shown in Appendix C.
Finally, hypothesis testing is conducted to evaluate whether
the three effects contributing to the differences in poverty between
Taiwan and another country are statistically significant. Following
the method developed by Wang, Rahman, Siegal, and Fisher
(2000), I generate 100 resamples from the original sample of each
country using the bootstrapping normal approximation method
and replicate decomposition to estimate the standard deviations of
decomposition analyses. According to Wang et al. (2000), the
standard deviations of the sampling distribution may be close to
the standard errors of the component effects with 50-200
resamples (Efron & Tibshirani, 1986). Each resample size accounts
for 20% of the original sample. Then I conduct 100 decomposition
analyses for each country (100 x 22), estimate standard deviations,
and conduct t-tests.
A. Household Composition
Table 1 shows household composition patterns by country.
Consistent with the living arrangements of young adults revealed in
previous studies, more than 60% of young Scandinavian adults
lives alone or with their spouses, although about 45% of young
adults stays with their parents. Meanwhile, the majority
of young respondents in Spain, East Asia, and post-socialist
countries lives with their parents. With 62% of respondents
residing with their coupled parents, the household composition of
Taiwan seems to be the most economically beneficial for young
adults. In addition, many young people live in households with
their grandparents, other relatives, or non-family household
members (i.e. category “others”).