You are on page 1of 13

Journal of Counseling Psychology © 2010 American Psychological Association

2010, Vol. 57, No. 1, 92–104 0022-0167/10/$12.00 DOI: 10.1037/a0017930

Shame, Internalized Heterosexism, Lesbian Identity, and Coming Out to Others:


A Comparative Study of Lesbians in Mainland China and Hong Kong

Pizza Ka-Yee Chow and Sheung-Tak Cheng


Hong Kong Institute of Education

The purpose of this study was to investigate coming out to family and friends and their relationships to
shame, internalized heterosexism, lesbian identity, and perceived social support in Chinese lesbians from
2 different cultural settings—Mainland China (N ⫽ 244) and Hong Kong (N ⫽ 234). Results of structural
equation modeling showed that, in both samples, a sense of shame was related to internalized hetero-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

sexism and a devaluation of one’s lesbian identity, which in turn was related to a decreased likelihood
This document is copyrighted by the American Psychological Association or one of its allied publishers.

of coming out to others. Shame was also associated with a reduced perception of support from friends,
which seemed in turn to exacerbate internalized heterosexism among lesbians. Family support was
generally unrelated to outness, except for outness to friends in the Hong Kong sample. Results are
discussed in relation to the cultural stigma attached to same-sex orientation and the cultural practice of
shaming that parents use to socialize children.

Keywords: coming out to others, shame, internalized heterosexism, lesbian identity, Chinese lesbians

Coming out to others, or outness, refers to verbally or behav- self (Corrigan & Matthews, 2003; Friedman & Downey, 1999;
iorally disclosing sexual orientation to others by sexual minorities. Herek, Chopp, & Strohl, 2007; Szymanski, Kashubeck-West, &
Research has demonstrated the importance of outness as the ac- Meyer, 2008). Despite changes in social attitudes over the years,
complished negotiation between self and society, and a sign of prejudice and discriminatory practices against lesbians are still
self-acceptance among lesbians and gays. Outness is a hallmark of common in contemporary societies, although the degree varies
lesbian and gay identity development and promotes self- from one society to another (National Coalition of Anti-Violence
integration, self-growth, and personal empowerment (Corrigan & Programs, 2007; Women Coalition of HKSAR, 2005). In a het-
Matthews, 2003; Herek, 2003). More important, outness denotes erosexist environment, internalized heterosexism is the norm
the resolution of inner stigma (Jordan & Deluty, 1998). Whereas rather than the exception for lesbians. This is especially true for
there is a small but respectable size of literature about coming out lesbians in Chinese societies, which are characterized by deeply
and identity development in Asian gay men (Harada, 2001; Jones, ingrained cultural attitudes against same-sex attraction. According
2007; Poon & Ho, 2008; Wong & Tang, 2004), little is known to Confucianism, daughters are expected to derive their identity
about Asian lesbians. In this study, we focused on the predictors of from relationships with men, performing their familial roles, and
outness in Chinese lebians living in two societies with different having children. The cardinal concept of Yin and Yang being in
socioeconomic and political systems. harmony further strengthens this gender role as fundamental to a
According to the modified labeling theory (Link, Cullen, proper social order. Undeniably, a daughter with same-sex orien-
Struening, Shrout, & Dohrenwend, 1989), minorities internalize tation is not just a disillusionment to parents, but a cause for family
cultural meanings about their condition and use such meanings to disgrace (Li & Orleans, 2001).
judge themselves. When growing up in a predominantly hetero- Because of internalized heterosexism, lesbians often construct a
sexist environment, internalized heterosexism often results as les- negative identity, even to the point of rejecting their lesbian
bians assimilate and internalize the majority view and public identity, and see themselves as inferior to other heterosexuals (see
stigma into schema about the self, leading to a devalued sense of review by Szymanski et al., 2008). Lesbian identity refers to the
acceptance of oneself and the comfort level in being a lesbian, and
is a strong predictor of outness (Ellen, Joy, Amber, Scales, & Sue,
Pizza Ka-Yee Chow and Sheung-Tak Cheng, Department of Psycho- 2008). According to McCarn and Fassinger (1996), lesbian iden-
logical Studies, Hong Kong Institute of Education, Hong Kong, China. tity is developed at both the personal and the collective level as
The Hong Kong data for this study were taken from Pizza Ka-Yee two parallel processes of construction. The identity construction is
Chow’s honors thesis completed under the supervision of Sheung-Tak expected to go through four phrases: awareness, exploration, com-
Cheng and submitted to the Department of Applied Social Studies, City mitment, and internalization. Although it is not clear whether
University of Hong Kong for partial fulfillment of the requirements of the
every lesbian would go through these stages in sequence, a positive
degree of bachelor of social sciences. We thank Michael Bender and Kin-Kit
development of lesbian identity along these lines requires a person
Li for their helpful comments on an earlier version of the manuscript.
Correspondence concerning this article should be addressed to Sheung- to acknowledge that she is different from the norm and to explore
Tak Cheng, Department of Psychological Studies, Hong Kong Institute her inner feelings for the same sex, as well as to see other lesbians
of Education, 10 Lo Ping Road, N.T., Hong Kong, China. E-mail: as a reference group for self-development and to recognize social
takcheng@ied.edu.hk oppression against sexual minorities. However, this process is

92
SHAME AND COMING OUT 93

often painful, associated with anger, guilt, and embarrassment, include gender-related norms. Conformity to these norms is essen-
because having grown up in a heterosexist environment, one has tial to maintaining harmony in collectivistic societies. However,
often engaged in heterosexist practices prior to coming to terms gay people tended to be more feminine, and lesbians more mas-
with one’s sexual orientation and struggling toward self- culine, in their plays and manners, and in choosing their hobbies,
acceptance as a lesbian (McCarn & Fassinger, 1996). Thus, re- interests, and occupational preference during childhood (Bailey &
search shows that internalized heterosexism is related negatively to Zucker, 1995), although such behaviors are not necessarily asso-
identity development and self-esteem, and positively to depression ciated with same-sex orientation in adulthood (Carr, 2007). Nev-
and hiding one’s sexual orientation (i.e., “passing”), among lesbi- ertheless, deviations from basic identity goals are likely to elicit
ans (Peterson & Gerrity, 2006; Szymanski, Chung, & Balsam, what Friedman and Downey (1999) called “gender-values self-
2001). condemnation” (p. 327). Because such identity goals are internal-
A more general concept related to internalized heterosexism is ized at a young age, shame is probably a more fundamental
self-stigma (Corrigan & Matthews, 2003; Mak et al., 2007). It is experience that predates internalized heterosexism in the develop-
sometimes used in an interchangeable fashion in the literature with ment of Chinese lesbian girls. Tracy and Robins (2004) further
another concept—shame. Despite their apparent similarity, there argued that shame would be more likely when the triggering events
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

