2011. Vol.2, No.2, 63-70
Copyright © 2011 SciRes. DOI:10.4236/psych.2011.22011
Self-Construal and Subjective Wellbeing in Two Ethnic
Communities in Singapore
Weining C. Chang1, Mohd Maliki Bin Osman2, Eddie M. W. Tong3, Daphne Tan3
1Division of Psychology, Nanyang Technological University, Singapore;
2Member of the Parliament, Republic of Singapore, Singapore;
3Department of Psychology, National University of Singapore, Singapore.
Received October 13th, 2010; revised December 23rd, 2010; accepted December 30th, 2010.
The study reported here addressed two questions concerning the applicability to other collectivist cultures of a
model developed on the basis of socially defined self-construals and their consequences for subjective wellbeing
(Kwan, Bond, & Singelis, 1997). In contrast to the social centrality of the Singaporean Chinese, Islam enjoys
centrality in the Malay community. Is subjective well-being in these two cultures attributable to independent or
interdependent self-construal? Do parallel paths to subjective wellbeing originate from these two forms of
self-construal? Are the respective mediators, self-esteem and relationship harmony, functioning in the same way
in these different communities? Two hundred and eighty-six participants (121 Malays, including 49 females and
72 males, and 165 Chinese, including 62 males and 103 females, average age, 18.52) were drawn from three ter-
tiary technical training institutes in Singapore. Independent and interdependent self-construal, Rosenberg’s
Self-esteem Scale (SE), the Relationship Harmony Scale (RH) (Kwan, et al. 1997) and the Satisfaction with Life
Scale (SL) (Diener, Emmons. Larsen, & Griffins, 1985) were administered to test the model. For the Chinese,
the best-fit model required a path between the two forms of self-construal, confirming their overlapping nature.
For the Malays, a radically different model fits the data: relationship harmony no longer functions as a mediator
but as an independent contributor in its own right. These results raise questions about the meaning and conse-
quences of socially defined self-construal in different collectivist cultures, and the path between the self and
Keywords: Subjective Wellbeing, Culture, Self, Chinese, Malays
Subjective wellbeing has been the focus of intense research
attention (Diener, 1984; Leung & Leung, 1992; Diener & Die-
ner, 1995). Recent cross-cultural research has provided pre-
liminary evidence for culture-specific correlates of life satisfac-
tion (for instance, Diener & Diener, 1995). Specifically, Kwan,
Bond and Singelis (1997) found that there are two conceptually
coherent and psychologically meaningful routes to life satisfac-
tion traceable to independent and interdependent self-construal.
Independent self-construal reaches life satisfaction through
mediation by individual self-esteem, while interdependent
self-construal reaches life satisfaction through mediation by
relationship harmony. These conclusions are acceptable if one
allows the assumption that the independent and the interde-
pendent variables are 1) orthogonal to each other and are 2)
discrete constructs. In other words, there are no psychological
overlaps between these two construals of the self. In cultures
where such assumptions cannot be met does the model still
hold? In terms of the mediating variables, this model relies on
the assumptions that 3a) individual self-esteem is conceptually
and psychologically related to the independent self construal
and 3b) relationship harmony is conceptually and psychologi-
cally related to interdependent self-construal. Could these as-
sumptions be met in other collectivist cultures?
Cross-culturally, relationship harmony and individual self-
esteem are both important for Chinese and Americans in terms
of life satisfaction, but the mediation effect of relationship
harmony on life satisfaction is greater, for instance, in Hong
Kong than in the United States (Kwan et al., 1997). This result
was interpreted in terms of the greater importance of interde-
pendent self-construal in Hong Kong, a Chinese community
and presumably collectivist, than in the United States, which is
presumably less collectivist. What might be the role of the self,
especially the culturally conditioned constructs of the self, in
the individual’s life-satisfaction or subjective wellbeing in an-
other Asian collectivist context remains to be investigated. It is
our contention that collectivism is not the only dimension of
cultures in Asia. In different Asian cultures, though they are
mostly collectivist, there are other dimensions of culture that
might interact with the social emphasis of collectivism to pro-
vide different meanings and consequences of socially defined
self-construal. The culture of the Chinese is a socio-centric
culture where interpersonal relationships play a central role in a
person’s daily life (Huang, 1999). Though social interdepend-
ence is a shared feature of many Asian cultures, some Asian
cultures structure themselves on the basis of religious beliefs,
such as the Muslims in many different countries in Asia. Can
their life satisfaction be explained by a model constructed on
the notion of human relationships?
