Victimisation and life satisfaction of gay and bisexual individuals in 44 European countries: the moderating role of country-level and person-level attitudes towards homosexuality

Abstract

We examined the link between victimisation and life satisfaction for 85,301 gay and bisexual individuals across 44 European countries. We expected this negative link to be stronger when the internalised homonegativity of the victim was high (e.g. because the victim is more vulnerable) and weaker when victimisation occurs in countries that express intolerance towards homosexuality (e.g. because in such contexts victims expect victimisation more and they attribute it to their external environment). Additionally, we expected internalised homonegativity to relate negatively to life satisfaction. Multilevel analyses revealed that victimisation (i.e. verbal insults, threats of violence, minor or major physical assaults) and internalised homonegativity were negatively related to life satisfaction. Furthermore, as we expected, the negative link between victimisation and life satisfaction was stronger when high internalised homonegativity was reported (and the interaction effect occurred for verbal insults and major assaults as outcome variables), while it was weaker when there was low national tolerance of homosexuality (and the interaction effect occurred for verbal insults and for minor assaults). Future research and social policy should consider how the consequences of victimisation are dependent on personal as well as national attitudes towards homosexuality.

Résumé

Nous avons examiné le lien entre la victimisation et la satisfaction de vivre chez 85,301 personnes gays et bisexuelles dans 44 pays européens. Nous nous attendions à ce que ce lien négatif soit plus marqué lorsque l’homonégativité intériorisée est forte (par exemple, parce que la personne qui souffre de cette victimisation est plus vulnérable), et plus faible lorsque la victimisation se manifeste dans les pays où l’intolérance vis-à-vis de l’homosexualité est exprimée (par exemple, parce que dans ce contexte, de manière plus forte, les victimes s’attendent à la victimisation et l’attribuent à l’environnement externe). De plus, nous anticipions un impact négatif de l’homonégativité intériorisée sur la satisfaction de vivre. Nos analyses multiniveaux ont révélé que la victimisation (c’est-à-dire, les insultes verbales, les menaces de violence, les agressions physiques mineures ou sévères) et l’homonégativité intériorisée avaient un impact négatif sur la satisfaction de vivre. En outre, comme nous nous y attendions, le lien négatif entre victimisation et satisfaction de vivre était plus marqué en présence d’une forte homonégativité intériorisée rapportée (et les effets d’interaction pour les insultes verbales et les agressions sévères se sont révélés en tant que variables des résultats), tandis qu’il était moins marqué dans un contexte de faible tolérance de l’homosexualité à l’échelle nationale (et les effets d’interaction se sont révélés pour les insultes verbales et les agressions mineures). À l’avenir, la recherche et les politiques sociales doivent prendre en compte la manière dont les conséquences de la victimisation dépendent des attitudes à la fois personnelles et nationales vis-à-vis de l’homosexualité.

Resumen

Aquí analizamos el vínculo entre victimización y satisfacción con la vida de 85.301 homosexuales y bisexuales de 44 países europeos. Previmos que este vínculo negativo sería más fuerte cuando la víctima presenta un nivel alto de homonegatividad interiorizada (p. ej., porque la víctima es más vulnerable) y más débil cuando la victimización ocurre en países que expresan intolerancia hacia la homosexualidad (p. ej., porque en tales contextos las víctimas esperan que la victimización sea mayor y la atribuyen a su entorno externo). Además, esperábamos que la homonegatividad interiorizada estuviese relacionada negativamente con la satisfacción con la vida. En análisis de varios niveles se observó que la victimización (es decir, insultos verbales, amenazas de violencia, agresiones físicas leves o graves) y la homonegatividad interiorizada estaban negativamente relacionadas con la satisfacción con la vida. Asimismo, tal como esperábamos, el vínculo negativo entre victimización y satisfacción con la vida era más fuerte cuando se informaba una alta homonegatividad interiorizada (y el efecto de interacción ocurría en casos de insultos verbales y agresiones graves como variables de resultado), mientras que era más débil cuando había un nivel bajo nacional de tolerancia hacia la homosexualidad (y el efecto de interacción ocurría en casos de insultos verbales y agresiones leves). En los estudios futuros y las políticas sociales debería tenerse en cuenta el modo en que las consecuencias de la victimización dependen tanto de la actitud personal como de las actitudes nacionales hacia la homosexualidad.