are subtle, yet important differences between them. Both con- involve global and stable aspects of the self, as in the case of a
structs assumes either actual or imagined other disapproval, a tomboyish girl or a feminine boy in a traditional Chinese family.
sense of powerlessness and worthlessness, and a desire to conceal It is likely that the experience of chronic shame colors the way
personal deficiencies and to avoid contact with others (see Corri- future prejudice and discrimination from the public is interpreted,
gan & Matthews, 2003; Tangney, Wagner, & Gramzow, 1992; increasing one’s proneness to internalized heterosexism, and the
H. B. Lewis, 1971; M. Lewis, 1998). However, self-stigma is devaluation of one’s identity as a sexual minority.
theoretically a more localized process and is a devaluation of that This leads to another related experience among lesbians—social
aspect of the self that relates to the stigmatized status (in this case, and psychological isolation, which probably starts early in life. It
being a lesbian). Shame, however, involves a negative evaluation is well documented that tomboyish girls and feminine boys expe-
of the entire self. In other words, “the ‘bad thing’ is experienced as rience more rejection from their peers and significant others,
a reflection of a ‘bad self,’ and the entire self is painfully scruti- especially parents (e.g., Carter & McCloskey, 1983–1984;
nized and negatively evaluated” (Tangney et al., 1992, p. 469). Landolt, Bartholomew, Saffrey, Oram, & Perlman, 2004), with
This distinction has significant import for the present study. adverse consequences on their social competencies and self-worth
Research shows that the manifestations of self-conscious emo- (Egan & Perry, 2001). In fact, once a lesbian internalizes dominant
tions (e.g., shame, guilt) differ markedly between individualistic social norms, expectations of prejudice from others often lead to
and collectivistic cultures (Eid & Diener, 2001; Kitayama, Markus, social withdrawal and secrecy as coping behaviors (Link et al.,
& Matsumoto, 1995). Indeed, compared with Western cultures, 1989). Indeed, studies found anticipatory rejection from parents to
shame is a more ubiquitous experience in the Chinese culture. It is be a major deterrent to outness (Li & Orleans, 2001; Savin-
well documented that shaming is a preferred and commonly adopted Williams & Dubé, 1998). Not only is support from parents typi-
technique used by Chinese parents to socialize their children (Ho, cally lacking, support from aged peers can be disappointing in a
1986). Chinese children are taught that if they do certain things, and heterosexist society (Rosario, Schrimshaw, & Hunter, 2005), al-
fail to do others, then they bring dishonor to themselves and to their though younger people are supposed to hold more liberal attitudes
families. A study showed that Chinese undergraduates responded toward lesbians and gays. Compared with friends with same-sex
with higher shame ratings to all hypothetical scenarios of poten- orientation, the support from heterosexual friends might be more
tially shameful events than their American counterparts (Tang, meaningful and significant to the lesbians, as it signifies approval
Wang, Qian, Gao, & Zhang, 2008). Moreover, unlike individual- from the heterosexual world. Research in other populations has
ists who see their selves as unique and independent, collectivists shown that a lack of social support can exacerbate the perception
see their selves as interconnected and interdependent with each of stigmatization a year or more later (Mueller et al., 2006). Beals
other (Markus & Kitayama, 1991), and their opinions of them- and Peplau (2005) suggested that supportive behaviors from sig-
selves are more strongly influenced by those who are close to them nificant others are indicative of their readiness to understand and
than those who are more distant in relationship (Tang et al., 2008). accept the person, which in turn helped to affirm the lesbian
Because of this, Chinese lesbians will experience greater diffi- identity of the person. In the present study, weinvestigated the
culty, compared with their Western counterparts, in forming opin- relative contributions of friends and family support, as well as the
ions of themselves that are independent of their significant others, effects of shame, internalized heterosexism, and lesbian identity,
most notably parents. on outness. Specifically, we hypothesized that social support me-
We believe that the experience of shame, in contrast to inter- diates the effect of shame on internalized heterosexism; individu-
nalized heterosexism, happens at a much earlier stage in the lives als who are ashamed may perceive lesser social support as they
of Chinese lesbians and gays. Whereas internalized heterosexism isolate themselves to avoid the social repercussions against their
is the eventual psychological consequence of living in a prejudiced deviant status, which ironically would reinforce further stigmati-
environment, shame is probably a deep-seated emotion that is zation (Herek et al., 2007).
socialized in the family. Unlike basic emotions such as sadness and
anger, self-conscious emotions involve evaluations of the self in Stigma and Outness in Two Chinese Societies
relation to ideal- and ought-self representations (Tangney,
Niedenthal, Covert, & Barlow, 1998). Some of these representa- Great strides in the social and legal rights of lesbians and gays
tions are based on identity goals (Tracy & Robins, 2004), which have been achieved in recent decades in China and Hong Kong.
94 CHOW AND CHENG

After years of advocacy by ardent gay activists, homosexuality China. Despite decriminalization, the deep-rooted prejudice
was decriminalized in 1991 in Hong Kong and 1997 in China. against same-sex orientation is still the norm, and social oppres-
Beginning in 2001, homosexuality was no longer included in the sion is commonplace. The situation is slightly better in urban
official nomenclature for mental illness in China. Lagging far settings, which are more open to Western influences, than in rural
behind many Western countries, however, both societies have yet areas. Across China, films with themes on same-sex attraction are
to grant sexual minorities equal civil rights (e.g., the Equal Op- still strictly prohibited in the public and can be viewed only in
portunities Ordinances in Hong Kong do not cover sexual minor- private. There are no such restrictions in Hong Kong, however,
ities), let alone the recognition and legalization of same-sex mar- where sexual minorities enjoy a lot more freedom. In Hong Kong,
riage. In December 2008, as many as 66 countries signed a joint gay bars, clubs, films, art exhibitions, websites, and organizations
United Nations (UN) declaration to decriminalize homosexuality are easy to find.
and to condemn all forms of discrimination and violence against Because of a rather different social, cultural, and political back-
people on the basis of their sexual orientation or gender identity. ground, we postulate that coming out, especially to one’s family,
This declaration, however, was not endorsed by China (Hong would be more common among lesbians in Hong Kong than
Kong is not an independent member of the UN). It is generally among those in China and that the sense of shame and internalized
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

believed that China adopts a “3-no” policy—no approval, no heterosexism would both be lower among lesbians in Hong Kong
disapproval, and no promotion— on homosexuality. than among those in China. In a similar vein, we assume that the
Despite decriminalization of homosexuality in both societies, preception of family and friends support would be higher among
there is still much prejudice against individuals with same-sex lesbians in Hong Kong than among those in China.
orientation, especially in mainland China. The Confucian notion of
filial piety stipulates the continuation of the family line as a basic The Present Study
duty of children to parents. A woman’s role is therefore to marry
to another man and give birth to children, especially sons (Cheng This study concerns the factors related to coming out to family
& Chan, 2006). The idea that same-sex attraction is deviant is and friends among lesbians in China and Hong Kong. Because
deeply rooted in the Chinese mind-set. However, Hong Kong was those with a less affirmative lesbian identity would tend to conceal
under British rule for nearly a century and was returned to China their sexual orientation from others, even to the extent of avoiding
only about 10 years ago. China, however, has been closed to the interactions (e.g., Szymanski et al., 2001), lesbian identity is
West until the last two decades. For this reason, Hong Kong may postulated as the proximal factor in outness. Shame and internal-
represent a rather Westernized and modern society, although it ized heterosexism are hypothesized to relate to outness through
retains many Chinese cultural traditions (Chia, Allred, & Jerzak, devaluation of one’s lesbian identity. Because of the potential
1997; King, 2004). This suggests that the cultural bias against effect of social support on internalized heterosexism (see Mueller
same-sex orientation should be stronger in China than in Hong et al., 2006), perceived social support from family and friends are
Kong. Moreover, because of Westernization, individualism has hypothesized to be related to outness by way of internalized
shaped Hong Kong people’s lives, whereby individual rights and heterosexism, and then lesbian identity. Finally, because of the
freedom to express oneself is also respected. Hong Kong can predominant nature of shame in Chinese societies, it was hypoth-
therefore be described as a mixture of individualism and collec- esized to be the common factor in both internalized heterosexism
tivism (King, 2004). Indeed, a meta-analysis showed that although and reduced support (i.e., individuals who felt more shameful of
Hong Kong is more collectivistic than Western societies, it is themselves would tend to have more intense internalized hetero-
nonetheless less collectivistic and more individualistic than China sexism about their same-sex orientation and tend to shy away from
(Oyserman, Coon, & Kemmelmeier, 2002). Hence, it can be ex- others). For this reason, shame also has an indirect effect on
pected that deviation from mainstream social norms is less of a internalized heterosexism via reduced social support. This is our
concern to lesbians in Hong Kong, who may value their lesbian theoretical model, and it is represented by Model A in Figure 1.
identity more than their Mainland Chinese counterparts (see Mao, In addition, we tested three alternate models. Model B includes,
McCormick, & van de Ven, 2002). on top of Model A, two additional paths from shame and inter-
As the awareness of being members of an oppressed minority nalized heterosexism to outness (path a and b, respectively, in
group is critical for lesbian identity development (e.g., McCarn & Figure 1). This model therefore postulates that shame and inter-
Fassinger, 1996), it is also noteworthy that social activism is nalized heterosexism related to outness directly, aside from the
tolerated much differently in the two societies. Prior to the decrim- indirect association by way of lesbian identity. Model C, however,
inalization legislation in Hong Kong, public interest groups were differs from Model A in that family and friends support are
primarily concerned with drawing the public’s attention to the postulated to be related to outness directly (path c and d, respec-
“invisible” group of lesbians and gays in the society. After the tively, in Figure 1). Finally, because internalized heterosexism can
enactment of decriminalization legislation in 1991, however, ac- lead to social withdrawal (Kaufman & Johnson, 2005), Model D
tivities become multifaceted and large scale, including gay pride postulates that the relationships between internalized heterosexism
parades and protests, in order to gain more rights for the group, and social support are actually in the opposite direction; that is,
including the legalization of homosexuality. These collective ac- internalized heterosexism is associated with reduced family and
tivities, some of which were joined by lesbians and gays from friends support, which in turn are associated with a higher likeli-
other parts of the world, are no doubt beneficial for the identity hood of coming out to others (see Figure 1b). Despite hypothesized
development of lesbians and gays in Hong Kong. Compared with mean differences between China and Hong Kong in the study
Hong Kong, lesbian and gay activities in China are carried out variables as mentioned above, we expected to find the best fitting
mostly “underground,” and there are no formal gay associations in models to be the same for both societies and that the way the
SHAME AND COMING OUT 95
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 1. Alternate models showing the pathways by which shame and internalized heterosexism decrease
outness in lesbians. Figure 1a displays Models A, B, and C, with Model B being identical to Model A except
for the additional paths a and b, and Model C being identical to Model A except for the additional paths c and
d. Figure 1b displays Model D. Paths in black denote hypothesized positive relationships; those in shaded lines
denote negative relationships.