Singapore has a Chinese majority (76%) but has a substantial
Malay minority (12%). Both the Malays and the Chinese in
Singapore have collectivist cultures (Chang, in press). There is
both divergence and similarity between these two ethnic groups.
W. C. CHANG ET AL.
In terms of cultural heritage, Singaporean Malays derive their
emphasis on the collective from the traditional cultures of the
indigenous Malay and Islam (Zarina, Isnis, & Chang, 2003);
the Chinese derive their emphasis on the collective from Con-
fucianism or its popular modern forms (Chang, Wong, & Koh,
2003). It can be hypothesized that the differences and similari-
ties between these two cultures would lead to differences and
similarities in their self-construal and life satisfaction.
Our past studies have found that in Singapore, instead of
forming two discrete constructs, independent and interdepend-
ent self-construal overlap when measured as individual differ-
ences of disposition variables. In daily social interactions, the
manifestation of the two forms of self construal in Singaporean
Chinese form a continuum from independence to increasing
interdependence according to the level of intimacy that the
person has with the other person (Chang, 2001). Much less is
known about the self-construal of Malays. Malay cultural heri-
tage has two distinct features: 1) traditional indigenous Malay
culture that emphasizes the interdependence between the self
and social others, and 2) Islam, where God is the creator of all
things including the self. Within this religion-centric culture,
the self might take on additional meaning compared to the so-
cially defined independent and interdependent self. The indi-
vidual’s life satisfaction is often attributed to one’s belief in
God (Shahiraa & Chang, 2004). Within this context, what im-
portance and what roles do the mediating variables, individual
self-esteem and relationship harmony, play in providing life
satisfaction? Is the model, based on the dichotomy of inde-
pendent and interdependent self construal, applicable to cul-
tures where the two forms of construal might overlap? What
might be the nature of the self-construal and its consequences in
a collectivist but highly religious culture?
Pathways to Life Satisfaction: Self-Construal,
Self-Esteem an d R el a tionship Harmony
In terms of the self and collectivist cultures, Kwan et al.
(1997) have shown that self-esteem and relationship harmony
mediate the relationship between self-construal (SC) and sub-
jective wellbeing (SWB). Relationship harmony was found to
be a mediator between interdependent self-construal and sub-
jective wellbeing. This path is logical and psychologically
meaningful only when the assumption that relationship har-
mony is derived from interdependent self-construal is met.
In Singapore, independent and interdependent self-construal
have been found to be positively correlated with each other
(Tong, Chang, & Koh, 2003). In our preliminary research on
the predictors of self-esteem, we also found that relationship
harmony is an important contributor to Singaporeans’ feelings
of self-esteem (Tong et al., 2003). That relationship suggests
that the individual’s self esteem might be drawn from both
independent and interdependent self-construal in this predomi-
nantly Chinese community. This finding of the intertwined
nature of the two forms of self-construal and their relationships
with relationship harmony and self-esteem further cast doubt on
the applicability of the Kwan model in Singapore.
In cultures where religion has a central role, such as that of
the Malays in Singapore, and where the self and the psycho-
logical constructs related to the self such as self-esteem and
self-other relationships are seen to be God-given, (Brenner,
1996; Shahiraa & Chang, 2003), what might the relationship be
between self-construal and the individual’s life satisfaction?
Hypothesizing different relationships between these predic-
tors of subjective wellbeing in different collectivist cultures, we
tested the Kwan model separately within two samples (i.e.,
Chinese and Malays) in Singapore. Attention was specifically
focused on the internal structure of self-construal in the Chinese
and the Malay communities.
Participants were drawn from three polytechnic institutes in
Singapore with a stratified random sampling method to derive
two matching samples of Malays and Chinese. As Malays con-
stitute only 12% of the general population, they were over-
sampled, whereas the Chinese, constituting 76% of the popula-
tion, were under-sampled. Two hundred and eighty six students
from three polytechnics and one technological-training institute
participated in the study. They comprised 165 Chinese (62
males and 103 females) and 121 Malays (49 males and 72 fe-
males). The mean age of the participants was 18.52 (SD =
The 10-item Rosenberg Self-Esteem Scale (RSE S: Rosenberg,
1965) was used to assess global self-esteem (SE). The partici-
pants responded to each item on a 5-point scale ranging from
“1” (strongly disagree) to “5” (strongly agree). The Cronbach
alphas for the overall sample, Chinese sample and Malay sam-
ple were .75, .77, and .72 respectively.
