Page 47 - IJPS-5-1

Mirembe, et al.

nine districts. The study was conducted through face to face interviews. Age was the key inclusion/exclusion criterion
whereby persons aged 18-35 were eligible for inclusion in the study. The study administered an individual questionnaire,
which meant that members of the household who were present at the time of the interviews and were in the age range of
18-35 years would be considered in the study. A total number of 1148 migrants and 376 nonmigrants were considered in
the study.

2.2. Study Design
The study considered four broad national regions (Central, East, North, and West) plus the capital city Kampala. From
each region, two districts were selected at random. These were Masaka and Mubende (Central region), Busia and Mbale
(Eastern region), Arua and Gulu (Northern region), and Mbarara and Hoima (Western region). Kampala the capital city
was purposely selected as the ninth district due to its primate city status, destination of 4612 large in-migrants and
prevalence of complex employment dynamics. Respondents were proportionally allocated to the nine districts factoring
in the proportion of youths in each district as informed by the National Population and Housing Census. Simple random
sampling was used to select the youths from each district for the interview. The study operationally considered youths to
be persons aged 18-35 years and this population subgroup constitutes about 33% of the population in the selected districts.
In addition, a total of 48 shortlisted enumerators and nine supervisors were ultimately recruited, trained, and deployed
to collect the data. A pre-test was carried out in November 2017 followed by the main data collection exercise in the
subsequent month. Both the pre-test and the main data collection exercises used computer-assisted personal interviewing
method. Uploads of data were effected onto the server for survey chief technology officer – a digital platform used for
data collection where information could be accessed in real time.

2.3. Outcome Variables
There are two dependent/outcome variables. The first (migration status) was a binary outcome (migrant vs. nonmigrant).
In the YEMESA project, internal migration was captured by asking for the district in which the respondent was born. With
such a question data pertaining to whether a person migrated or not was obtained. The second dependent variable was the
current employment status of the migrants at the time of the interview with three outcomes: Employed, self-employed,
and unemployed.

2.4. Explanatory Variables

The explanatory or independent variables were the individual-level factors which influenced migration status. These factors
were divided into demographic and socioeconomic factors. The demographic factors included age, sex, marital status,
and number of children. The socioeconomic factors considered were urban or rural residence, region, educational level,
reasons for migration, and social networks. In the study, social network was defined as a network of social interactions
and personal relationships with friends, relatives or coworkers. Reasons for migration included economic reasons for a
better job, new business or higher income in destination; social reasons for better or new relations with relatives, friends
or partner in the destination; and forced reasons where an individual unwillingly left the place of origin due to factors such
as wars, insecurity, disease outbreak, hunger, and other environmental disasters.

2.5. Analytical Strategies
The univariate analysis involved the use of frequency distributions for the explanatory variables both demographic
and socioeconomic factors as well as the outcome variables. Multivariable binary logit model was done to test which
explanatory variables affected migration status among all the respondents. The multinomial logit model was adopted to
assess whether migration status affected employment in the presence of other covariates and what factors associated with
employment. The model used relative risk ratios (RRR), which mean that for a unit change in the predictor variable, the
logit of an outcome relative to the reference group is expected to change by its respective parameter estimate given that
the variables in the model are held constant. For this case, all variables included at bivariate analysis were included in the
multinomial logit model. In the model, un-employed response was chosen as the comparison category.
3. Results

Results in Table 1 show that 75.3% of the sample were migrants, 46.0% were self-employed and 40.9% were employed.
A majority (40.2%) of the sample were aged 23-27, 55.6% were males and 76.1% were coming from rural areas. Almost

International Journal of Population Studies | 2019, Volume 5, Issue 1 41