variables relate to each other (i.e., the strengths of the associations) about what other people think of your appearance?” In none of the
are also the same between them. items was the term homosexuality, gay, lesbian, or sexual orien-
tation mentioned. Each item was rated on a 5-point scale ranging
Method from 1 (not at all) to 5 (very much), with higher scores indicating
more shame. The scale was shown to have high internal consis-
Participants. Lesbians were recruited from relevant websites
tency (␣ ⫽ .92) and 11-week test–retest reliability (r ⫽ .83).
(see below) in China (n ⫽ 244) and in Hong Kong (n ⫽ 237).
Convergent validity was demonstrated by correlations with other
However, three Hong Kong participants with response set (filling
measures of shame and guilt, and nomological validity was sup-
in the same score for every question) were removed, leaving 234
ported by predicted changes in depressive symptoms over an
participants for the Hong Kong sample. Characteristics of the two
11-week interval (Andrews et al., 2002). Two items were removed
samples are presented in Table 1. Expectedly, the samples were
due to low item-total correlations, leaving 23 items for analysis.
relatively young; almost all the participants were aged 30 or
younger. However, the Hong Kong participants were much Alpha coefficients for the 23-item scale equaled .92 and .94 for the
younger, with almost half the sample teenagers; on the contrary, China and the Hong Kong sample, respectively.
most of the participants from Mainland China were in their 20s.1 Internalized heterosexism. Internalized heterosexism was as-
Partly because of this, participants from Hong Kong had on the sessed using the Self-Stigma Scale, originally developed for Chi-
average lower educational attainment and were more likely to live nese populations (Mak et al., 2007; Mak & Wu, 2006). It measures
with parents, than those from China. Moreover, perhaps due to the the affective, cognitive, and behavioral dimensions of the experi-
differential living standard between the two places, it was much ence of stigma. Compared with other measures of internalized
less likely for lesbians in Hong Kong to live alone or with their heterosexism (Szymanski et al., 2008), this scale was tailor-made
partners anyway. Consistent with census figures (Census and Sta- in the Chinese context and was intended to serve as a common
tistics Department, 2007; National Bureau of Statistics of China, basis for assessing internalized stigmatization across different mar-
2002), the majority of participants in both samples declared to ginalized groups. Items were therefore developed on the basis of
have no religion. Finally, nearly half in both samples was perform- focus groups with various social minorities, including lesbians,
ing the butch role, a quarter femme, and one third undifferentiated.
Measures. Except for the Self-Stigma Scale (which was de- 1
veloped in Chinese), all measures were translated into Chinese Because we do not have access to membership information of the
various websites, we do not know to what extent membership character-
using the back-translation procedure to ensure semantic equiva-
istics determine the different age distributions of the two samples. How-
lence with the original English version. ever, by examining activities on the various discussion forums belonging to
Shame. The 25-item Experience of Shame Scale (Andrews, different age groups, the age distributions in both samples reflect the
Qian, & Valentine, 2002) measures affective, cognitive, and be- relative frequency of discussions as seen on these webs. Note also that
havioral aspects of shame. Sample items are “Have you felt admission to an Internet bar/cafe is limited to those aged 18 or over in
ashamed of the sort of person you are?” and “Have you worried China.
96 CHOW AND CHENG

Table 1 comfortable knowing that others judge me negatively for being


Sociodemographic Characteristics, China (N ⫽ 244) and Hong lesbian” and “I would rather be straight if I could.” Exploratory
Kong (N ⫽ 234) Samples factor analyses supported a six-factor structure—identity confu-
sion, identity superiority, internalized homonegativity, need for
% acceptance, need for privacy, and difficult processes—for both
Variable China Hong Kong ␹2 df lesbians and gays in the United States. Alpha coefficients for the
six subscales ranged from .65 to .81. Construct validity was

Age 118.20 4 supported by negative correlations with same-group orientation,
15–18 9.0 48.9 the internalized/synthesis phase of individual sexual identity de-
19–22 31.6 32.9
23–26 35.5 13.4 velopment, and self-esteem (Mohr & Fassinger, 2000). However,
27–30 17.2 2.6 in our pilot study of 49 lesbians in Hong Kong, the Difficult
30⫹ 6.7 2.2 Processes subscale (five items) correlated at r ⫽ .78 with the
Gender role as lesbian 4.78 2 Self-Stigma Scale, raising questions as to whether the two mea-
Butch 43.1 48.1
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Femme 22.1 25.1 sures were sufficiently distinct. Moreover, these five items, along
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Undifferentiated 34.8 26.8 with four others (distributed across the other subscales) that did not
Education 134.02ⴱ 3 function equivalently across the two samples were removed, leav-
Junior high 3.3 44.2 ing 18 items for the measure of lesbian identity.2 This 18-item
Senior high 21.3 25.5
College 67.2 29.0
scale had alpha coefficients of .82 and .85 for the China and the
Master’s degree or above 8.2 1.3 Hong Kong sample, respectively. A higher score indicates a less
Living arrangement 94.53ⴱ 2 affirmed identity.
With parents 47.1 89.2 Perceived social support. The Perceived Social Support
With partners 24.2 4.8 Scales (Procidano & Heller, 1983) measure perceived support
Alone 28.7 6.1
Employment 69.65ⴱ 4 from family and friends, each with 20 yes–no items. Sample items
Student 34.4 72.7 were “My family enjoys hearing what I think” and “My friends are
Blue collar, sales, clerical 25.4 17.3 sensitive to my personal needs.” Alpha coefficients for the Family
Professional 15.6 5.6 and Friends support subscales were reported to be .88 and .79 in
Unemployed 7.4 4.3
Others 17.2 0.0 Asian American college students (Gloria & Ho, 2003), and .89 and
Religion 64.27ⴱ 3 .85, respectively, in college students with bisexual orientation
No religion 76.2 65.8 (Sheets & Mohr, 2009). In support of their nomological validity,
Christianity 4.1 28.6 both subscales were found to be associated with depression and
Buddhism 16.8 5.2
Others 2.9 0.4
life satisfaction in bisexual individuals (Sheets & Mohr, 2009).
Again, due to nonequivalence across samples, three Family and

p ⬍ .001. two Friends support items were omitted, leaving 17 items for the
Family subscale and 18 for the Friends subscale (see Footnote 2).
For the China and the Hong Kong sample, respectively, alpha
gays, bisexuals, mental health consumers, and the like. It has a coefficients, based on the reduced set of items, were .91 and .91 for
long (22 items) and a short (15 items) version. The short version the Family subscale, and .83 and .80 for the Friends subscale.
was used in the present study, with the alpha coefficient reported Outness. The Outness Inventory (Mohr & Fassinger, 2000)
to be .95 in a previous Hong Kong sample (Mak & Wu, 2006). measures, using a 7-point Likert scale, the degree to which the
Sample items are “I feel pity for myself as being a _____,” and “I
participant’s same-sex orientation is known to another person and
don’t dare to make new friends lest they find out that I am
has been openly discussed with him or her (1 ⫽ definitely does not
a _____” and the word lesbian was inserted in the blank space in
know, 2 ⫽ might know but never talked about, 3 ⫽ probably knows
each item. The items were rated on a 4-point scale ranging from 1
but never talked about, 4 ⫽ probably knows but rarely talked
(strongly disagree) to 4 (strongly agree), with higher scores indi-
about, 5 ⫽ definitely knows and rarely talked about, 6 ⫽ definitely
cating more intense internalized heterosexism. In a sample of
knows and sometimes talked about, 7 ⫽ definitely knows and
Chinese adults with HIV/AIDS, exploratory factor analysis of the
openly talked about). On the basis of exploratory factor analysis in
long version supported the scale’s unidimensionality, and, in sup-
lesbians and gays, Mohr and Fassinger categorized the target
port of the measure’s nomological validity, it was found to be
persons into three categories: (a) family members (mother, father,
associated with less social support and more psychological distress
siblings, extended family, and relatives), (b) world (old and new
(Mak et al., 2007). The 15-item short form had alpha coefficients
heterosexual friends, strangers, work peers, and work supervisors),
of .92 and .95 for the China and the Hong Kong sample, respec-
and (c) members of religious groups (religious leader and religious
tively.
member). Alpha coefficients were reported to be .74, .79, and .97
Lesbian identity. Compared with other measures of identity
for outness to family, world, and religion, respectively. In support
development that focus on identity stages (e.g., Brady & Busse,
1994), the Lesbian and Gay Identity Scale (Mohr & Fassinger,
2000) focuses on dimensions of the lesbian and gay experience 2
We examined each scale one by one for cross-cultural equivalence first
that are relevant throughout the entire identity development pro- using the method of invariance analysis in confirmatory factor analysis.
cess. It contains 27 items rated on a 7-point scale (1 ⫽ disagree Items that were nonequivalent were removed prior to forming the item
strongly, 7 ⫽ agree strongly). Sample items are “I can’t feel parcels for the main analyses.
SHAME AND COMING OUT 97