Keywords:

• Anti-gay victimisation
• life satisfaction
• internalised homonegativity
• Europe
• minority stress

Introduction

The victimisation of gay individuals is still a widespread phenomenon in Europe (ILGA-Europe 2015). Victimisation may be defined as ‘harms that occur to individuals because of other human actors behaving in ways that violate social norms’ (Finkelhor and Kendall-Tackett 1997, 2). Consequently, gay victimisation refers to being harmed by another person who acts outside the norms of acceptable behaviour due to the sexual orientation of the victim (Bradbury et al. 2016). Following previously proposed typologies of gay victimisation (e.g. D’Augelli, Hershberger, and Pilkington 1998), for the purposes of this paper we define gay victimisation as negative acts, including verbal insults, threats of violence and minor or major physical assaults, experienced by victims due to their sexual orientation and/or sexual identity. Such intimidating behaviours are known to have detrimental effects on the well-being, mental health and life satisfaction of lesbian, gay, bisexual and transsexual persons (Hershberger and D'Augelli 1995).

The aim of this paper is to examine further the previously identified (Bachmann and Simon 2014) relationship between anti-gay victimisation and life satisfaction (see Figure 1 for our hypothesised model).

Figure 1. Our hypothesised model.

First, drawing on the minority stress model (Meyer 1995), we expect this relationship to be negative and, moreover, stronger when individuals report high internalised homonegativity because they are more influenced by the victimisation acts. Second, drawing on views on the self-protective properties of social stigma (Crocker and Major 1989; Crocker and Quinn 2000) and social identity theory (Tajfel and Turner 1979), we expect this relationship to be weaker in countries that have a low tolerance of homosexuality. This is because victimisation in such countries is more expected and likely to be attributed by the victim to national norms rather than his or her own beliefs. Finally, although not the primary focus of our paper, we additionally acknowledge the previously theorised and identified negative link between internalised homonegativity and life satisfaction (Meyer 1995).

Anti-gay victimisation and life satisfaction

According to the minority stress model (Meyer 2003), being gay in a heterosexist society leads gay individuals to experience minority stressors, namely, distressful events as a result of their social minority position. Victimisation acts can constitute such typical distressful events. Victimisation does not always have to be a major, or explicit, incident. Implicit behaviours, ranging from seemingly benign remarks such as ‘That’s so gay!’ (Charlesworth 2015) to overhearing homophobic remarks about oneself (Jewell et al. 2012), also count as victimisation acts, albeit of a minor character. Victimisation also includes more serious acts, such as threats and physical violence (Mustanski, Andrews, and Puckett 2016). No matter how aggressive these behaviours are, they all have something in common: they remind the victims that they belong to a minority group, a group of lower status and power. According to the minority stress model, this evokes in the victim deep feelings of rejection (Meyer 1995) and fails to address the basic human needs of gay individuals (Meyer 2003). In other words, being victimised provides gay individuals with fewer chances to experience social recognition (Bachmann and Simon 2014) and a meaningful life (Janoff-Bulman and Frieze 1983), which could lead to a global feeling of life dissatisfaction.