The Interpersonal Relationship Harmony Inventory (IRHI:
Kwan & Bond, 1996) was used to assess perceived relationship
harmony (RH). Participants were instructed to think (for about
5 minutes) of five different dyadic relationships and to rate each
of these relationships on four items. For the first two items,
participants indicated the nature of the relationship (e.g. father,
friend) and the gender of the target in the relationship. The last
two items asked for the participant’s rating of the relationship
in terms of perceived degree of harmony and the contributions
of the relationship to one’s happiness on a 4-point scale ranging
from “1” (very low) to “4” (high). A relationshi p harmony index
for each participant was computed by multiplying the harmony
score of each relationship with its contribution-to-happiness
score and then summing the resulting values across the five
Singelis’s (1994) Self-Construal Scale (SCS) was employed
to assess the extent of interdependent and independent SC.
Twelve items in the SCS were designed to measure interde-
pendent self-construal and another 12 to measure independent
self-construal. Participants indicated the extent to which they
agreed (or did not agree) with each of the items on a 5-point
scale that ranged from “1” (strongly disagree) to “2” (strongly
agree). For independent SC, the alpha values for the overall
sample, Chinese, and Malays were .65 each. For interdependent
SC, the internal reliabilities for the whole sample and the Chi-
nese were .53 and .71, respectively. However, one item (#16) “I
am comfortable being singled out for praise or rewards” had to
be removed from the Malay data to achieve an acceptable reli-
ability of .63. It is our understanding that within Malay culture,
W. C. CHANG ET AL. 65
being humble is an important virtue; this item was considered
inappropriate by many Malays.
Perception of satisfaction with one’s life was assessed by the
Satisfaction with Life Scale (SWLS: Diener, Emmons, Larsen,
& Griffin, 1985). Participants responded on a 5-point scale
ranging from “1” (strongly disagree) to “5” (strongly agree) on
each of the five items that measured the extent to which they
were satisfied with their lives (e.g. “I am satisfied with my
life”). Previous studies had demonstrated strong reliabilities and
convergence validity of the SWLS (c.f., Larsen, Diener, &
Emmons, 1985; Pavot & Diener, 1993). In this study, the alpha
value the overall sample was 0.73, while those of the Chinese
and Malays were 0.75 and 0.69 respectively.
The participants completed a questionnaire that contained the
abovementioned scales arranged in random order. Permission to
conduct the study in the four institutions was given by their
respective administrative departments. The student participation
was voluntary and anonymous.
All participants were carefully briefed by the researchers
about completing the questionnaires and the confidentiality of
the replies. The participants were also assured of anonymity in
their responses. No difficulties in answering the questionnaires
were reported; each session took approximately 30 minutes.
Table 1 presents the means and standard deviations of the
seven variables examined. One clear result is that the scores for
Malays on these variables were higher than those of the Chi-
nese; t-tests on each variable verified that the Malays were
significantly higher than the Chinese on every variable. How-
ever, any conclusion of real ethnic difference at this point is
unwarranted because there is the possibility of culturally based
response bias, or response mode (Van de Vijver & Leung,
1997). We thus de signed a way to adjust for this potential bias.
Assuming that the differential response bias is evenly dis-
tributed across measures of all psychological constructs, we
constructed a correction term (Chang, 2003) by averaging the
between-group differences of all items of all measures used in
the study. The between-group difference of each construct was
the tested against this grand mean difference. With this analysis,
we found that none of the between-group differences reached
Independent and Interdepe nde nt Se lf-Construal in
Submitting the 24 items of the SCS to factor analysis with
varimax rotation, we found an acceptable structure, as pre-
sented in Table 2, with the interdependent and independent
items meaningfully loaded onto two factors. However, as we
predicted, the inter-factor correlation is very high, –0.592, sug-
gesting that in Singapore these two forms of self-construal are
not independent of each other but are highly overlapping with
the interdependent self more salient than the independent self.
Similar results were obtained when the analysis was repeated
with each of the ethnic groups. We again tested the two factor
structures by constraining the solution to two factors. As can be
seen in Table 2, the items fell into the two meaningful groups
of Factor 1 - Interdependent Self-construal, and Factor 2 - In-
dependent Self-construal. However, the two factors accounted
for only 24.59%and 24.56% of the total variance for the Chi-
nese and the Malay data, respectively. What is striking about
this result is the highly intertwined nature of these two factors.
For the Malays and the Chinese, the inter-factor correlations are
–0.566, and –0.576, respectively.