of construct validity of the scale, outness to the different targets complexity, with a lower value indicating a better model fit.
was associated positively with same-group orientation and the According to Burnham and Anderson (2002), an AIC difference ⱖ
internalized/synthesis phase of individual sexual identity develop- 10 is a strong indication that one model is inferior to the other, a
ment, but negatively with the deepening/commitment phase of difference of 3–9 suggests different degrees of departure, and a
group membership identity (Mohr & Fassinger, 2000). Similar to difference ⱕ 2 does not warrant interpretation. Finally, the joint
what Mohr and Fassinger (2000) found in their U.S. sample, significance test (i.e., all mediating pathways need to be significant)
missing values were plenty for the religious targets (in both soci- was used to identify significant mediation effects (MacKinnon,
eties, people with religious beliefs are a minority; Census and Lockwood, Hoffman, West, & Sheets, 2002; Taylor, MacKinnon,
Statistics Department, 2007; National Bureau of Statistics of & Tein, 2008).3
China, 2002). Moreover, most of our participants were not em-
ployed (see Table 1). Hence the “religious” and “work” targets
Results
were dropped. Furthermore, because it was extremely unlikely for
our participants to disclose to strangers (M ⫽ 1.49, SD ⫽ 1.27 for Coming out to family and friends. Table 2 presents data on
China; M ⫽ 1.62, SD ⫽ 1.18 for Hong Kong), this item was also outness (scoring 5 or above on the Outness Inventory) to various
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

dropped. Thus, our data focused on outness to two broad catego- targets. Across the two samples, a lot more lesbians disclosed to
ries of relationship—family and friends. The alpha coefficients for friends (particularly long-term heterosexual friends) than to family
the reduced set of items were .82 and .85 for the China and the members. With both samples combined, over half disclosed to
Hong Kong sample, respectively. A higher score indicates more friends but only one third disclosed to any family member. Prom-
openness about one’s sexual orientation. inent targets for disclosure within the family were siblings and the
Procedure. Lesbians are a group of sexual minorities who are mother. Across all targets, disclosure was much more common
more or less invisible in Chinese societies. Hence, for recruitment, among lesbians in Hong Kong than among those in China. Such a
an anonymous online questionnaire link was posted in five lesbian difference was in general more pronounced with heterosexual
web forums in each society. These forums, connected to public friends than with family members as targets.
lesbian and gay organizations, are presented as a supportive com- Descriptive statistics of observed variables. The means and
munication platform where plenty of self-identified lesbians standard deviations of the observed variables (the outness mea-
gather, interact, and exchange assorted information. Information sures here were composites of the raw scores of constituent items)
on (mostly social) activities of interest to lesbians is also publi- are shown in Table 3. To further examine differences between
cized on these webs. Recruitment was limited to those who were China and Hong Kong, we conducted a multivariate analysis of
ethnically Chinese, who provided informed consent on the web variance with shame, internalized heterosexism, lesbian identity,
questionnaire. Questionnaires printed in simplified Chinese char- and family and friend support as dependent variables, and country
acters were placed on the China websites, and questionnaires in as the between-subjects factor. No multivariate outliers were iden-
traditional Chinese characters were placed on the Hong Kong tified using Mahalanobis distances. An overall main effect of
websites. The internet protocol (IP) address of each submission country, Pillai’s F(5, 472) ⫽ 3.89, p ⬍ .01, ␩2 ⫽ .04, was found.
was carefully checked to avoid possible duplicate submissions Further univariate tests showed that lesbian identity, F(1, 477) ⫽
from the same computer user. There were no identical IP addresses 7.95, p ⬍ .01, ␩2 ⫽ .02, and friends support, F(1, 477) ⫽ 6.83, p ⬍
in both samples, and hence all submissions were considered valid .01, ␩2 ⫽ .01, were significantly different between the two sam-
entries. ples, with lesbians in Hong Kong reporting more positive identity
Data analysis. The different theoretical models (see Figure 1) with their sexual orientation and better support from friends than
were tested using structural equation modeling. The covariance those in China. Thus, not only were lesbians in Hong Kong more
matrix of the indicators was subject to the maximum likelihood likely to disclose their sexual orientation to family and friends,
estimation using LISREL Version 8.52. Because the outness data they also had a slightly more positive lesbian identity and received
were skewed, the Satorra-Bentler scaled chi-square is reported more support from friends than the lesbians in China. Furthermore,
(Satorra & Bentler, 1994), which corrects the chi-square statistic a paired-sample t test showed more perceived support from friends
(and all fit indices based on it) as well as the standard errors of than from family, whether in Hong Kong, t(233) ⫽ 11.16, p ⬍
parameter estimates for nonnormality when the maximum likeli- .001, d ⫽ 1.39, or in China, t(243) ⫽ 10.55, p ⬍ .001, d ⫽ 1.23.
hood estimator is used. Besides chi-square, the comparative fit Measurement invariance. Before proceeding to test the full
index (CFI), nonnormed fit index (NNFI), standardized root-mean- structural models, it is paramount that the latent constructs are
square residual (SRMR), and root-mean-square error of approxi- equivalent and relate to each other the same way across groups
mation (RMSEA) were used to evaluate model fit. In general, CFI (i.e., measurement invariance). Given the relatively small within-
and NNFI ⱖ .95, SRMR ⱕ .08, and RMSEA ⱕ .06 (Hu & Bentler,
1999) represent good fit to the data, but these should not be taken
3
as rigid rules (Marsh, Hau, & Wen, 2004). In addition, to evaluate In the simplest A 3 B 3 C mediating model, the joint significance test
the relative fit of nested models (Model B/Model C vs. Model A), determines the existence of a mediation effect by variable B when the paths
A 3 B and B 3 C are both statistically significant. Although not
the difference between two scaled chi-square values were adjusted
necessarily the most powerful test among different ways to test mediation
using the procedure described in Satorra and Bentler (2001) so that effects, it provides an effective control for Type I error rate but is not
the critical values for normal theory chi-square can be used. necessarily the most powerful test (MacKinnon et al., 2002; Taylor et al.,
However, because Model D is not nested within Model A, the two 2008). We compared the results using this test and those from the more
models in which the Akaike information criterion (AIC) was used powerful bias-corrected bootstrapping procedure (Mallinckrodt, Abraham,
were compared, which corrects the chi-square statistic for model Wei, & Russell, 2006). The results were the same.
98 CHOW AND CHENG

Table 2 the number of estimated parameters, we conducted the analyses


Targets of Disclosure separately for outness to family and outness to friends as depen-
dent variables. The results based on the scaled chi-square statistics
% are reported in Table 6. (An estimation of standardized residuals
China Hong Kong Total ␹2(1)a
using multiple regressions with latent variable scores revealed no
outliers in all models.)
Family members Next, we examined whether Models B and C provided better fit
Father 4.9 12.8 8.8 9.31ⴱⴱ to the data than Model A. For outness to family, both Model B,
Mother 11.1 25.7 18.2 17.04ⴱⴱⴱ
adjusted ⌬␹2(2) ⫽ ⫺2.97 and 0.02, respectively, both ns, and
Brothers and sisters 18.9 28.6 23.6 6.33ⴱ
Other relatives 3.7 9.0 6.3 5.67ⴱ Model C, adjusted ⌬␹2(2) ⫽ ⫺3.71 and ⫺2.59 respectively, both
Any family member 27.5 40.6 33.9 9.20ⴱⴱ ns, were not significantly different from Model A in the China and
Friends the Hong Kong sample. Model D, however, was clearly less fitting
Old heterosexual friends 34.9 50.0 42.3 11.25ⴱⴱⴱ than Model A due to the substantial difference in AIC values in
New heterosexual friends 21.3 39.3 30.1 18.40ⴱⴱⴱ
both samples. Model A was therefore accepted as the model of
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Any heterosexual friend 47.1 62.0 54.4 10.60ⴱⴱ


This document is copyrighted by the American Psychological Association or one of its allied publishers.

