Page 125 - IJPS-11-6

International Journal of
Population Studies Internal migration in Indonesia

Technically, we first defined the migration status for statistical associations rather than causal relationships.
each individual within the age range of 12–50 years. Interpretations of the results should therefore be
Migration status was classified into three categories: “stay,” understood as correlational.
“rural migration,” and “urban migration.” Individuals
who had not migrated were categorized as “stay.” “Rural 3. Key findings
migration” referred to those who moved to another rural 3.1. Describing the Indonesian rural-urban
district, while “urban migration” described those who migration pattern
relocated to an urban district. In addition, we distinguished
between the stages of movement (first, second, third, Table 2 presents an in-depth examination of internal
and more) to capture the sequence of migration events migration patterns in Indonesia, which commences
(Bernard, 2022b). Using the optimal matching algorithm, in rural regions and encompasses various movement
we calculated dissimilarities between each migration sequences between rural and urban destinations. These
sequence (Needleman & Wunsch, 1970) and then grouped sequences vary in the number of movements and types
similar migration sequences into specific clusters using of destinations, offering a comprehensive perspective on
hierarchical cluster analysis (Piccarreta & Billari, 2007), migration dynamics. Among single movement patterns,
with a ward linkage approach (Brzinsky-Fay et al., 2006). migration within rural areas was the most prevalent,
comprising 642 cases (19.93%), followed by migration
In the second stage, we employed a multinomial
logistic regression model to identify the sociodemographic from rural to urban areas, with 620 cases (19.25%). These
figures indicate that despite significant urbanization, a
characteristics associated with each migration cluster. This
modeling approach was appropriate because the dependent substantial portion of the population remains in rural
variable, the migration trajectory cluster identified in the areas, suggesting that pull factors are not exclusive to urban
first stage, was categorical and nominal with more than areas and that strong sociocultural ties to rural origins may
two unordered outcomes. Multinomial logistic regression influence migration decisions (Debray & Ruyssen, 2023;
allowed us to examine the statistical association between He et al., 2023). Internal migration from rural to rural is
multiple independent variables with the probability of primarily driven by economic opportunities, improved
belonging to each cluster. The model used in this second living conditions, and social networks (Kumar, 2020; Liang
stage is shown in Equation (I), and the variables involved et al., 2002; Lucas, 2015; Shen et al., 2024; Sugiyarto et al.,
in the model are listed in Table 1. 2019). Bazzi et al. (2016) revealed that transmigration to
obtain access to new land has caused the transmigration of
Y = β + β gender + n mar + β edu + u gender.edu + n villagers from Java and Bali to other villages outside. This
5
4
0
ij
1
ij
2
ij
ij
3
ij
age group + o welfare + l island + l motives + t migrate is believed to be one of the causes of the high percentage
ij
8
ij
9
ij
ij
6
7
with ij (I) of migration from rural to rural areas in Indonesia. These
Given the observational nature of the IFLS data and the patterns resonate with the life-course perspective, which
use of a non-experimental design, the analysis identifies suggests that migration decisions are rarely isolated
Table 1. The variables involved in the multinomial logistic regression model
Variables Definitions and coding
Y The likelihood of migrant i being included in migration cluster j
ij
β β₀ denotes the constant term, while β₁–β₉ represent the regression coefficients that measure the relationship between the
independent (predictor) variables and the dependent (outcome) variable with more than two categories
gender ij Gender of migrant i in migration cluster j, 0=male, 1=female
mar Marital status of migrant i in migration cluster j, 0=unmarried, 1=married, 2=ever married
ij
edu Level of education of migrant i in migration cluster j, 0=low (elementary school and lower), 1=middle (junior high school or
ij
equivalent), 2=high (senior high school and higher)
gender.edu Interaction of gender and education level of migrant i in migration cluster j, 0=others, 1=female with high education level
ij
age group ij Age group of migrant i in migration cluster j, 0=<20 year, 1=20–30 year, 2=31–44 year, 3=45 year+
welfare ij Welfare status of migrant households of migrant i in migration cluster j, 0=poor, 1=near poor, 2=not poor
island Island of origin of migrant i in migration cluster j, 0=other, 1=Java, 2=Sumatera
ij
motives The migration motive of migrant i in cluster j, 0=others, 1=work, 2=education, 3=marriage, 4=migration with family, 5=to be closer
ij
with family, 6=pregnancy/other family reasons
migrate with With whom migrants i in migration cluster j migrate, 0=alone, 1=others
ij
Volume 11 Issue 6 (2025) 119 https://doi.org/10.36922/IJPS025190084