Goffman ^{Reference Goffman1} identified homosexuality as falling into a stigmatised category based on moral failing (i.e. being responsible and therefore being blamed for the condition) and rooted in patriarchal and heteronormative positions as well as the main, mostly monotheistic, religions. The most commonly used term to describe the stigmatisation of lesbian, gay and bisexual (LGB) people is ‘homophobia’ as the dislike or hatred towards homosexuals. ^{Reference Sullivan2} Homophobia is often experienced by gay people from their early years at school, within their families, within their immediate daily environment and in broader society including its institutions, such as the media, workplaces or legislative bodies. Homophobia takes many forms that have been described extensively in the literature in terms of discrimination, violence, prejudice and stereotypes and harassment. ^{Reference Baker3-Reference Hunt and Jensen6} Importantly, homophobia appears to affect gay people's own beliefs and attitudes towards themselves. This is known as ‘internalised homophobia’, the phenomenon where gay people are to a variable extent unhappy about their sexual orientation because of societal homophobia. ^{Reference Hunter, Shannon, Knox and Martin7}

In general, stigmatisation of an attribute may have a negative effect on a person's positive sense of self. In the case of LGB people, there is evidence that homophobia, both perceived and internalised, affects them in terms of their overall well-being and often in terms of their self-esteem. ^{Reference Berg, Ross, Weatherburn and Schmidt8-Reference Szymanski, Chung and Balsam11} Self-esteem is an individual's own evaluation of their personal worth and, according to Rosenberg, ^{Reference Rosenberg12} it refers to the notion of a relatively stable sense of an overall or global, as it is often described, personal worth. There is a voluminous amount of work which has found a relationship between self-esteem and depression, ^{Reference Rosenberg and Leahy13} anxiety, ^{Reference Rosenberg, Schooler and Schoenbach14} anger, hostility and aggressiveness ^{Reference Baumeister, Tice and Hutton15} and life satisfaction. ^{Reference Diener and Diener16} Therefore, self-esteem is an important construct that mental health professionals need to take into account when dealing with mental health problems.

There are many recent surveys in the UK suggesting that homophobia is still widespread and experienced by gay men in their daily lives, affecting their psychological well-being. A number of studies have also shown the direct relationship between homophobia (or fear of being discriminated against) and mental health problems. ^{Reference Dick4,Reference Hunt and Jensen6} Furthermore, recent epidemiological studies found that LGB groups have a significantly higher prevalence of mental health problems than the general population; in particular, conditions including common mental disorders and alcohol and substance misuse, and attempted suicide. ^{Reference King, McKeown, Warner, Ramsay, Johnson and Cort17,Reference Warner, McKeown, Griffin, Johnson, Ramsay and Cort18} In addition to a higher prevalence of mental health problems in gay men and LGB groups in general, further evidence suggests that there are disparities in access to mental healthcare. For example, LGB people report a mixed reception from mental health services. ^{Reference King, McKeown, Warner, Ramsay, Johnson and Cort17} In particular, mental healthcare for LGB groups is very often associated with depreciation of their domestic circumstances, refusal to accept partners as next of kin, professionals' excessive curiosity about LGB lives, concern about confidentiality, and fear that their sexuality will be regarded as the ‘pathology’ requiring attention. ^{Reference Golding19,Reference McFarlane20} Along these lines, Kitzinger & Coyle ^{Reference Kitzinger, Coyle, Coyle and Kitzinger21} argue that psychological research has always presented homosexuality as a form of pathology and, when dealing with gay or lesbian issues, it is concentrated in clinical and psychotherapeutic psychology.

Societal homophobia is not evident or perceived to be at the same level across all countries or societies. There are historical, cultural and institutional factors that affect its factual or perceived prevalence. ^{Reference Berg, Ross, Weatherburn and Schmidt8,Reference Kitzinger, Coyle, Coyle and Kitzinger21,Reference Graupner22} For example, in institutional terms, there is a broad spectrum that includes countries where homosexual activity is punished to countries where same-sex marriage is recognised. Between these two ends, there are countries where homosexuality is neither illegal nor there are provisions for equal treatment of people irrespective of their sexuality. In relation to the context of this study, there is evidence that Greek society is more homophobic than the UK society ^23,24 and that Greek LGB people tend to face greater difficulties in disclosing their sexuality than Anglo-Americans. ^{Reference Fygetakis and Greene25-Reference Phellas27} In Greece, sex between men was decriminalised in 1951 while there was never any mention of lesbians in Greek law. Today, the age of consent for gay men is set at 17, higher by 2 years than for heterosexuals. There is no official recognition of same-sex couples in any terms. In the UK, although ‘homosexuality’ was decriminalised later, in 1967, legislation today does not discriminate on the basis of sexuality. Currently, same-sex couples can marry and adopt children.

