary education 0.519 0.184 0.064
Secondary plus 0.365 0.152 0.015
Residence
Urban (Ref.) - - -
Rural 1.182 0.413 0.632
Region of residence
Western (Ref.) - - -
Nairobi 0.969 0.564 0.957
Central 0.577 0.231 0.169
Coast 1.145 0.503 0.758
Eastern 0.525 0.196 0.085
Nyanza 0.735 0.259 0.381
Rift Valley 0.909 0.333 0.795
North Eastern 0.870 0.541 0.824
Marital status
Never married (Ref.) - - -
Currently married 1.855 1.475 0.437
Widowed/Divorced 0.833 0.691 0.826
Religion
Catholic (Ref.) - - -
Protestant 1.210 0.307 0.452
Muslim 1.394 0.590 0.432
Other 0.714 0.432 0.478
Wealth index
Poorest (Ref.) - - -
Poorer 0.741 0.234 0.342
Middle 0.529 0.177 0.057
Richer 0.487 0.165 0.034
Richest 0.336 0.146 0.012

Age at start of interval 0.879 0.020 0.000

Ever use of contraception
Never used (Ref.) - - -
Used traditional 1.036 0.555 0.948
Used modern 0.546 0.156 0.034
Prior child death
Alive (Ref.) - - -
Dead 1.473 0.356 0.109
Desire for more children
Wants no more (Ref.) - - -
Wants within two years 0.335 0.122 0.003
Wants after two years 0.514 0.143 0.017

36
Table 8. Logistic regression results for transition from second birth to third birth in the five-year
period before the survey: KDHS 2003-Model 1: background variables

Variable Exp. (β) S.E. (β) P-value

Educational level
No education (Ref.) - - -
Primary education 1.345 0.454 0.380
Secondary plus 0.851 0.318 0.666
Residence
Urban (Ref.) - - -
Rural 0.954 0.278 0.872
Region of residence
Western (Ref.) - - -
Nairobi 0.474 0.205 0.084
Central 0.325 0.121 0.003
Coast 0.513 0.224 0.126
Eastern 0.645 0.260 0.276
Nyanza 0.381 0.145 0.011
Rift Valley 0.585 0.214 0.142
North Eastern 0.641 0.426 0.504
Marital status
Never married (Ref.) - - -
Currently married 9.317 7.931 0.009
Widowed/Divorced 6.735 5.900 0.029
Religion
Catholic (Ref.) - - -
Protestant 1.343 0.291 0.173
Muslim 2.511 1.146 0.044
Other 2.182 1.314 0.195
Wealth index
Poorest (Ref.) - - -
Poorer 0.239 0.088 0.000
Middle 0.311 0.116 0.002
Richer 0.174 0.064 0.000
Richest 0.123 0.053 0.000

Age at start of interval 0.896 0.021 0.000

37
Table 9. Logistic regression results for transition from second birth to third birth in the five-year
period before the survey: KDHS 2003- Model 2: all variables

Variable Exp. (β) S.E. (β) P-value

Educational level
No education (Ref.) - - -
Primary education 1.671 0.610 0.160
Secondary plus 1.281 0.527 0.548
Residence
Urban (Ref.) - - -
Rural 0.969 0.288 0.915
Region of residence
Western (Ref.) - - -
Nairobi 0.444 0.196 0.066
Central 0.315 0.122 0.154
Coast 0.528 0.236 0.256
Eastern 0.655 0.271 0.306
Nyanza 0.343 0.134 0.006
Rift Valley 0.578 0.216 0.143
North Eastern 0.753 0.517 0.679
Marital status
Never married (Ref.) - - -
Currently married 7.961 6.799 0.015
Widowed/Divorced 4.552 4.012 0.086
Religion
Catholic (Ref.) - - -
Protestant 1.341 0.297 0.185
Muslim 2.918 1.411 0.027
Other 1.951 1.865 0.272
Wealth index
Poorest (Ref.) - - -
Poorer 0.259 0.098 0.000
Middle 0.369 0.143 0.010
Richer 0.195 0.074 0.000
Richest 0.143 0.064 0.000

Age at start of interval 0.879 0.021 0.000

Ever use of contraception
Never used (Ref.) - - -
Used traditional 1.561 0.798 0.383
Used modern 0.403 0.109 0.001
Prior child death
Alive (Ref.) - - -
Dead 1.056 0.286 0.840
Desire for more children
Wants no more (Ref.) - - -
Wants within two years 0.269 0.087 0.000
Wants after two years 0.759 0.183 0.253

38
Table 10. Logistic regression results for transition from third birth to fourth birth in the five-year
period before the survey: KDHS 2003- Model 1: background variables

Variable Exp. (β) S.E. (β) P-value

Educational level
No education (Ref.) - - -
Primary education 1.123 0.365 0.721
Secondary plus 0.837 0.337 0.659
Residence
Urban (Ref.) - - -
Rural 1.766 0.633 0.113
Region of residence
Western (Ref.) - - -
Nairobi 1.332 0.762 0.616
Central 0.368 0.141 0.009
Coast 0.532 0.252 0.183
Eastern 0.484 0.193 0.069
Nyanza 1.037 0.429 0.930
Rift Valley 0.893 0.333 0.761
North Eastern 4.003 3.705 0.134
Marital status
Never married (Ref.) - - -
Currently married 2.610 2.423 0.301
Widowed/Divorced 1.276 1.225 0.799
Religion
Catholic (Ref.) - - -
Protestant 0.899 0.239 0.689
Muslim 1.485 0.741 0.428
Other 3.044 3.441 0.325
Wealth index
Poorest (Ref.) - - -
Poorer 0.419 0.152 0.016
Middle 0.319 0.113 0.001
Richer 0.415 0.158 0.021
Richest 0.260 0.123 0.005

Age at start of interval 0.904 0.025 0.000

39
Table 11. Logistic regression results for transition from third birth to fourth birth in the five-year
period before the survey: KDHS 2003- Model 2: all variables

Variable Exp. (β) S.E. (β) P-value

Educational level
No education (Ref.) - - -
Primary education 1.169 0.406 0.654
Secondary plus 1.015 0.440 0.973
Residence
Urban (Ref.) - - -
Rural 1.940 0.709 0.070
Region of residence
Western (Ref.) - - -
Nairobi 1.280 0.752 0.674
Central 0.395 0.153 0.016
Coast 0.468 0.228 0.119
Eastern 0.472 0.191 0.063
Nyanza 0.966 0.408 0.934
Rift Valley 0.882 0.334 0.741
North Eastern 3.890 3.697 0.153
Marital status
Never married (Ref.) - - -
Currently married 1.864 1.723 0.500
Widowed/Divorced 0.851 0.816 0.866
Religion
Catholic (Ref.) - - -
Protestant 0.887 0.241 0.659
Muslim 1.349 0.697 0.562
Other 2.076 2.330 0.515
Wealth index
Poorest (Ref.) - - -
Poorer 0.408 0.153 0.017
Middle 0.351 0.128 0.004
Richer 0.422 0.165 0.027
Richest 0.310 0.150 0.016

Age at start of interval 0.909 0.026 0.001

Ever use of contraception
Never used (Ref.) - - -
Used traditional method 0.448 0.211 0.088
Used modern method 0.441 0.123 0.003
Prior child death
Alive (Ref.) - - -
Dead 1.343 0.325 0.223
Desire for more children
Wants no more (Ref.) - - -
Wants within two years 0.374 0.148 0.013
Wants after two years 1.288 0.404 0.419

40