These results suggested that although the independent and
interdependent dichotomy might not be most applicable to
self-construal, the exact nature of self-construal in different
Asian communities might be more complicated than the
two-way division. More importantly, the independent and in-
terdependent forms of self construal were highly inter-related in
both ethnic samples. They are certainly not discrete variables
but are continuations of each other: that is, the relationship
between the independent and the interdependent aspects of the
self might be interdependent or reciprocal. This would have
implications in the psychological consequences predicted by
these forms of self-construal (Chang, 2001).
Subjective Wellbeing (SWB)
The results of factor analyses reveal a one-factor solution for
the overall sample, the Chinese, and the Malays, accounting for
49.8%, 51.9%, and 45.6% of the respective data. Thus, it ap-
pears that the factor structure for SWB is replicated among our
Table 3 presents the correlations among the variables. Across
the two ethnic groups, SWB positively correlated with the four
predictor variables. The strongest association was found be-
tween SWB and SE, consistent with research showing that
self-esteem is a strong predictor of life satisfaction (Diener,
1984; Diener & Di e n e r , 1995 ) .
However, more interestingly, RH correlated almost as highly
as SE with SWB. This is consistent with the findings among
Hong Kong participants in the Kwan et al. (1997) study show-
ing that in a collectivistic culture like that of the current sample,
RH is as strong a predictor of SWB as SE. Furthermore, con-
sistent with previous findings (Singelis, Bond, Lai, & Sharkey,
Means and standard deviations of five variables in the overall sample,
Chinese, and Malays.
Chinese Malay Total
M SD M SD M SD
Wellbeing 15.283.57 16.34 3.28 15.733.48
Self-esteem 32.965.15 34.17 5.01 33.475.12
Harmony 77.7024.9588.80 20.52 82.3023.81
Self-construal 3.49 0.42 3.51 0.43 3.69 0.41
Self-construal 3.61 0.43 3.85 0.70 3.73 0.53
. Calculated as per item mean.
W. C. CHANG ET AL.
Factor structure of Singaporea n se l f-construal.
Whole ChineseMalay Whole ChineseMalay
6. I will sacrifice my self-i nterest for the benefit of the group. 0.656 0.487 0.68
5. I respect p eople who are modest about themselves. 0.573 0.437 0.616
9. It is important to me to respect decisions made b y the group. 0.561 0.504 0.7
3. My happiness depends on the happiness of those around me. 0.545 0.566 0.551
10. I will stay in a group if they need me, even when I am not happy with
the group. 0.545 0.463 0.51
7. I often have the fee li ng th at my relationships w i th others are more
important than my own. 0.526 0.453 0.632
2. It is important for me to maintain harmony w it hin my group. 0.483 0.521 0.495
4. I would of f er my seat in a bus to my prof essor. 0.475 0.437 0.448
1. I have res pect for the authority figures with whom I interact. 0.46 0.616 0.265
12. Even when I stron gly disagre e with group members I avoid argum ents. 0.319 0.201 0.325
8. I should take into consideration my parents’ advice when making educa-
tion/ career plans. 0.237 0.46 0.239
14. Speaki n g up during class is not a problem for me. 0.61 0.592 0.653
19. I act the same way no matter who I am with. 0.55 0.49 0.62
21. I prefer to be direct and forth r ight when dealing wit h p eople. 0.544 0.573 0.579
22. I enjoy be ing unique and different from others in many respects. 0.504 0.526 0.407
17. I am the same person at home and at school. 0.482 0.404 0.533
23. My personal identity indepe n dent of others is very im por tant. 0.421 0.316 0.304
13. I’d rather say “no” directly, t han risk bein g misunderstood. 0.41 0.476 0.336
18. Being abl e to take care of myself is a primary concern for me. 0.404 0.35 0.376
16. I am comfortable with being singled out for praise or rewards. 0.403 0.542 0.222
20. I feel comfortable u si ng someone’s first name soon after I meet t he m even
when they are much older than I am. 0.342 0.328 0.476
15. Having a lively imagination is important to me. 0.328 0.358 0.311
24. I value being in good health above everything. 0.309 0.148 0.458
11. If my brot her or sister fa i ls, I feel r esponsible. 0.198 0.12 0.234
1995), SE correlated with independent SC but not with inter-
dependent SC. However, RH correlated positively with both
independent and interdependent SC in our sample, which is
similar to what Kwan et al. (1997) found in their Hong Kong
The lower two sections of Table 3 present the inter-correla-
tions among the 5 variables according to the ethnic groups. For
the Chinese, SWB was positively correlated with SE, RH, and
interdependent SC. Again, both SE and RH were equally strong
predictors of SWB. Interestingly, among the Chinese, SE not
only correlated positively with independent SC, but also with
interdependent SC. RH was positively associated with both
For the Malays, SWB correlated positively with all variables
except interdependent SC. In fact, the interdependent SC of the
Malays only correlated with independent SC, and not with any
other variables included in this model.