It needs to be noted that metropolitan and cosmopolitan cities such as London consist of people from a variety of countries and cultures. People who live in London or Athens are not exclusively of White English or Greek ethnic background, nor have they been born and lived in either city all their life. So, the term cross-cultural should be avoided in studies such as this one; instead, cross-city comparisons resemble more and should be referred to as cross-national comparisons. The term ‘national’ entails the space and the local rules of law, it does not necessarily refer to participants' nationalities in legal terms and in this case, it can include MSM from potentially any cultural communities that live within the national space, although the existence of a predominant cultural and ethnic majority should be acknowledged.

Method

Participants

The sample included 188 MSM living in London and 173 MSM living in Athens, their mean age being 32 (range 16-64) and 27 (range 16-50) years respectively. The mean time of living in the city was longer for Athens than for London MSM. There was ethnic variation within both samples, but to a lesser degree in Athens than in London; the majority of both samples were White. In terms of sexuality, 89% of London MSM reported that they were sexually attracted to men only, with the remaining 11% being attracted to both men and women. The corresponding figures for Athens MSM were 71% and 29% respectively. In relation to educational level, about 7 in 10 London MSM and 8 in 10 Athens MSM reported to have at least a university degree. Finally, nearly half of the Athenians were recruited and completed the questionnaire online and the rest used a hard copy whereas a small majority of Londoners (56%) filled out the survey online. Table 1 provides a summary of this demographic information.

Table 1 Sample characteristics by city

	London (n = 188)	Athens (n = 173)
Age, years (mean)	32	27
Living in the city, mean	12 years 4 months	20 years 3 months
Ethnic group, %
White	56.3	86.1
Other	43.7	13.9
Sexuality, a %
Gay	89.1	71.4
Bisexual	10.9	28.6
Participation method, %
Offline	44.1	52
Online	55.9	48

a. London n = 183, Athens n = 168

As there was some scepticism as to whether living in a city for only a few years would be considered sufficient for the participants to be ‘typical’ Londoners or Athenians and because such participants may not have had informed views about how their fellow citizens view gay men in their city, differences in the responses given on all variables of the questionnaire between participants living in London for less than 5 years or more than 5 years were investigated. No statistically significant differences were found. A similar analysis could not be conducted for the Athenian sample because only a handful of participants lived in the city for less than 5 years.

Procedure

A survey took place in the capital cities of Greece and the UK in Greek and English respectively. Participants were recruited using two methods: the distribution of questionnaires at gay venues (the questionnaires were then self-completed) and through an online questionnaire advertised in several popular gay-themed websites. The study was presented as one examining gay men's views about their social environment, being part of a broader research programme into how gay men see themselves and deal with everyday issues. Potential participants had to be residents of London or Athens for at least the past 12 months. They were assured that their responses and their participation in this research project would remain entirely anonymous and they were informed of their right of withdrawal. Data collection took place over a period of the same 3 months for both cities, although the majority of offline survey data were collected over 2-week periods in each city within those 3 months when the online survey link was live.

Materials

The study was conducted through a survey that, apart from questions on demographics, included four scales: ‘disclosure of one's sexuality’, ‘perceived homophobia of the general public and of people close to the participants’, MSM's ‘internalised homophobia’, and ‘global self-esteem’. The scales, based on existing English-language scales or developed first in English, were translated to Greek and back-translated until they matched each other so that similar items were asked to both English-speaking participants in London and Greek-speaking participants in Athens. Because of the cross-national element of the study, original scales were subjected to psychometric testing to standardise the scales between the two samples. In particular, all Likert-type scales were subjected to factor analyses for the English and Greek versions separately. Cross-language structurally identical scales were formed following the factor analyses solutions as well as conceptual interpretations, and the reliability of each emerging scale was measured in terms of Cronbach's α in English and Greek separately. Details of the measurements used are given below.

Disclosure of one's sexuality

Participants were asked to report the extent to which they were ‘out’ at work, to friends and family. For example, participants had to state whether they had discussed their sexuality with all, some or none of their friends.