The typology of gay victimisation used in this paper comprises verbal insults, threats of violence and (minor or major) physical assaults. Verbal violence is intimidating since it represents a symbolic form of violence acting as a reminder of actual physical violence (Garnets, Herek, and Levy 1990). Because of its power and symbolisms, the threat of violence can be as distressing as actual violence (Rogers and Kelloway 1997). Finally, physical victimisation is often what verbal aggression escalates to (Maher 2007) and is clearly more overt and intense. It is likely to lead to post-assault trauma as well as continuous vigilance and decreased psychosocial functioning (Willis 2008). Moreover, it is more likely to manifest against sexual minority rather than sexual non-minority groups, and thus could explain their poor mental health (Friedman et al. 2011) and the fact that they experience consequences as extreme as suicide and suicide ideation (Eisenberg and Resnick 2006). All in all, victimisation makes it more difficult for gay individuals to view their lives as meaningful, autonomous, predictable and worthy (Garnets, Herek, and Levy 1990) and is therefore expected to relate negatively to life satisfaction:

Hypothesis 1: Anti-gay victimisation, including acts such as verbal insults (1a), threats of violence (1b), minor physical assaults (1c) and major physical assaults (1d), experienced by gay individuals negatively relates to their life satisfaction.

The role of internalised homonegativity

According to the minority stress model (Meyer 2003), being gay in a heterosexist society not only leads to distal stressors such as those addressed above (i.e. victimisation) but also to proximal stressors, such as internalised homonegativity. Internalised homonegativity has been defined as ‘the internal reaction to stigma associated with being homosexual’ (Ross et al. 2008, 547). By inducing self-blame and guilt (Morandini et al. 2015) and impairing psychosocial functioning (Rowen and Malcolm 2003; Van Wijngaarden and Ojanen 2016), internalised homonegativity has a detrimental impact on gay individuals’ life satisfaction. This proposition is rooted in the minority stress model (Meyer 1995) and it is also supported by extensive empirical evidence (for a review of the literature, see Berg, Munthe-Kaas, and Ross 2016; Newcomb and Mustanski 2010). Although our paper acknowledges this link, our scope and contribution lies more within our aim to address internalised homophobia as a moderator (rather than a predictor) in the link between victimisation and life satisfaction.

Although both proximal and distal minority stressors induce stress, proximal stressors can also moderate the link between distal stressors and well-being outcomes. For instance, prejudice events are more detrimental for victims trying to adjust to internalised homonegativity experiences (Meyer 1995). This moderating link is addressed by the minority stress model but hardly ever tested by empirical research. Two reasons have been proposed by Meyer (2003) as to why this link occurs. First, victims with low internalised homonegativity (i.e. high self-acceptance) are more resilient against stress. This is supported by stress theories (e.g. Lazarus and Folkman 1984), according to which individuals perceive stress to the extent that they appraise that an event exceeds their capacities to deal with it. Resilient individuals (e.g. individuals with low internalised homonegativity) are thus more capable of cognitively reframing and undermining the negative events. Second, individuals with low internalised homonegativity have more opportunities to feel solidarity with other gay individuals (e.g. Sowe, Brown, and Taylor 2014), and are consequently able to collectively undermine and make sense of the events in a more positive way.

Similarly, interview studies among lesbian, gay, and bisexual youth show that individuals who have been stigmatised are more likely to cope with the stigmatism in a healthy way when they hold positive attitudes towards their homosexuality (van Bergen and Spiegel 2014). Therefore, based on all the aforementioned arguments, we expect internalised homonegativity to not only relate negatively to the victim’s life satisfaction, but to also moderate the link between victimisation and life satisfaction in the following ways:

Hypothesis 2: Internalised homonegativity relates negatively to the life satisfaction of gay individuals.

Hypothesis 3: The negative relationship between anti-gay victimisation, namely verbal insults (3a), threats of violence (3b), minor physical assaults (3c) and major physical assaults (3d), and life satisfaction is stronger when internalised homonegativity is higher.