We first tested the applicability of the Kwan, et al. (1997)
model to the Singaporean context. Then we sought to delineate
a model that would best fit Singapore. To this end, we first
performed a series of mediation analyses procedures outlined
by Baron and Kenny (1986) to deduce the pathways leading
from each form of self-construal to SWB. In doing so, we were
not only able to test the mediation process proposed by Kwan et
al. (1997) (see Figure 1), but also determine whether there
could be “cross-mediation”: that is, whether RH and SE also
W. C. CHANG ET AL. 67
Intercorrelations between five variables in the overall sample, Chinese,
1 2 3 4 5
Independe nt Self-con strual 0.15**0.37** 0.30**
Interdepe ndent Self -c ons trua l 0 . 13 *0 . 10 0.24** 0.27**
Relationshi p Harmony 0.31**0.29**
Independe nt Self-con strual 0.11 0.34** 0.37**
Interdepe ndent Self -c ons trua l 0.25**0. 19 * 0.33** 0.27**
Relationship Harmony 0.23*0.08
Independe nt Self-construal 0.18*0.38** 0.17
Interdepe nde nt Sel f -cons t rual 0.00 0.01 0.13 0.27**
** p < .01 ; * p < .05
All path coefficients are significant at p = .05.
The Kwan et al. model.
mediate the effect on SWB by independent SC and interdepend-
ent SC respectively. This procedure allowed us to construct a
model that may provide a more precise explanation of our data.
Section 1: Fitting of the Kwan et al. model. The result of
the analysis revealed only a moderate fit of the Kwan, et al.
model to the data of the overall sample. 2 (5) = 25.75, p < .001,
RMSEA = 0.12, GFI = 0.97, NFI = 0.84, CFI = 0.86, and IFI =
0.86. The fit was good but not excellent because of the rela-
tively large RMSEA.
Chinese. The same analysis was conducted on the Chinese
sample. Again the results showed a poor fit of the Kwan, et al.
model to the Chinese data: 2 (5) = 29.10, p < .001, RMSEA =
0.17, GFI = 0.94, NFI = 0.74, CFI = 0.76, and IFI = 0.77. The
fit was good but not excellent due to the relatively large
Malays. In contrast, when the data for the Malays were
tested, the results revealed an excellent fit, 2 (5) = 5.59, p
= .35, RMSEA = 0.032, GFI = 0.98, NFI = 0.89, CFI = 0.99,
and IFI = 0.99. However, the path between interdependent SC
and relationship harmony was non-significant.
Additional analysis. We further tested this ethnic difference
by using the multi-sample option in LISREL that allows com-
parison of models across independent groups. Consistent with
the foregoing results, this test revealed a poor fit, 2 (19) =
63.76.10, p < .001, RMSEA = 0.091, GFI = 0.96, NFI = 0.66,
CFI = 0.73, and IFI = 0.73. This indicated that the Kwan et al.
(1997) model failed to fit the two samples equally well.
Section 2: In search of a Singaporean model - mediation
analysis. Kwan et al. (1997) argued that the relationship be-
tween independent SC and SWB and that between interde-
pendent SC and SWB were mediated by SE and RH, respec-
tively. We sought to extend this argument by testing the extent
to which the relationship of SWB with each form of SC was
mediated by both SE and RH. Thus, we performed a series of
mediation analyses. We first investigated the extent to which
the effect of independent SC on SWB is mediated by both SE
and RH. Subsequently we performed the same test on the rela-
tionship between interdependent SC and SWB.
We performed four sets of mediation analyses. The first two
analyses pertained to the effect of independent SC (the predic-
tor) on SWB (the criterion), one analysis examined the mediat-
ing role of SE, and one examined that of RH. The next two sets
of analyses pertained to the effect of interdependent SC (the
predictor) on SWB as mediated by SE and SH.
Independent SC on SWB. As can be seen in Table 3, SWB
was significantly associated with independent SC. Additionally,
SWB was significantly related to SE. As the two pre-requisites
of mediation analysis were met, SWB was regressed simulta-
neously onto independent SC and SE. The relationship between
independent SC and life satisfaction was reduced to being
non-significant, = .03, t (286) = 0.56, p > .05, whereas SE
remained a significant predictor, = .32, t (286) = 5.33, p < .01,
thus demonstrating a total mediation. The procedure was re-
peated with RH as the mediator. The relationship between in-
dependent SC, and SWB became non-significant, = .07, t
(286) = 1.21, p > .05, while that between RH and SWB was
significant, = .26, t (286) = 4.38, p < .01, once again demon-
strating total mediation. Thus, the relationship between inde-
pendent SC and SWB in Singapore is not only totally mediated
by SE as sho wn by Kwan et al. (1997), but is also mediated by
RH. In our sample an individual’s self-esteem (SE) and their
relationship harmony (RH) were closely intertwined.