Perceived homophobia of the general public and of people close to the participants

Participants' perceptions of other people's homophobic feelings and attitudes were assessed using 36 items based on the Modern Homophobia Scale developed by Raja & Stokes. ^{Reference Raja and Stokes28} Eight items were preselected according to their original loadings in the factor analysis that Raja & Stokes ran. Preference was shown to items with higher loadings under each of the three factors of the authors' factor analysis solution as well as items with relevance to this study and its cross-national context. For example, an item referring to gay men being allowed to join the military rather than the item that referred to openly gay celebrities advertising products was chosen despite the lower loading of the former; the reason was that both countries have armies but there are no openly gay celebrities in Greece. Furthermore, the selected items were modified to reflect representations of homophobia; instead of using the original statements written in the first person such as ‘I wouldn't mind going to a party that included gay men’, the statements were modified to read ‘I think that most Londoners/Athenians wouldn't mind going to a party that included gay men’. The same preselected eight items were repeated four times each to capture the perceptions of our participants on how homophobic they think that (a) the general public, (b) their family members, (c) their friends, and (d) their colleagues are. So, the expression ‘most Londoners’ or ‘most Athenians’ was replaced by ‘most members of my family’, ‘most of my friends’, and ‘most of my colleagues’ accordingly. Note that the wording of these representations assessed homophobia in terms of attitudes and behaviour towards gay men only rather than sexual minorities in general. All statements were scored on a 1 to 5 Likert-type scale with 1 meaning ‘strongly disagree’ and 5 meaning ‘strongly agree’. Reliability alphas for the scales made up of 8 items each and assessing perceived homophobia of the 4 different groups of people in both London and Athens were good and varied from 0.76 to 0.93.

Internalised homophobia

Mayfield's ^{Reference Mayfield29} Internalized Homonegativity Inventory was used, including its three factors referring to ‘personal homonegativity’ (e.g. ‘I feel ashamed of my homosexuality’), ‘gay affirmation’ (e.g. ‘I believe being gay is an important part of me’) and ‘morality of homosexuality’ (e.g. ‘I believe it is morally wrong for men to be attracted to each other’). Factor analyses run for each sample confirmed the existence of these factors. One item was excluded from the ‘morality of homosexuality’ factor as it was reducing the α of the Greek scale below the 0.60 level; the same item had to be removed from the English version for equivalence. The alphas of the three factors were 0.69, 0.77, 0.90 and 0.63, 0.76, 0.88 for the English and Greek versions respectively.

Self-esteem scale

Finally, Rosenberg's ^{Reference Rosenberg12} Global Self-Esteem Scale consisting of ten items was used in full to provide a measure of the participants' perception of self-worth. The items of the original scale were rated on a 4-point ‘strongly disagree’ to ‘strongly agree’ scale but we added a middle fifth option of ‘neither disagree nor agree’ to increase variance in the data. A single-factor solution was produced for both city samples with reliability α 0.88 for London and α 0.87 for Athens.

Results

In analysing the data, missing values met within any section of this survey were not treated in any way, and cases with missing values were excluded analysis by analysis. Because numbers of valid cases for each analysis conducted were adequate, treating of missing values with the potential to affect results was seen as erroneous.

Initially, in establishing potential differences in the views and attitudes of Athens and London participants themselves, an important observation was that there were statistically significant differences between the two city groups in relation to the disclosure of their sexuality to other people (Fig. 1). The scores obtained on this measure were subjected to a 3×2 chi-squared analysis. Differences between the samples were found to be highly significant when discussion with family members (χ² = 45.25, d.f. = 2, P<0.001) and friends (χ² = 39.15, d.f. = 2, P<0.001) was concerned and as far as hiding (χ² = 65.96, d.f. = 2, P<0.001) or revealing (χ² = 72.46, d.f. = 2, P<0.001) sexuality from colleagues at work or university was concerned.

Fig. 1 Sexuality disclosure in the study sample.

(a) Discussed sexuality with family members; (b) Discussed sexuality with friends; (c) Hiding sexuality from colleagues; (d) Colleagues know about one's sexuality.

A multivariate analysis of variance (MANOVA) was then conducted to investigate the differences seen in Table 1 between London and Athens MSM in their views on how they think the general local public and people close to them see gay men (Table 2). Overall, Athens MSM perceived other people's homophobia, whether general public, friends, family members or colleagues, to be higher compared with London MSM perceptions. There was a significant overall difference between the two populations on the combined four dependent variables (F _{(4, 298)} = 36.63, P<0.001; Wilks's λ = 0.67, partial η² = 0.33). In the separate analyses for each dependent variable, all differences between Londoners and Athenians were also found to be highly statistically significant (Table 3).