choice for the sake of parsimony. In other words, a strong lesbian


Note. Figures were based on responses to the Outness Inventory, with a identity was the key factor in coming out to family members, and
score of 5 (“definitely knows”) as the cutoff. Numbers add up to ⬎ 100% it completely mediated the effects of shame and internalized het-
because each person could have multiple targets of disclosure.
a erosexism on coming out. Thirteen percent of the variance in
Chi-square tests of differences between China and Hong Kong.

p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001. outness to family was explained by lesbian identity. Social support
was not related to outness directly, but only indirectly through
reduced internalized heterosexism. We further tested whether
group sample sizes and the dichotomous response format of the Model A was invariant between the two samples. On top of the
social support items, items (except those for outness to friends) equality of factor loadings, we forced all structural path coeffi-
tapping the same construct were randomized into three parcels, cients to be equal across the two groups (i.e., another nested
each being the average score of the items, because parcels help to model). This resulted in a nonsignificant increase in chi-square,
correct distribution problems, are more reliable than the individual adjusted ⌬␹2(8) ⫽ 4.20, ns, over and above the model that spec-
items themselves, and improve the ratio of sample size to indica- ified only factor loadings to be equivalent. Thus, the latent vari-
tors (Little, Cunningham, Shahar, & Widaman, 2002). Four items ables related to each other in the same way in both samples when
measuring outness to family were also randomized, with outness to accounting for outness to family. Figure 2 shows the estimates for
siblings and outness to other relatives being combined into one the common metric completely standardized solution. All paths
parcel (see Parcel 3 in Table 4), and outness to mother (Parcel 1) were significant and in the expected direction, except for the path
and outness to father (Parcel 2) being single-item parcels. As for leading from family support to internalized heterosexism.
outness to friends, the two constituent items each served as an For outness to friends in China, again, both Model B, adjusted
indicator for the latent factor, with old and new heterosexual ⌬␹2(2) ⫽ ⫺4.13, ns, and Model C, adjusted ⌬␹2(2) ⫽ ⫺4.18, ns, was
friends being Indicators 1 and 2, respectively (see Table 4). The not significantly different from Model A. The results, however, were
original 7-point items were used for forming the indicators for the different for the Hong Kong sample, in which Model C fit the data
outness constructs. better than Model A, adjusted ⌬␹2(2) ⫽ ⫺7.81, p ⬍ .05, but Model
Results showed that the indicators loaded on the same factors B was not, adjusted ⌬␹2(2) ⫽ ⫺2.36, ns. Moreover, Model D was
(i.e., factor-pattern invariance) in both samples, ␹2(298) ⫽ 397.21, inferior than Model A in the China sample, and it was also inferior to
p ⬍ .001; CFI ⫽ .97, NNFI ⫽ .97, SRMR ⫽ .05, and RMSEA ⫽ Model C in the Hong Kong sample. Thus, for outness to friends,
.04. We then forced all factor loadings to be equal across the two Model A and Model C should be adopted for the China and the Hong
groups, over and above factor-pattern invariance (i.e., a nested Kong sample, respectively. The completely standardized solutions for
model), resulting in adjusted ⌬␹2(20) ⫽ 21.78, ns. Thus, the same these models are displayed in Figures 3a and 3b, respectively. Com-
set of latent constructs was being measured, whether among les-
bians in China or those in Hong Kong, and shame, internalized
heterosexism, and lesbian identity could be distinguished from one
Table 3
another. Moreover, outness to family could also be distinguished
Descriptive Statistics of Observed Study Variables
from outness to friends. We report the overall factor loadings for
the combined sample (N ⫽ 478) in Table 4. All parcels loaded China Hong Kong
highly on their factors. Intercorrelations of the latent variables are
displayed in Table 5. Measure M SD M SD
Evaluating structural models. The establishment of mea-
Shame 58.17 17.58 58.93 17.91
surement invariance set the stage for testing structural models. Internalized heterosexism 21.81 7.40 21.54 7.84
Because of age and educational differences between the two sam- Lesbian identity 62.09 18.40 59.69 17.17
ples, we first compared models with and without controlling for Family support 16.50 9.93 17.42 9.67
the potential effects of age and education on outness. Results Friends support 27.04 6.91 28.59 5.95
Outness to family 7.68 4.23 10.43 5.35
showed that the path coefficients as well as the fit indices were Outness to friends 5.80 3.39 7.90 3.57
basically identical with or without the controls; hence, we elimi-
nated them from the model. To improve the ratio of sample size to Note. Ns ⫽ 244 for China sample and 234 for Hong Kong sample.
SHAME AND COMING OUT 99

Table 4
Factor Loadings for the Measurement Model, China and Hong Kong Samples Combined

Unstandardized Standardized
Latent factor and indicators factor loading SE Z factor loading

Experience of shame
Shame Parcel 1 0.78 0.03 26.95 .89
Shame Parcel 2 0.79 0.02 32.33 .94
Shame Parcel 3 0.67 0.02 27.65 .88
Internalized heterosexism (IH)
IH Parcel 1 0.48 0.03 18.23 .89
IH Parcel 2 0.48 0.03 16.23 .92
IH Parcel 3 0.52 0.03 18.93 .94
Lesbian identity
Identity Parcel 1 0.80 0.04 19.64 .76
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Identity Parcel 2 0.89 0.04 20.48 .85


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Identity Parcel 3 0.75 0.04 18.87 .78


Perceived social support from family
Family support Parcel 1 0.54 0.02 32.79 .90
Family support Parcel 2 0.47 0.02 30.10 .88
Family support Parcel 3 0.44 0.01 31.84 .88
Perceived social support from friends
Friends support Parcel 1 0.33 0.02 18.46 .78
Friends support Parcel 2 0.32 0.02 15.68 .77
Friends support Parcel 3 0.25 0.02 14.86 .70
Outness to family
Family outness Parcel 1 0.98 0.09 10.40 .67
Family outness Parcel 2 1.44 0.09 16.87 .87
Family outness Parcel 3 0.57 0.08 7.42 .44
Outness to friends
Friend outness Indicator 1 1.45 0.12 12.62 .67
Friend outness Indicator 2 1.27 0.12 10.73 .58

Note. N ⫽ 478. All factor loadings were significant at the .001 level.

pared with the earlier analysis on outness to family, lesbian identity heterosexism had small to moderate indirect effects on outness
accounted for more variance (34%) in outness to friends in the China across models; nevertheless, it was noteworthy that they had a
sample and, together with family support, accounted for 27% of the combined massive effect size of ⫺0.56 ([⫺0.29] ⫹ [⫺0.27]) on
variance in outness in the Hong Kong sample. The China model was outness to friends among lesbians in China. Across models, the
generally the same for the one with outness to family as the dependent effect sizes of friends and family support tended to be noticeably
variable, and we do not discuss it further. Interestingly, however, for smaller than those for shame and internalized heterosexism.
the Hong Kong model, it was family support, not friends support, that
was directly associated with coming out to friends. Friends support Discussion
continued to exert its indirect effect on outness through the internal-
ized heterosexism 3 lesbian identity pathway, but its direct effect on This study investigated whether outness, the disclosure of one’s
outness to friends was not significant. same-sex orientation, was related to lesbian identity, internalized
The standardized direct and indirect effects of the different heterosexism, and social support from family and friends, within
latent variables are shown in Table 7. Both shame and internalized the larger context of the experience of shame. To our knowledge,

Table 5
Within-Group Correlations Among Latent Variables

Latent variable 1 2 3 4 5 6 7
ⴱⴱⴱ ⴱⴱⴱ ⴱⴱ ⴱⴱⴱ ⴱ
1. Shame — .38 .52 ⫺.18 ⫺.26 ⫺.21 ⫺.12
2. Internalized heterosexism .43ⴱⴱⴱ — .60ⴱⴱⴱ ⫺.19ⴱⴱ ⫺.47ⴱⴱⴱ ⫺.15ⴱ ⫺.33ⴱⴱ
3. Lesbian identity .47ⴱⴱⴱ .53ⴱⴱⴱ — ⫺.25ⴱⴱⴱ ⫺.33ⴱⴱⴱ ⫺.41ⴱⴱⴱ ⫺.60ⴱⴱⴱ
4. Family support ⫺.12 ⫺.13ⴱ ⫺.05 — .40ⴱⴱⴱ .19ⴱ .09
5. Friends support ⫺.38ⴱⴱⴱ ⫺.34ⴱⴱⴱ ⫺.31ⴱⴱⴱ .28ⴱⴱⴱ — .18ⴱ .45ⴱⴱⴱ
6. Outness to family ⫺.17ⴱ ⫺.19ⴱ ⫺.32ⴱⴱⴱ .17ⴱ .05 — .60ⴱⴱⴱ
7. Outness to friends ⫺.32ⴱⴱⴱ ⫺.38ⴱⴱⴱ ⫺.47ⴱⴱⴱ .20ⴱ .27ⴱⴱ .40ⴱⴱⴱ —