The moderating role of national tolerance levels of homosexuality

The question of when victimisation is most detrimental to the victims would receive an incomplete answer if one views the victims’ attitudes towards homosexuality only as internal and stable, irrespective of the national or social context in which they live. Indeed, existing theoretical and empirical work has tried to place stigmatisation and victimisation phenomena within the social context. This constitutes an attempt to explain mixed empirical evidence on the self-efficacy of stigmatised groups, which is inconclusive as to whether being stigmatised actually affects one’s feelings of self-worth (Gray-Little and Hafdahl 2000). To address such findings, Crocker and Quinn (2000) have, for example, proposed that stigmatised targets reflect upon their stigma in two distinct ways: either on the basis of their own stable (and perhaps internalised) beliefs about their stigma, or on the basis of their situation or the social beliefs around their stigma. While positive internalised beliefs of the victim protect him/her from stigmatisation, and thus also from victimisation (as we have argued above), the role of social beliefs around stigma is somewhat less straightforward. To that end, Crocker and Major (1989) proposed that social stigma (i.e. negative public attitudes against a certain social group) may, in fact, protect the targets of stigmatisation. This is because victims within such aggressive or intolerant societies have learned to expect those victimisation acts and are, therefore, less affected by them. Furthermore, they are more likely to attribute experienced victimisation to social injustice and public attitudes rather than to their inner self, thus protecting their feelings of self-worth (Crocker and Quinn 2000). This line of reasoning correlates with social identity theory (Tajfel and Turner 1979), which suggests that when an in-group is stereotyped by an out-group, the targets of the stereotyping identify more with their in-group, which has been found to protect their well-being (Latrofa et al. 2009).

These propositions have been applied and tested through empirical research on a wide range of discrimination types, such as racial or sexual discrimination. For instance, racial discrimination within any life domain of African-American individuals has a stronger detrimental effect on their mental health if those individuals have a low (rather than high) awareness and experience of racism, because they have fewer opportunities to understand and make sense of the racist events as a social phenomenon (Fischer and Shaw 1999). In other words, the mere awareness of the fact that a minority group member lives in a society that holds predominantly racist views may lead him or her to experience fewer detrimental consequences of future racism. Similarly, in one cross-cultural study it was found that anti-gay workplace bullying had more detrimental effects on gay employees’ satisfaction in Australia, where such behaviours are not tolerated, than in Singapore, where such behaviours can be viewed as more common because of cultural and organisational values of power distance (Loh, Restubog, and Zagenczyk 2010). Therefore, we expect that the victimisation of gay individuals will have fewer detrimental effects on their life satisfaction for those living in countries with low tolerance of homosexuality, because in these countries victimisation is more expected and, therefore, less intimidating.

Hypothesis 4: The negative relationship between anti-gay victimisation, namely, verbal insults (4a), threats of violence (4b), minor physical assaults (4c) and major physical assaults (4d), and life satisfaction is weaker when national tolerance of homosexuality is lower.

Methods

Sample

The respondents to our survey comprised 85,301 men, living in 44 European countries, who were users of a global dating website. Most of them identified as gay individuals (85.7%), while the rest identified as bisexual (14.3%). Their mean age was 38.2 (SD = 12.4). The majority came from Germany (33.8%), followed by Italy (11.6%) and France (8.2%; see Table 1 for a country-by-country breakdown). Most respondents lived in a metropolis (28%), city (20.7%) or town (21.1%). When asked about their current relationship status, most stated that they were not currently in a committed relationship (56.8%), some that they were in a relationship with a man (35.7%) or a woman (4.4%), while others indicated ‘other’ (1.5%) or preferred not to answer (1.6%). Half of the respondents had received a university-level education (50.7%), followed by secondary/high school (34.2%), basic school (15.8%) and no formal education (0.7%).

Table 1. Participation and mean scores/frequencies of the study variables per country (N = 85,301 individuals, 44 countries).