Table 2 Descriptive statistics of the Likert-type variables of the study

5-point Likert-type variables (n items)	London n (mean) s.d.	Athens n (mean) s.d.
Perceived homophobia
General public (8)	176 (2.38) 0.59	166 (3.14) 0.63
Friends (8)	176 (1.75) 0.63	156 (2.40) 0.68
Family (8)	175 (2.37) 0.96	156 (3.23) 0.83
Colleagues (8)	173 (1.98) 0.73	155 (2.67) 0.73
Gay men's personal homonegativity (11)	184 (1.87) 0.72	167 (1.95) 0.70
Gay men's negative views on morality of homosexuality (4)	185 (1.25) 0.46	167 (1.37) 0.51
Gay affirmation (7)	186 (3.72) 0.66	166 (3.60) 0.67
Self-esteem (10)	172 (3.92) 0.68	155 (4.10) 0.65

Table 3 Between-subjects effects on the variables referring to perceived homophobia of others

Variables	F a	ηp 2
Perceived homophobia of general public (8)	111.889	0.271
Perceived homophobia of friends (8)	85.349	0.221
Perceived homophobia of family (8)	75.872	0.201
Perceived homophobia of colleagues (8)	72.925	0.195

a. d.f. = 1, d.f. for error = 301, P = 0.000 for all variables.

In investigating the differences between the two samples in relation to ‘internalised homophobia’ as again seen in Table 1, a one-way between-groups MANOVA showed that there was a narrowly statistically significant difference between Londoners and Athenians on the combined dependent variables (F _{(3, 340)} = 2.65, P<0.05; Wilks's λ = 0.98, partial η² = 0.02). When the results for the three dependent variables were considered separately, Athens MSM scored significantly higher only on the ‘morality of homosexuality’ variable (F _{(1, 342)} = 6.545, P<0.05, partial η² = 0.02).

Third, an independent-samples t-test was conducted to compare the self-esteem scores (Table 2) for the London and Athens samples. There was a significant difference in scores between the two groups with Athenians reporting a higher self-esteem than Londoners (t = −2.421, d.f. = 325, P<0.05, r γ λ = 0.13).

Correlational analysis showed, as expected, positive relationships between most perceived and internalised homophobia scales and negative relationships between homophobia and self-esteem scales. Online tables DS1 and DS2 show that these findings are relatively consistent across the two city samples. Further regression analyses were carried out to test the relationship of both internalised and perceived homophobia of others controlling for city. Correlations showed that factors for each scale were significantly associated with each other; this was expected given that they are subscales of the same construct. Therefore, and to avoid multi-collinearity, the composite scores of the scales were used. To test whether city explained the relationship between homophobia (internalised and perceived) and self-esteem, a stepwise hierarchical regression was conducted. The first step tested the relationship between homophobia and self-esteem and the second step tested whether the addition of city had a significant impact on the model. The inter-correlations between ‘self-esteem’ and ‘perceived homophobia of others’ as well as ‘internalised homophobia’ were significant (r = −0.15, P<0.05 and r = −0.27, P<0.001 respectively). The correlation between perceived and internalised homophobia was also significant (r = 0.29, P<0.001). Model 1 is statistically significant (adjusted R ² = 0.070, P<0.001) but self-esteem is explained significantly only by internalised homophobia (Table 4). City, in model 2 has a significant impact on the relationship between internalised and perceived homophobia of others and self-esteem (adjusted R ² = 0.12, P<0.001). In particular, city explains an additional 5% of the model. Moreover, both homophobia scales in the model independently explain self-esteem significantly.

Table 4 Explanation of self-esteem by homophobia scales (composite scores) and the role of city context

Independent variables	Standardised β	T	P⩽
Step 1
Perception of homophobia in others	−0.079	−1.403	n.s.
Internalised homophobia	−0.242	−4.313	~0
F(2, 321) = 13.1, R = 0.275, R 2 = 0.076, adjusted R 2 = 0.070, P<0.001
Step 2
Perception of homophobia in others	−0.228	−3.479	0.001
Internalised homophobia	−0.210	−3.805	~0
City context	0.260	4.147	~0
F(3, 320) = 14.9, R = 0.351, R 2 = 0.123, adjusted R 2 = 0.12, P<0.001.