Note. Ns ⫽ 244 for China sample and 234 for Hong Kong sample. Correlations for the China sample are displayed above the diagonal; those for the Hong
Kong sample are below.

p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
100 CHOW AND CHENG

Table 6
Fit Indices for Alternate Models for Outness to Family and Outness to Friends

␹2 df CFI NNFI SRMR RMSEA AIC

Outness to family
China
Model A 213.18 127 .97 .97 .08 .05 301.18
Model B 209.74 125 .97 .96 .08 .05 301.74
Model C 209.35 125 .97 .96 .08 .05 301.35
Model D 230.38 126 .96 .95 .10 .06 320.38
Hong Kong
Model A 208.30 127 .97 .96 .06 .05 296.30
Model B 208.33 125 .97 .96 .06 .05 301.39
Model C 205.21 125 .97 .97 .06 .05 297.21
Model D 233.64 126 .96 .96 .08 .06 323.64
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Outness to friends
China
Model A 194.20 111 .97 .97 .08 .06 278.20
Model B 189.79 109 .97 .97 .08 .06 277.79
Model C 189.96 109 .97 .97 .08 .06 277.96
Model D 216.33 110 .96 .95 .11 .06 302.33
Hong Kong
Model A 206.02 111 .97 .96 .06 .06 290.02
Model B 203.07 109 .97 .96 .06 .06 291.07
Model C 198.80 109 .97 .96 .06 .06 286.83
Model D 231.41 110 .96 .95 .09 .07 317.41

Note. Ns ⫽ 244 for China sample and 234 for Hong Kong sample. All chi-square values were significant at the .001 level. CFI ⫽ comparative fit index;
NNFI ⫽ nonnormed fit index; SRMR ⫽ standardized root-mean-square residual; RMSEA ⫽ root-mean-square error of approximation; AIC ⫽ Akaike
information criterion.

this is the first empirical study ever conducted in Chinese lesbian friends was less common in the former. Identity, in turn, was
populations. Consistent with our predictions, outness was much related to a sense of shame and internalized heterosexism. More-
less likely for those who had a weak and negative lesbian identity. over, shame had both direct and indirect (via negative associations
Because lesbian identity was more negative among lesbians in with perceived support from friends) effects on internalized het-
China than among those in Hong Kong, coming out to family and erosexism. These findings were by and large similar between
China and Hong Kong. The findings strongly supported the need
to examine the simultaneous effects of internalized heterosexism
and shame on the self-disclosure of sexual orientation by lesbians.
However, because coming out to family was more difficult for
Chinese lesbians, the models were more effective in explaining
outness to friends than in explaining outness to family.
In a heterosexist society, those with same-sex orientation are
often seen as deviant and treated unfairly. Such prejudice and
associated practices may come not only from the public but also
from close others, such as family members. Kanuha (1997) argued
that prejudice and discrimination against the outgroup were due to
the aversive valuations (i.e., same-sex orientation is morally wrong
and unacceptable) that culture delivers toward lesbians, or stigma.
Such cultural views become personally relevant and form the basis
for self-evaluation (Link et al., 1989). The paramount negative
valuations result in a stigmatized self, which in turn leads to an
ambiguous identity. Such an identity development makes disclos-
ing a socially ostracized status even more unlikely (Anhalt &
Morris, 1998; Szymanski et al., 2008; Tangney & Dearing, 2002).
It is therefore not surprising that coming out to others is not a
common phenomenon in Chinese lesbians, especially within fam-
ilies.
Figure 2. Final structural equation model for outness to family (Model Noticeably, this study showed that the effect of shame has to be
A), common metric solution for China (N ⫽ 244) and Hong Kong (N ⫽ distinguished from that of internalized heterosexism. Although
234) samples. ⴱⴱⴱ p ⬍ .01. shame and internalized heterosexism may appear to be similar
SHAME AND COMING OUT 101
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 3. Final structural equation model for outness to friends for (a) China sample (N ⫽ 244; Model A) and
(b) Hong Kong sample (N ⫽ 234, Model C). ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱ p ⬍ .01.

constructs, they were only moderately correlated, and were both may also be another factor that contributes further to the devalu-
uniquely associated with identity formation in this study. The ation of their identity. Accordingly, parents were unlikely targets
experience of shame involves discrepancies between actual self of disclosure within the immediate family in both samples. In fact,
and ideal selves, as well as those between actual self and ought disclosing to fathers was as unlikely as disclosing to distant rela-
selves (Tangney et al., 1998). The ideal self represents hopes and tives, as fathers tend to be less tolerant and react more negatively
aspirations, whereas the ought self encompasses duties and obli- to their children’s same-sex orientation (D’Augelli, Hershberger,
gations; both of these selves receive input from others as well as & Pilkington, 1998). By comparison, siblings are typically per-
from oneself (Higgins, 1987). In this connection, the Chinese ceived as more like peers and tend to be more supportive (Cain,
parenting practice of shaming (Ho, 1986) is expected to accentuate 1991).
the effect of self-discrepancies so that Chinese are more likely than On the contrary, the lesbians were more willing to disclose to
Westerners to feel ashamed in response to potentially shameful friends, whose support facilitated coming out. It should be noted
events (Tang et al., 2008), although this hypothesis is yet to be that the Perceived Social Support Scales taps support from friends
tested. In any case, this study showed that shame was moderately in general, without differentiation in sexual orientation. Neverthe-
related to outness, partly through increasing internalized hetero- less, data from the Outness Inventory suggested that lesbians were
sexism and weakening lesbian identity. Thus, those who are more quite ready to disclose to heterosexual friends, especially when
ashamed of themselves as a person are also more likely to inter- these were long-term relationships with probably a greater degree
nalize the public stigma about same-sex attraction. They are the of trust. This suggested that support was relatively forthcoming
ones who see themselves in more negative ways globally, and from age peers, although they might not share the same sexual
when they encounter prejudice and discrimination from others, orientation. Compared with the parent– child relationship, which is
they are more ready to incorporate such stereotypical views into hierarchical in nature, the relationship between friends is charac-
their already negative self. The foregoing discussion implies a link terized by equal status and mutuality that facilitate sharing and
between parenting practice and internalizing heterosexism by les- disclosure. Moreover, the younger age groups have generally more
bians in Chinese societies, an important question that should be tolerant attitudes toward same-sex orientation than the older co-
investigated in future research. horts (i.e., parents; Dube & Savin-Williams, 1999).
The relationship between social support and outness was more Although family support was not significantly related to coming
complicated than was initially assumed. We expected that per- out by way of internalized heterosexism and lesbian identity, it
ceived support from both friends and family would mediate the had, surprisingly, a small direct effect on coming out to friends in
effect of shame on internalized heterosexism. However, only per- the Hong Kong sample, but not on coming out to family members.
ceived support from friends was associated with a more positive We suspect that this finding might be peculiar to our sample of
lesbian identity through reduced internalized heterosexism, younger lesbians. As can be seen from Table 1, about half the
whereas family support did not. A possible explanation is that, Hong Kong sample was teenagers. Observations suggest that par-
under the principle of filial piety, one should honor the parents ents often treat nonconforming gender-typed behaviors during
(Cheng & Chan, 2006), and anything, such as a same-sex orien- adolescence as experimental and transient. When parents are more
tation, that brings disgrace to them and to the family name should supportive, a teenage girl may have more room exploring her
be avoided. As a result, it may be difficult for Chinese lesbians to sexuality with her peers, some of whom may self-identify as
develop a positive lesbian identity, whether the parents appear to lesbian, and some as heterosexual, at various stages of develop-
be understanding or not. The prospect of losing parental affection ment (see Diamond, 2005). Thus, such girls might report more
102 CHOW AND CHENG

Table 7 2005; Sheets & Mohr, 2009), and whether the support comes from
Standardized Direct and Indirect Effects of Different Variables friends/relatives with cross-sex orientation versus support from
on Outness, Based on Different Structural Equation Models and those with same-sex orientation, would provide a more compre-
the Joint Significance Test hensive picture of the effect of support on coming out. Finally,
although the present study demonstrated the importance of a
Standardized effect size positive lesbian identity in coming out to others, it did not reveal
Independent and mediator variable Direct Indirect Total
the dynamic process of identity development in Chinese lesbians.
Future research should address this important issue.
Outness to family Despite these limitations, this study provides valuable insights
Shame 0.00 ⫺0.18 ⫺0.18 into the psychological issues faced by Chinese lesbians day in and
Via identity ⫺0.12 day out. Whether in China or in Hong Kong, the experience of
Via IH 3 identity ⫺0.05 lesbians with regards to shame, internalized stigma, self-
Via friend support 3 IH 3 identity ⫺0.01 devaluation, and the dilemmas of coming out to others are rather
IH 0.00 ⫺0.15 ⫺0.15
similar. We suspect that similar identity and outness issues are also
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Via identity ⫺0.15