Interdependent SC on SWB. The procedure was then re-
peated with the relationship between interdependent SC and
SWB. First, as can be seen in Table 3, interdependent SC was
related to SWB, thus fulfilling the first prerequisite of media-
tion analysis. Interdependent SC was associated with RH but
not SE. This indicates that only RH could serve as the mediator
of the relationship between interdependent SC and SWB. Thus,
SWB was regressed simultaneously onto interdependent SC
and RH. The results revealed a total mediation of RH between
interdependent SC and SWB because RH remained a signifi-
cant predictor of SWB, = .27, t (286) = 4.63, p < .01, whereas
when RH was partialled out, interdependent SC no longer pre-
dicted SWB, = .08, t (286) = 1.31, p ns. Hence, this series of
W. C. CHANG ET AL.
mediation analyses showed that it was RH, and not SE, that
served as the mediator between interdependent SC and SWB.
The Singapore model. The combined results of this series of
mediation analyses thus suggested a model that might better fit
the data from Singapore. This model is presented on Figure 2.
This model is similar to that of Kwan et al., (1997) in that the
relationship between interdependent SC and SWB was medi-
ated by RH. However, it is different from that of Kwan et al.,
(1997) in certain key issues: our analyses revealed that the ef-
fect of independent SC on SWB was not only mediated by SE,
but also by RH.
Testing the Singaporean model. Further testing of the Sin-
gaporean model indicated that it fit the data well: 2 (4) = 6.98,
p = .14, RMSEA = 0.005, GFI = 0.99, NFI = 0.96, CFI = 0.98,
and IFI = 0.98.1 We tested whether our model differed signifi-
cantly from the Kwan et al. model; this comparison yielded the
following results: 2 = 18.77, df = 1, p < .001. Hence, taken
together, the Singaporean model constructed through mediation
analysis fitted the data better than the Kwan et al. (1997)
Section 3: Ethnicity specific models. We repeated the series
of analyses used in Section 2 on the two ethnic samples sepa-
rately with the newly identified Singaporean model. The results
showed that data of the two groups cannot be equally fit by the
Singaporean model: 2 (18) = 44.44, p < .001, RMSEA = 0.072,
GFI = 0.97, CFI = 0.84, and IFI = 0.84. This suggested the
possibility that the data obtained from these two ethnic groups
might be best described by ethnicity specific models.
Chinese model. We first consider a specifically Chinese
Mediation analysis: independent SC on SWB. As seen in Ta-
ble 3, independent SC did not show significant correlation with
SWB for the Chinese data. No mediation analysis was con-
ducted for the effect of independent SC on SWB.
Mediation analysis: interdependent SC on SWB. As the sec-
ond section of Table 2 shows, interdependent SC was signifi-
cantly associated with SWB, SE, and RH. Following this, we
simultaneously regressed SWB onto SE and interdependent SC.
The results indicate partial mediation because, although SE
predicted SWB, = .35, t (165) = 4.76, p < .01, the effect of
interdependent SC on SWB was reduced but remained signifi-
cant, = .19, t (165) = 2.56, p < .05.
Similarly, we regressed SWB onto interdependent SC and
RH to test for any mediating effect of RH. In this case, RH
totally mediated the relationship between interdependent SC
and SWB, because it remained a significant predictor of SWB,
= .28, t (165) = 3.31, p < .001, while the effect of interde-
pendent SC on SWB was reduced to non-significance, = .11, t
(165) = 1.31, p ns.
Structural equation modeling. Therefore, mediation analysis
suggests a Chinese model in which the relationship between
interdependent SC and SWB was mediated by both SE and RH.
Submitting this model to SEM yielded an inadequate fit to the
data, 2 (2) = 14.46, p < .001, RMSEA = 0.20, GFI = 0.96, NFI
= 0.80, CFI = 0.81, and IFI = 0.82 and modification indices
suggested an additional pathway connecting SE and RH. Fur-
ther analysis was conducted to determine the causal direction
between SE and RH. We regressed SWB onto both RH and SE.
