Page 43 - IJPS-9-1

International Journal of
Population Studies Transportation assimilation in Hong Kong

Table 2. (Continued)
Full sample Non‑minibus Minibus users Difference
(n=169,766) users (n=145,770) (n=23,996)
Mean/% (SD) Mean/% (SD) Mean/% (SD)
Skilled agricultural and fishery workers 0.11 0.10 0.22
Craft and related workers 9.66 9.75 9.13
Plant and machine operators and assemblers 5.56 5.65 5.00
Elementary occupations 23.60 23.64 23.35
Others 0.02 0.03 0.01
Note: ***p<0.001 (two-tailed tests).

immigrants arriving in or after 1997, East Asian, South 1988). We followed Shevky and Bell’s method to obtain the
Asian, South-east Asian, white, and others. As provided in index of interaction, in which P* represents the probability
the existing literature on immigration issues in Hong Kong, of a randomly selected member of a particular ethnic
1997, the year of the Handover, marks a special watershed group meeting (in his census tract) another member of the
that differentiates a Crown Colony of the United Kingdom same ethnic group (Bell, 1954; Shevky & Bell, 1955). The
from a Special Administrative Region under the People’s * 1 k a i 2
Republic of China. The different sociopolitical contexts equation for P* is P A b . We obtained P* by
before and after the Handover may result in different social i1 i
sentiments toward Chinese immigrants, especially those dividing the number of coethnic immigrants living in the

from mainland China. Other ethnic groups are broader same residential district by the total number of residents in
that district in each census year.
categories compared to the Chinese. East Asian includes
Japanese and Korean. South Asians refer to Bangladeshi, In addition to the independent variables mentioned
Indian, Nepalese, Pakistani, and Sri Lankan. South-east above, we also control a list of covariates, including
Asian includes Filipino, Indonesian, Thai, and Vietnamese. educational attainment, gender, life stages, monthly
Mixed races, blacks, other Asians, and others are all income, industry, occupation, residential district, working
categorized under others. As shown in Table 2, in the full district, year dummies, and several district-year dummies.
sample, 67.43% are Chinese immigrants who migrated To control for residential district-fixed effects, for instance,
before 1997, 23.76% are Chinese immigrants who migrated to compare the residents living in the same district, we
in or after 1997, 0.80% are East Asians, 2.18% are South included 24 residential district dummies in our analysis.
Asians, 2.25% are South-east Asians, 2.70% are whites, Similarly, to control for working district-fixed effects, such
and 0.89% are others. In the minibus users (N = 23,996), as comparing the passengers working in the same district,
69.65% are Chinese who migrated before 1997, 24.10% we included 24 working district dummies throughout the
are Chinese who migrated in or after 1997, 0.48% are East models. We also included the interactions between place
Asians, 1.25% are South Asians, 1.79% are South-east of residence and place of work as well as the interactions
Asians, 1.97% are whites, and 0.76% are others. between census year and residential/working district.

2.2.4. Cantonese ability 2.3. Methodology
The ability to speak Cantonese is a binary variable, with 1 We applied linear probability models (LPM) with the
referring to yes, and 0 referring to no. In our full sample, as binary variable, such as the likelihood of taking the
shown in Table 2, those who indicated the ability to speak minibus as the dependent variable. We started with a
Cantonese consisted of 92.18% of all the individuals. This base model (Model 1), in which we only included our
percentage is 94.72% for the minibus subsample, which is key independent variables by themselves. In this way, the
slightly higher than the full sample. base model captures the static effects of individual-level
and district-level variables on the probability of taking a
2.2.5. Index of interaction minibus. From Model 2 to Model 5, we added interaction
The index of interaction aims to measure the possibility of terms between certain key independent variables and the
meeting other co-ethnic immigrants in one’s residential years of duration in Hong Kong one by one. In Model 4
area. The index of interaction depicts the dimension of and Model 5, we included three-way interaction terms
exposure, a rather standard indicator to measure the level to explore inter-ethnic differences in taking minibuses
of spatial segregation by sociologists (Massey & Denton, along with different lengths of stay in Hong Kong. Model

Volume 9 Issue 1 (2023) 37 https://doi.org/10.36922/ijps.0386