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Friend support 0.00 0.05 0.05 common among Chinese/Asian lesbians beyond these two societ-
Via IH 3 identity 0.05 ies, including those residing in foreign countries such as the United
Outness to friends: China States. Such experiences would be especially relevant for those in
Shame 0.00 ⫺0.29 ⫺0.29 the first or second generations, due to entrenched cultural attitudes
Via identity ⫺0.19 against same-sex orientation. Psychologists working with these
Via IH 3 identity ⫺0.07 populations should be aware of the potential cultural issues that
Via friend support 3 IH 3 identity ⫺0.03 they face. It is recommended that counselors should help clients to
IH 0.00 ⫺0.27 ⫺0.27
identify oppressive experiences and suggest strategies for stigma
Via identity ⫺0.27
Friend support 0.00 0.11 0.11 management. Our findings suggest that the process of constructing
Via IH 3 identity 0.11 a lesbian identity takes place in the context of deep-rooted shame
Outness to friends: Hong Kong that involves the entire self. Thus, counselors working with Chi-
nese/Asian lesbians will probably need to deal with global issues
Shame 0.00 ⫺0.23 ⫺0.23
Via identity ⫺0.14 of self-image as well as issues related directly to lesbian identity.
Via IH 3 identity ⫺0.06 Moreover, because harmony and familism (including filial piety)
Via friend support 3 IH 3 identity ⫺0.01 are highly valued in these communities, it is likely that Chinese/
Via family support ⫺0.02 Asian lesbians will have strong hesitations engaging in actions that
IH 0.00 ⫺0.18 ⫺0.18
Via identity ⫺0.18 lead to confrontations with family members, including coming out
Friend support 0.00 0.04 0.04 and avoiding “fake” heterosexual marriage. Although some Chi-
Via IH 3 identity 0.04 nese/Asian lesbians may choose to come out to family members,
Family support 0.14 0.00 0.14 others may find a dual lesbian and mainstream (e.g., a good
daughter) identity (see Fingerhut, Peplau, & Ghavami, 2005) to be
Note. IH ⫽ internalized heterosexism.
a more acceptable resolution in the long run. Counselors will need
to be sensitive to the needs of different subgroups of lesbians when
disclosures, although one should caution against interpreting the working in a multicultural setting.
finding to mean that family support led directly to disclosure to
friends. To test our assumption that the relationship between References
family support and outness to friends was especially true for the
teenage lesbians, we ran correlational analysis for the two vari- Andrews, B., Qian, M., & Valentine, J. D. (2002). Predicting depressive
ables, broken down by age groups (15–18 and 19⫹). Product– symptoms with a new measure of shame: The Experience of Shame
Scale. British Journal of Clinical Psychology, 41, 29 – 42.
moment correlations equaled .21 ( p ⬍ .05) for the former, but only
Anhalt, K., & Morris, T. L. (1998). Developmental and adjustment issues
.06 (ns) for the latter. Further controlling for friends support and of gay, lesbian and bisexual adolescents: A review of the empirical
lesbian identity (see Figure 1b) resulted in little changes to these literature. Clinical Child and Family Psychology Review, 1, 215–230.
coefficients. Thus, the way family (or parental) support works may Bailey, J. M., & Zucker, K. J. (1995). Childhood sex-typed behavior and
depend on the age of the lesbian, a topic that begs further research. sexual orientation: A conceptual analysis and quantitative review. De-
Several limitations need to be mentioned before closing. First, velopmental Psychology, 31, 43–55.
participants were mainly teenagers and young adults; future re- Beals, K. P., & Peplau, A. (2005). Identity support, identity devaluation,
search should cover a broader age range in order to gain a more and well-being among lesbians. Psychology of Women Quarterly, 29,
thorough understanding about Chinese lesbians. Second, the 140 –148.
present cross-sectional data do not allow inferences about causal- Brady, S., & Busse, W. J. (1994). The Gay Identity Questionnaire: A brief
measure of homosexual identity formation. Journal of Homosexuality,
ity, and thus cannot illustrate the dynamic nature of identity
26, 1–22.
development overtime in lesbians. Third, although it is not possi- Burnham, K. P., & Anderson, D. R. (2002). Model selection and multi-
ble to recruit representative samples of lesbians, samples of much model inference: A practical information theoretic approach (2nd ed.).
larger sizes would enhance generalizability of the results. Fourth, New York, NY: Springer-Verlag.
in addition to assessing general perceived support, the inclusion of Cain, R. (1991). Relational contexts and information management among
measures of support for same-sex orientation (Beals & Peplau, gay men. Families in Society, 72, 344 –352.
SHAME AND COMING OUT 103

Carr, C. L. (2007). Where have all the tomboys gone? Women’s accounts tity, community and sexual risk among men who have sex with men in
of gender in adolescence. Sex Roles, 56, 439 – 448. China. Journal of Homosexuality, 53, 83–115.
Carter, D. B., & McCloskey, L. A. (1983–1984). Peers and the mainte- Jordan, K., & Deluty, R. H. (1998). Coming out for lesbian women: Its
nance of sex-typed behavior: The development of children’s conceptions relations to anxiety, positive affectivity, self-esteem, and social support.
of cross-gender behavior in their peers. Social Cognition, 2, 24 –314. Journal of Homosexuality, 35, 41– 63.
Census and Statistics Department. (2007). Hong Kong 2006 population Kanuha, V. K. (1997). Stigma, identity, and passing: How lesbians and gay
by-census main report, Vols. I & II. Hong Kong, China: Author. men of color construct and manage stigmatized identity in social inter-
Cheng, S.-T., & Chan, A. C. M. (2006). Filial piety and psychological action. Dissertation Abstracts International, 58, 4811A. (UMI No.
well-being in well older Chinese. Journal of Gerontology: Psychological 9819258)
Sciences, 61B, P262–P269. Kaufman, J. M., & Johnson, C. (2005). Stigmatized individuals and the
Chia, R. C., Allred, L. J., & Jerzak, P. A. (1997). Attitudes toward women process of identity. The Sociological Quarterly, 45, 807– 833.
in Taiwan and China. Psychology of Women Quarterly, 21, 137–150. King, A. (2004). ₼⦚䤓䚍ⅲ懘⚠ [Modern turn of China]. Hong Kong:
Corrigan, P. W., & Matthews, A. K. (2003). Stigma and disclosure: Oxford University Press.
Implications for coming out of the closet. Journal of Mental Health, 12, Kitayama, S., Markus, H. R., & Matsumoto, H. (1995). Culture, self, and
235–248. emotion: A cultural perspective on “self-conscious” emotions. In J. P.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

D’Augelli, A., Hershberger, S., & Pilkington, N. (1998). Lesbian, gay, and Tangney & K. W. Fischer (Eds.), Self-conscious emotions: The psychol-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

