Page 45 - IJPS-9-1

International Journal of
Population Studies Transportation assimilation in Hong Kong

minibus routes) that may affect minibus ridership (Xu, models, we only performed within-district comparisons.
2018). Our analysis suggests that within each residential district,
We employed LPM with robust standard errors the population proportion of one’s ethnic group has no
correcting heteroskedasticity to test our hypotheses. We association with the likelihood of taking the minibus.
acknowledge that our dependent variable is of binary Second, this study depends on a context different from
nature, which generally should rate logistic regression prior research. Unlike relatively sparsely populated areas,
with higher preference. However, since we will incorporate such as cities in Europe and North America, in the previous
several interaction terms in our full model, the statistical research, Hong Kong is a highly compact city with a high
interaction in models with categorical outcomes may result population density.
in inaccurate P-values (Mustillo et al., 2018). In addition, Model 2 incorporates the interactions between
logistic regressions would eliminate the observations certain individual characteristics and the duration of
in which no minibus takers were present, which might residence in Hong Kong to capture the “dynamic mode”
unnecessarily reduce our sample size. Therefore, to avoid of transportation assimilation. Throughout the models,
any inaccurate interpretation and maintain the high continuous variables are centered around their means, such
statistical power of our analysis, we simply relied on linear as duration of residence in Hong Kong, age at migration,
probability models. Nevertheless, for robustness check, logged personal income, and others. Since the interaction
please refer to Table S1 in the Online Supplementary between years of staying in Hong Kong and age at migration
Materials for the results from logic regressions, in which is statistically insignificant, we cannot suggest that age
all major conclusions stand. at migration moderates the effect of residence duration

To ensure that our results are robust, we also carried among all immigrants. The main effect of the ethnic
out a series of analyses, including only immigrants whose group indicates the probability of taking the minibus by
main mode of transportation is a minibus. We reran the ethnic group when the duration of residence in Hong
Model 1, Model 2, Model 4, and Model 5, and all the major Kong is equal to its mean (i.e., 15 years) and holding other
conclusions also stand. Please find the results in Table S2 in factors constant. The interactions between the duration
the Online Supplementary Materials as well. of residence in Hong Kong and ethnic group show that
compared to Chinese immigrating in or after 1997, South-
3. Results east Asians are significantly less likely to take minibus with
longer years of staying in Hong Kong.
Tables 3 and 4 present the results of our five regression
models, each of which has controlled for a collection of Model 3 adds three-way interactions among duration
sociodemographic characteristics, year-specific fixed of residence in Hong Kong, age at migration, and ethnic
effects, and location-specific fixed effects. Model 1 is the group, as well as interactions between age at migration and
base model that does not include any interaction terms. ethnic group. The main effect of duration of residence in
Model 1 suggests that the duration of residence in Hong Hong Kong reveals that those staying in Hong Kong for
Kong is positively associated with the likelihood of taking more years are more likely to take the minibus. The main
the minibus. By contrast, a negative relationship exists effect of age at migration shows that those arriving in
between age at migration and the probability of taking the Hong Kong at a younger age are more likely to take the
minibus. In regard to the role of ethnic groups, we found minibus. Since we used Chinese immigrating in or after
ethnic differences in the likelihood of taking the minibus 1997 as the reference group for ethnicity, the statistically
to be statistically significant. Specifically, South Asians are significant interaction between years of living in Hong
significantly less likely to take the minibus than Chinese Kong and age at arrival suggests that, for recent Chinese
immigrating in or after the year of the Handover (i.e., immigrants, the effect of duration of residence in Hong
1997). The latter has a slightly but insignificantly lower Kong on minibus taking is more pronounced for those
chance of taking minibus than other ethnic groups except immigrating at a younger age. Figure 2 summarizes the
for South Asians. In addition, Cantonese ability is another association between years of living in Hong Kong and the
powerful determinant of immigrants’ minibus taking. predicted probability of taking the minibus by age group
Inconsistent with the findings of prior research that reveal at migration among Chinese immigrating in or after 1997.
a negative link between ethnic density and transportation Recent Chinese immigrants arriving at a younger age have
assimilation (Liu & Painter, 2012), our analysis indicates a higher initial probability of taking the minibus than
no substantial effect of district-level ethnic density. There those arriving at an older age, and such a difference grows
are two possible explanations for such a result. First, with more years of staying in Hong Kong. Similar to Model
since residential district dummies are included in all the 2, Model 3 also confirms South-east Asians’ steeper decline

Volume 9 Issue 1 (2023) 39 https://doi.org/10.36922/ijps.0386