Page 100 - IJPS-11-3

International Journal of
Population Studies Male fertility in Uganda

demographic and proximate factors associated with male level, occupation, wealth status, number of current
fertility using the 2006 and 2016 UDHS. wives, access to mass media (through television, reading
newspapers, and listening to radio), contraceptives use,
2. Data and methods age of first child’s birth, duration of cohabitation and

2.1. Study design and scope marriage, lifetime sexual partners, number of women with
whom men fathered children, and partner age difference.
This study used cross-sectional data from the 2006 – All these variables included in the conceptual framework
2016 UDHS rounds. These surveys used a two-stage (Figure 1) were significantly associated with CEB at the
sampling design to generate countrywide representative bivariate analysis level and were considered for inclusion
household samples. The first stage involved the selection of in the multivariate analysis. During the examination of
enumeration areas from a list of clusters generated in the the determinants, all study variables were recoded into
2002 and 2014 Uganda National Population and Housing customized categories to simplify and ensure meaningful
Census sampling frames. The second sampling stage interpretation of analysis, as shown in Tables 1-3.
included randomly selecting households in each cluster Partner age difference, age of the respondent, wealth
from an updated list of eligible households (UBOS and ICF, status, and number of women with whom men fathered
2018). Although the authors initially considered using all children were excluded from multivariate analysis due to
existing UDHS rounds that included men as respondents multicollinearity.
to gain substantial insights into male fertility dynamics in
Uganda, considerable missing data from non-matching 2.5. Data analysis
variables in the 2000/2001 survey round led to its exclusion. To guide the selection of the UDHS survey rounds for
Thus, only the 2006, 2011, and 2016 survey datasets were inclusion in the regression model analysis, the authors
used for the estimation of male age-specific fertility rates applied the own-children method as described by
(ASFRm). For examining the determinants of male fertility, Schoumaker (2017). The results from this application,
the 2006 and 2016 surveys were analyzed—2006 serving as presented in Table 1 and Figures 2 and 3, are very useful
the baseline and 2016 as the most recent survey dataset. for understanding the trends in prevailing male fertility
patterns. In brief, the own-children method (developed
2.2. Source of data
by Grabill and Cho in 1965) was used due to its success
Secondary data for analysis were obtained from the UDHS in estimating current fertility, especially in contexts where
conducted in 2006, 2011, and 2016 by the UBOS and vital registration systems are almost non-existent, as is the
ICF International. The Household Recode file provided case in many developing countries such as Uganda (Abbasi-
household schedule data on member characteristics. The Shavazi, 1997; Dubuc, 2009; Schoumaker, 2017). This
Individual Woman’s Recode file provided data on child method is accurate and recommended for the estimation
characteristics from the woman’s individual full birth of current fertility using survey data, as children living
history, while the Man’s Recode file provided data on men with each household member are listed in the household
aged 15 – 54 years. The Individual Woman’s Recode file schedules during survey enumeration (Abbasi-Shavazi,
was used because the Man’s Recode file does not capture 1997; Avery et al., 2013; Dubuc, 2009; Schoumaker, 2017).
the full birth history of male respondents. The own-child This method is particularly important for this study
method used both the Household and Individual Woman’s because birth histories for men are not regularly recorded,
Recode files, while the regression model for examining resulting in a lack of available data, unlike for women
determinants used only the Man’s Recode file. (Schoumaker, 2017).
2.3. Sample size In this method, all children aged 0 – 4 years in each
household were linked to their fathers if their survival
During the analysis to examine the determinants of CEB, status was known to be alive. All records for children whose
the sample sizes for men aged 15 – 54 years were 2503 in biological father’s survival status was known to be dead or
2006 and 5336 in 2016. All men who responded to the uncertain were excluded from the analysis. This data were
Man’s Recode questionnaire were included in the study, extracted from the Household Recode files for each of the
irrespective of whether they had ever had sex. In total, the consecutive survey rounds (2006, 2011, and 2016). To
actual sample size was 7839 male respondents. control for fertility underestimation due to selection bias
from the assumption that all children must have lived with
2.4. Study variables
their biological parents at the time of household survey
The variables included in the conceptual framework are as enumeration, an estimation of the age of any surviving
follows: man’s age, religion, place of residence, education fathers who were not living with their biological children

Volume 11 Issue 3 (2025) 94 https://doi.org/10.36922/ijps.461