Country	Total number of respondents	%	Verbal assaults (%)	Threats of violence (%)	Minor physical (%)	Major physical (%)	Internalised homonegativity [0–6]	National tolerance of homosexuality [1–10]	Life satisfaction [0–6]
Germany	28822	33.8	17.0	5.0	2.4	1.0	1.52	5.69	4.03
Italy	9856	11.6	11.7	2.6	1.6	0.8	2.41	3.79	3.29
France	6997	8.2	14.0	5.4	2.6	0.9	2.02	5.65	3.74
Spain	3670	4.3	10.6	2.2	1.4	0.5	1.73	6.01	3.92
Switzerland	3091	3.6	12.4	3.1	2.1	0.7	1.61	6.35	4.32
The Netherlands	2967	3.5	13.4	4.6	2.6	0.6	1.49	7.53	4.34
Greece	2847	3.3	15.7	4.4	1.7	0.8	2.53	3.71	3.36
Belgium	2715	3.2	14.0	4.5	2.5	1.4	1.80	5.83	3.98
Romania	2502	2.9	23.6	9.1	4.5	1.8	2.84	2.10	3.25
Austria	2462	2.9	19.0	6.1	2.9	1.1	1.66	5.42	4.12
Hungary	2091	2.5	16.8	7.2	2.2	0.9	2.07	3.26	2.71
Poland	1984	2.3	27.9	9.5	4.9	1.9	2.42	2.86	3.20
Turkey	1768	2.1	18.9	9.3	7.1	3.7	2.81	1.48	2.96
Serbia	1719	2.0	23.1	11.2	6.1	3.1	2.62	1.82	2.45
United Kingdom of Great Britain and Northern Ireland	1492	1.7	17.8	5.4	2.4	1.3	1.56	5.40	3.68
Russian Federation	1292	1.5	25.1	13.5	7.3	3.6	2.63	2.23	3.78
Bulgaria	668	0.8	25.3	8.4	3.9	1.3	2.64	2.78	3.25
Finland	654	0.8	13.8	3.1	1.5	0.8	1.40	6.67	4.20
Sweden	606	0.7	11.9	4.0	1.8	0.8	1.29	7.76	4.25
Croatia	547	0.6	19.2	7.7	6.2	2.9	2.43	2.49	3.15
Czech Republic	542	0.6	7.4	3.0	2.8	0.7	1.86	4.85	4.19
Norway	508	0.6	8.9	2.8	1.6	1.0	1.30	7.18	4.25
Portugal	505	0.6	14.9	5.0	4.0	0.8	2.21	3.68	3.52
Denmark	466	0.5	9.9	3.2	3.0	2.4	1.29	7.25	4.38
Bosnia and Herzegovina	416	0.5	22.4	10.1	6.7	2.6	3.05	1.73	2.56
Ireland	413	0.5	15.5	3.9	2.4	1.9	1.70	5.20	3.62
Slovakia	396	0.5	17.2	6.1	3.8	1.8	2.21	4.79	3.59
Slovenia	386	0.5	17.6	3.1	3.9	1.3	2.24	3.91	3.50
Ukraine	353	0.4	22.4	11.6	5.7	2.8	2.58	1.61	3.58
Cyprus	345	0.4	20.9	7.5	4.9	3.2	2.65	2.19	3.16
Latvia	283	0.3	22.3	7.8	3.5	2.5	2.75	2.41	3.53
Luxembourg	280	0.3	11.1	2.9	2.1	0.4	1.68	6.51	4.37
Estonia	280	0.3	25.0	6.1	3.2	1.8	2.20	2.30	3.69
The former Yugoslav Republic of Macedonia	240	0.3	22.9	9.6	10.0	5.4	2.83	2.05	2.64
Lithuania	235	0.3	28.1	7.2	8.5	2.6	2.57	1.95	3.40
Malta	198	0.2	15.2	3.0	2.0	0.5	1.90	3.87	3.78
Montenegro	122	0.1	18.0	17.2	9.8	4.9	2.78	1.73	2.45
Iceland	118	0.1	5.1	0.8	0.8	0.8	1.47	8.34	4.37
Belarus	116	0.1	25.0	8.6	5.2	0.9	2.82	2.74	3.61
Republic of Moldova	101	0.1	26.7	11.9	6.9	3.0	3.02	1.72	3.14
Georgia	78	0.1	34.6	19.2	11.5	7.7	2.98	1.14	3.18
Albania	69	0.1	30.4	13.0	13.0	4.3	3.49	2.11	2.58
Kosovo	54	0.1	29.6	14.8	14.8	14.8	3.25	1.29	2.46
Armenia	47	0.1	29.8	23.4	12.8	4.3	3.47	1.19	2.80