The results show that while the relationship between SWB and
SE was not reduced, = .38, t (145) = 5.02, p < .001, the rela-
tionship between SWB and RH was substantially reduced,
while remaining significant, = .20, t (145) = 2.61, p = .01.
This suggests a stronger path from RH to SE than from SE to
Following this, we derived the Chinese model as shown in
Figure 3. SEM testing revealed a good fit of the model to the
data: 2 (1) = 3.63, p = .57, RMSEA = 0.13, GFI = 0.99, NFI =
0.95, CFI = 0.96, and IFI = 0.96. The best fit model and stan-
dardized path coe ffi cients are present ed in Figu re 2.
Malay model. A Malay model can also be derived from the
Mediation analysis: independent SC on SWB. Table 2 indi-
cates that independent SC predicts SWB and SE but not RH.
We regressed SWB ont o inde pendent SC and SE simultaneou sly .
The relationship between SE and SWB remained, = .19, t
(119) = 2.01, p < .05, whereas the effect of independent SC on
SWB was reduced to non-significance, = .12, t (119) = 1.10,
Mediation analysis: interdependent SC on SWB. As Table 3
indicates, interdependent SC did not show significant correla-
tion with SWB in the Malay sample. Thus mediation analysis
was not carried out. RH, however, showed significant correla-
tion with SWB, suggesting an independent contribution of RH
Structural equation modeling. The Malay model derived
from the mediation analysis is one whereby independent SC is
the only form of self-construal that predicts SWB via the me-
diation effect of SE. In addition, as RH also correlated with
SWB, we included a path from RH to SWB. The resulting pat-
tern is shown in Figure 4. Subjecting this model to SEM, it
provided an excellent fit to the data: 2 (2) = 0.63, p = .73,
RMSEA = 0.00, GFI = 1.00, NFI = 0.98, CFI = 1.00, and IFI =
1.04.3 This model is more parsimonious than the Kwan et al.
(1997) model, to which the Malay data also fit. In comparing
the two models, it was found that they did not differ significantly
All path coefficients are significant at p = .05.
Path coefficients for Singaporean model.
All path coefficients are significant at p = .05.
The Singaporean Chinese model .
W. C. CHANG ET AL. 69
All path coefficients are significant at p = .05.
The Singaporean Malay model.
from each other, 2 = 4.96, df = 3, p ns. Thus, the simpler
model as illustrated in Figure 4 was identified as the Malay
We set out to ask whether a model of subjective wellbeing
developed within one collectivist culture is applicable to other
collectivist cultures in Asia, in this case Singaporeans and the
Chinese and Malay subgroups therein. These are groups whose
culture(s) overlap with that of the Hong Kong Chinese in terms
of collectivism - the emphasis on the collective and interper-
sonal relationships. Singaporean Malays are 99% Muslim.
Their religious belief to a great extent provides the basic
worldview on which to construct their self-construal, and their
view of interpersonal relationships and subjective wellbeing.
We found that although self-esteem and relationship harmony
were significant contributors to life satisfaction, the relation-
ships between the predictors, independent and interdependent
self-construal, and the mediator variables, self-esteem and rela-
tionship harmony, seem to work differently in different ethnic
Relationsh ip b e tw e e n Independent and
To begin with, independent and interdependent self-construal
were found to be positively correlated to a significant extent for
the entire Singapore sample and for the Malays and the Chinese
separately. This indicates that the two forms of self-construal
might not be clearly separable in terms of their impacts on the
criterion measures. More importantly, the intertwining might
mean that they could be two sides of the same construct in these
Asian communities (Tong, Chang, & Koh, 2003) rather than
psychologically separate constructs. Seen in this light, the fol-
lowing results will be logically explicable.
Self-Esteem an d R el a tionship Harmony
The two variables, hypothesized to mediate the two forms of
self-construal and subjective wellbeing, were positively corre-
lated to each other in the entire sample, and in the Chinese
sample, but not in the Malay sample. We will discuss this result
separately for each ethnic group. For the Chinese, individual
self-esteem does not exist independently of relationship har-
mony. This conclusion is in line with the traditional Chinese
belief that the individual’s self-esteem is derived from, and is
conferred to, the individual by people who are related to the self.