bisexual youth and their families: Disclosure of sexual orientation and its ogy of shame, guilt, embarrassment, and pride (pp. 439 – 464). New
consequences. American Journal of Orthopsychiatry, 68, 361–371. York, NY: Guilford Press.
Diamond, L. M. (2005). A new view of lesbian subtypes: Stable versus Landolt, M. A., Bartholomew, K., Saffrey, C., Oram, D., & Perlman, D.
fluid identity trajectories over an 8-year period. Psychology of Women (2004). Gender nonconformity, childhood rejection, and adult attach-
Quarterly, 29, 119 –128. ment: A study of gay men. Archives of Sexual Behavior, 33, 117–128.
Dube, E. M., & Savin-Williams, R. C. (1999). Sexual identity development Lewis, H. B. (1971). Shame and guilt in neurosis. New York, NY:
among ethnic sexual-minority male youths. Developmental Psychology, International Universities Press.
35, 1389 –1398. Lewis, M. (1998). Shame and stigma. In P. Gilbert & B. Andrews (Eds.),
Egan, S. K., & Perry, D. G. (2001). Gender identity: A multidimensional Shame: Interpersonal behavior, psychopathology, and culture (pp. 126 –
analysis with implications for psychological adjustment. Developmental 140). New York, NY: Oxford University Press.
Psychology, 37, 451– 463. Li, L., & Orleans, M. (2001). Coming out discourses of Asian American
Eid, M., & Diener, E. (2001). Norms for experiencing emotions in different lesbians. Sexuality & Culture: An Interdisciplinary Quarterly, 5, 57–78.
cultures: Inter- and intra-national differences. Journal of Personality and Link, B. G., Cullen, F. T., Struening, E., Shrout, P. E., & Dohrenwend,
Social Psychology, 81, 869 – 885. B. P. (1989). A modified labeling theory approach to mental disorders:
Ellen, R. D. B., Joy, W. S., Amber, O., Scales, R. S., & Sue, S. (2008). The An empirical assessment. American Sociological Review, 54, 400 – 423.
positive aspects of being a lesbian or gay man. Professional Psychology: Little, T. D., Cunningham, W. A., Shahar, G., & Widaman, K. F. (2002).
Research and Practice, 39, 210 –217. To parcel or not to parcel: Exploring the question, weighting the merits.
Fingerhut, A. W., Peplau, L. A., & Ghavami, N. (2005). A dual-identity Structural Equation Modeling, 9, 151–173.
framework for understanding lesbian experience. Psychology of Women MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., &
Quarterly, 29, 129 –139. Sheets, V. (2002). A comparison of methods to test mediation and other
Friedman, R. C., & Downey, J. I. (1999). Internalized homophobia and intervening variable effects. Psychological Methods, 7, 83–104.
gender-valued self-esteem in the psychoanalysis of gay patients. Psy- Mak, W. W. S., Cheung, R. Y. M., Law, R. W., Woo, J., Li, P. C. K., &
choanalytic Review, 86, 325–347. Chung, R. W. Y. (2007). Examining attribution model of self-stigma on
Gloria, A. M., & Ho, T. A. (2003). Environment, social, and psychological social support and psychological well-being among people with HIV⫹/
experiences of Asian American undergraduates: Examining issues of AIDS. Social Science & Medicine, 64, 1549 –1559.
academic persistence. Journal of Counseling & Development, 81, 93– Mak, W. W. S., & Wu, C. F. M. (2006). Cognitive insight and causal
105. attribution in the development of self-stigma among individuals with
Harada, M. (2001). Japanese male gay and bisexual identity. Journal of schizophrenia. Psychiatric Services, 57, 1800 –1802.
Homosexuality, 42, 77–100. Mallinckrodt, M., Abraham, T. W., Wei, M., & Russell, D. W. (2006).
Herek, G. M. (2003). Why tell if you’re not asked? Self-disclosure, Advances in testing the statistical significance of mediation effects.
intergroup contact, and heterosexuals’ attitudes toward lesbians and gay Journal of Counseling Psychology, 53, 372–378.
men. In L. D. Garnets & D. C. Kimmel (Eds.), Psychological perspec- Mao, L., McCormick, J., & van de Ven, P. (2002). Ethnic and gay
tives on lesbian, gay, and bisexual experiences (2nd ed., pp. 270 –298). identification: Gay Asian men dealing with the divide. Culture, Health
New York, NY: Columbia University Press. & Sexuality, 4, 419 – 430.
Herek, G. M., Chopp, R., & Strohl, D. (2007). Sexual stigma: Putting Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications
sexual minority health issues in context. In I. H. Meyer & M. E. for cognition, emotion, and motivation. Psychological Review, 98, 224 –
Northeidge (Eds.), The health of sexual minorities: Public health per- 253.
spectives on lesbian, gay, bisexual and transgender populations (pp. Marsh, H. W., Hau, K.-T., & Wen, Z. (2004). In search of golden rules:
171–208). Berlin, Germany: Springer. Comment on hypothesis-testing approaches to setting cutoff values for
Higgins, E. T. (1987). Self-discrepancy: A theory relating self and affect. fit indexes and dangers in overgeneralizing Hu and Bentler’s (1999).
Psychological Review, 94, 319 –340. findings. Structural Equation Modeling, 11, 320 –341.
Ho, D. Y. F. (1986). Chinese pattern of socialization: A critical review. In McCarn, S. R., & Fassinger, R. E. (1996). Revisioning sexual minority
M. H. Bond (Ed.), The psychology of the Chinese people (pp. 1–37). identity formation: A new model of lesbian minority and its implications
New York, NY: Oxford University Press. for counseling and research. The Counseling Psychologist, 24, 508 –534.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance Mohr, J. J., & Fassinger, R. E. (2000). Measuring dimensions of lesbian
structure analysis: Conventional criteria versus new alternatives. Struc- and gay male experience. Measurement and Evaluation in Counseling
tural Equation Modeling, 6, 1–55. and Development, 33, 66 –90.
Jones, R. H. (2007). Imagined comrades and imaginary protections: Iden- Mueller, B., Nordt, C., Lauber, C., Rueesch, P., Meyer, P. C., & Roessler,
104 CHOW AND CHENG

W. (2006). Social support modifies perceived stigmatization in the first Szymanski, D. M., Chung, Y. B., & Balsam, K. F. (2001). Psychosocial
years of mental illness: A longitudinal approach. Social Science and correlates of internalized homophobia in lesbians. Measurement and
Medicine, 62, 39 – 49. Evaluation in Counseling and Development, 34, 27–38.
National Bureau of Statistics of China. (2002). Tabulation of the 2000 Szymanski, D. M., Kashubeck-West, S., & Meyer, J. (2008). Internalized
population census of the People’s Republic of China, Vol. 3. Beijing, heterosexism: Measurement, psychosocial correlates, and research di-
China: China Statistics Press. rections. The Counseling Psychologist, 36, 525–574.
National Coalition of Anti-Violence Programs. (2007). Anti-lesbian, gay, Tang, M., Wang, Z., Qian, M., Gao, J., & Zhang, L. (2008). Transferred
bisexual, and transgender violence in 2006. New York, NY: Author. shame in the cultures of interdependent-self and independent self. Jour-
Oyserman, D., Coon, H. M., & Kemmelmeier, M. (2002). Rethinking nal of Cognition and Culture, 8, 163–178.
individualism and collectivism: Evaluation of theoretical assumptions Tangney, J. P., & Dearing, R. L. (2002). Shame and guilt. New York, NY:
and meta-analyses. Psychological Bulletin, 128, 3–72. Guilford Press.
Peterson, T. L., & Gerrity, D. A. (2006). Internalized homophobia, lesbian Tangney, J. P., Niedenthal, P. M., Covert, M. V., & Barlow, D. H. (1998).
identity development, and self-esteem in undergraduate women. Journal Are shame and guilt related to distinct self-discrepancies? A test of
of Homosexuality, 50, 49 –75. Higgins’s (1987) hypotheses. Journal of Personality and Social Psy-
Poon, M. K., & Ho, P. T. (2008). Negotiating social stigma among gay
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

chology, 75, 256 –268.


Asian men. Sexualities, 11, 245–268.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Tangney, J. P., Wagner, P., & Gramzow, R. (1992). Proneness to shame,


Procidano, M. E., & Heller, K. (1983). Measure of perceived social support
proness to guilt, and psychopathology. Journal of Personality and Social
from friends and from family: Three validation studies. American Jour-
Psychology, 101, 469 – 478.
nal of Community Psychology, 11, 1–24.
Taylor, A. B., MacKinnon, D. P., & Tein, J.-Y. (2008). Tests of the
Rosario, M., Schrimshaw, E. W., & Hunter, J. (2005). Psychological
three-path mediated effect. Organizational Research Methods, 11, 241–
distress following suicidality among gay, lesbian, and bisexual youths:
Roles of social relationships. Journal of Youth and Adolescence, 34, 269.
149 –161. Tracy, J. L., & Robins, R. W. (2004). Putting the self into self-conscious
Satorra, A., & Bentler, P. M. (1994). Corrections to test statistics and emotions: A theoretical model. Psychological Inquiry, 15, 103–125.
standard errors in covariance structure analysis. In A. von Eye & C. C. Women Coalition of HKSAR. (2005). Sexual orientation discrimination in
Clogg (Eds.), Latent variables analysis: Applications for developmental Hong Kong. Hong Kong, China: Author.
research. (pp. 399 – 419). Thousand Oaks, CA: Sage. Wong, C., & Tang, C. S. (2004). Coming out experiences and psycholog-
Satorra, A., & Bentler, P. M. (2001). A scaled difference chi-square test ical distress of Chinese homosexual men in Hong Kong. Archives of
statistic for moment structure analysis. Psychometrika, 66, 507–514. Sexual Behavior, 33, 149 –157.
Savin-Williams, R. C., & Dubé, E. M. (1998). Parental reactions to their
child’s disclosure of a gay/lesbian identity. Family Relations, 47, 7–13.
Sheets, R. L., Jr., & Mohr, J. J. (2009). Perceived social support from Received December 19, 2008
friends and family and psychosocial functioning in bisexual young adult Revision received October 6, 2009
college students. Journal of Counseling Psychology, 56, 152–163. Accepted October 7, 2009 䡲

You might also like