Note

Descriptive statistics may differ from Berg, Lemke, and Ross (2017) because of different methods of dealing with missing data.

Procedure

All users of the dating site planetromeo.com were invited to participate in an online survey by a large banner placed on the start page of the dating site as well as by three individual messages to each member (two were sent in December 2014 and one in February 2015). While originally founded in Germany, the site planetromeo.com has since expanded to become a global platform. As with most dating sites or dating apps, it is characterised by the following: users have to create a personal profile page (using a nickname) that is visible to other users within the closed system. The profile page documents a range of personal data along predefined categories, including information that is not visible during face-to-face encounters, while a search engine allows users to filter other members based on these categories (e.g. location, gender, age, sexual preference, etc.). Users can display the profiles of other users sorted by offline local structures (e.g. members from the same city, by physical proximity in ascending order, etc.). Finally, a messaging system allows private communication between two users. The use of Internet dating is particularly popular among gay and bisexual individuals (Bolding et al. 2005; Davis et al. 2008); indeed, Miller (2015, 479) describes dating sites for gay and bisexual men as ‘the modern-day gay bar’. For gay and bisexual men, online dating sites are not just a medium to facilitate finding a romantic partner. Many men keep their profiles online even when they are in a long-term relationship in order to stay in touch with friends and to find new sexual or social friends (as mentioned above, in our sample only 57% declared that they were single at the time of the survey). For these reasons, studies that have focused on different questions related to gay and bisexual men frequently used these sites as the main or even the sole forum for recruiting subjects (see also Gudelunas 2012; Miller 2015; Lemke and Weber 2017 for the role played by dating sites among gay and bisexual men). Such dating sites facilitate the recruitment of gay and bisexual men who keep their same-sex sexual attraction secret in offline surroundings (and thus avoid offline gay bars and respective places) in particular (Lemke and Weber 2017). Taken together, we are confident that our recruitment method gives wide access to the population under examination.

Anonymity as well as voluntary participation were explained and guaranteed to all participants. The data we report here form part of a larger research project involving a longer questionnaire on personal and situational factors connected to life satisfaction levels of gay and bisexual men (Lemke, Tornow, and planetromeo.com 2015). In total, all 1.8 million members of the dating site were addressed. A total of 165,319 participants began to answer the survey. Data were excluded from the survey sample if the questionnaire was not completed; the questionnaire was completed in less than 300 s; or the participants did not indicate their country of residence, age or gender. This led to a remaining sample of N = 118,165 respondents from 86 countries and thus a response rate of 6.4%. For the present analysis, we further reduced the sample to include only those men who chose bisexual or gay as their sexual orientation (thereby excluding anyone who refused to choose a label or who chose heterosexual despite being sexually attracted to men and those who identified as transgender; N = 114,452 remaining). Because of the European scope of the present paper, only the European sub-sample is used. Specifically, we decided to retain only the 44 European countries that had an available score on public opinion towards homosexuality according to the European Values Study (EVS 2010). Thus, a sample of N = 85,832 participants from 44 European countries remained. Finally, 531 subjects were excluded due to their not responding to questions of their city size or education because these variables, together with age and sexual orientation, were used as deck variables in a hot deck imputation procedure (Myers 2011), leading to a final sample of N = 85,301 respondents.