For the Malays, relationship harmony seemed to be independent
of individual self-esteem, and was also curiously independent of
the interdependent self-construal. Interdependent self-construal
in the Malays was correlated with neither self-esteem nor rela-
tionship harmony. The SEM analysis of interdependent
self-construal was not found to contribute to the subjective
wellbeing of the Malays. This is a surprise to us. The Malays as
a community place high emphasis on the collective and inter-
personal relationships. However, being Muslim, they also be-
lieve in God and attribute life events mostly to God (Brenner,
1996; Shahiraa & Chang, 2003). Under the influence of Islam,
Malays in Southeast Asia have been observed to construct their
self-concepts on the basis of religion rather than social rela-
tionships (Brenner, 1996). This religious orientation might
suggest that independent/interdependent self-construal dichot-
omy might not capture the shared beliefs and meaning of the
self within the Malay Muslim community. More studies need to
be conducted to identify the conceptualization and manifesta-
tion of the self within the Malay community before conclusions
can be drawn about the two forms of self-construal and their
psychological c onsequences.
Predicting Subjective Wellbeing
Our study has extended the conclusions drawn by Kwan et al.
(1997) in that we have identified that the two mediators,
self-esteem and relationship harmony, work differently within
different Asian ethnic groups. Both self-esteem and relationship
harmony are important mediators of subjective wellbeing, but
in the Singaporean sample they were intertwined and did not
operate independently of each other. This intertwining might be
due to the equally intertwined nature of independent and inter-
dependent self-construal in the Singaporean context.
The general Singapore context is one with a high emphasis
on the collective and interpersonal harmony, but within each
ethnic group, the shared emphasis on the collective - or collec-
tivism-might take on different institutional and behavioral
forms. Collectivism, or interdependence, in the Chinese and the
Malay cultures of Singapore might originate from different
cultural heritages; for the Chinese, relationships constitute the
fundamental core of their world view, while for the Malays,
relationships are God-given and thus independent of construals
of the self. These differences could lead to differences in many
self-related psychological processes and constructs, including
subjective wellbeing and the paths that lead to subjective well-
Baron, R. M., & Kenny, D. A. (1986). The Moderator-Mediator vari-
able distinction in social psychological research: Conceptual, strate-
gic and statistical considerations. Journal of Personality and Social
Psychology, 51, 1173-1182.
Chang, W. C. (2003). Correction for differential response bias for be-
tween group comparison. Research Note, Asian Cultural Psychology
Project, National University of Singapore.
Chang, W. C. (2001, April). Multifaceted Chinese self. Paper presented
at the 59th Nebraska Symposium of Motivation, Lincoln, Nebraska:
University of Nebraska .
Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95,
Diener, E., & Diener, M. (1995). Cross-cultural correlates of life satis-
W. C. CHANG ET AL.
faction and self-esteem. Journal of Personality and Social Psychol-
ogy, 68, 653-663. doi:10.1037/0022-3518.104.22.1683
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The
satisfaction with life scale. Journal of Personality Assessment, 46,
Huang, L. L. (1999). Harmony and conflict of interpersonal relation-
ships: Indigenized theorization and research (in Chinese). Taipei,
Taiwan: Laurate Books.
Kwan, S. Y., Bond, M. H., & Singelis, T. M. (1997). Pancultural ex-
planations for life satisfaction: Adding relationship harmony to
self-esteem. Journal of Personality and Social Psychology, 73, 1038-
Larsen, R. J., Diener, E., & Emmons, R. A. (1985). An evaluation of
subjective well-being measures. Social Indicators Research, 17, 1-18.
Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implica-
tions for cognition, emotion and motivation. Psychological Review,
98, 224-253. doi:10.1037/0033-295X.98.2.224
Mohammad, N. B., Mohammad, I. B., & Chang, W. C. (2003). Stress-
and-coping of Malay children in Singapore. Paper presented at the 5th
International Conference of the Asian Association of Social Psy-
chology, Manila, Philippines.
Pavot, W., & Diener, E. (1993). Review of the satisfaction with life
scale. Psychological Assessment , 5, 164-172.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton,
NJ: Princeton University Press.
Shahiraa, Binte, M., & Chang, W. C. (2004). Stress-and-coping in the
Malay organizational context. Working paper, Singapore: Center of
Asian Cultural Psychology , Nanyang Technological University.
Singelis, T. M. (1994). The measurement of independent and interde-
pendent self-construals. Personality and Social Psychology Bulletin,
20, 580-591. doi:10.1177/0146167294205014
Singelis, T. M., Bond, M. H., Sharkey, W. F. & Lai, C. S. Y. (1999).
Unpacking culture’s influence on self-esteem and embarrassability.
Journal of Cross-Cultural Psychology, 30, 315-341.
Tong, E. M. W., Chang, W. C., & Koh, J. B. K. J. (2003). Differential
psychological functioning based on the orthogonality of self-con-
struals. Manuscript under review, Asian Cultural Psychology Project,
National University of Singapore.