The survey was translated by translators of the dating website into 25 different languages. Respondents could thus choose only one of them before they started the survey. The majority chose German (38.7%), followed by Italian (11.6%), French (10.4%), English (9.3%), Dutch (4.6%), Spanish (4.2%), Greek (3.5%), Russian (2.6%), Hungarian (2.7%), Romanian (2.8%), Polish (2.6%), Serbian (2.4%), Turkish (2%) and the remaining 3% chose other languages (each amounting to less than 1%). Further details can be found in Lemke, Tornow, and planetromeo.com (2015).

Measures

Anti-gay victimisation. Following existing literature on this topic (e.g. D’Augelli and Grossman 2001), we measured anti-gay victimisation using a checklist that comprised four categories, namely verbal insults, threats of violence, minor physical assaults and major physical assaults. The question that this checklist referred to was ‘Have you ever experienced victimisation due to your sexual orientation and/or gender identity?’ For each category, respondents could indicate whether they experienced it more than a year ago, less than a year ago or never. Participants could omit any of the four categories they did not want to answer. If participants had missing values for one or two categories, those values were imputed via a hot deck procedure based on the remaining three or two items as deck variables (Myers 2011). If participants had missing values for three or all four categories of victimisation, the missing values for those three or four items were imputed via hot deck procedure based on participants’ education, sexual orientation (gay vs. bisexual), age and size of city of residence as deck variables. For the analyses, for each category we created one categorical variable comparing victimisation experiences in the last year (1) to those occurring more than a year ago or never (0).

Internalised homonegativity. To assess internalised homonegativity we used a modified six-item scale (Smolenski et al. 2010) based on the Internalised Homonegativity Scale developed by Ross and Rosser (1996). Sample items included statements such as ‘I would prefer to be solely or more heterosexual’ or ‘I feel comfortable about being seen in public with an obviously gay person’ (reverse). The answering scale ranged from 0 = does not apply to me, to 6 = applies to me, with five (unlabelled) intermediate points enabling participants to provide an answer between those two extremes. Cronbach’s alpha was .79. Participants could omit any item they did not want to answer. If participants had missing values for one or two items, those values were imputed via a hot deck procedure based on the remaining four or five items as deck variables (Myers 2011). Participants were then assigned a mean score for all items representing their value of internalised homonegativity. If participants had missing values for three or more items in this scale, they received no scale value. In those cases, their value for internalised homonegativity was also imputed via a hot deck procedure (Myers 2011) based on their education, sexual orientation (gay vs. bisexual), age and size of city of residence as deck variables.

Life satisfaction. To measure life satisfaction, we used the five-item scale by Diener et al. (1985). Again, items (e.g. ‘I am satisfied with my life’ or ‘If I could live my life over, I would change almost nothing’) were rated on an answering scale ranging from 0 = strongly disagree, to 6 = strongly agree, with five (unlabelled) intermediate points. Cronbach’s alpha was .94. Once again, participants could omit any item they did not want to answer. If participants had missing values for one or two items, those values were imputed via a hot deck procedure based on the remaining three or four items as deck variables (Myers 2011). Participants were then assigned a mean score for all items representing their satisfaction with life. If participants had missing values for three or more items in this scale, they received no scale value. In those cases, their value for satisfaction with life was also imputed via a hot deck procedure based on their education, sexual orientation (gay vs. bisexual), age and size of city of residence as deck variables.

National tolerance of homosexuality. To capture national tolerance levels of homosexuality in an objective and independent way we used existing data from the 2008 European Values Study (EVS 2010). In this survey, a large number of respondents (ranging between 1000 and 2384) from the general population of each of the 44 countries used in the present paper reported the extent to which they justify homosexuality. Answers ranged from 1 = never to 10 = always, meaning that a high EVS score represents higher tolerance of homosexuality. To assess the stability of this score, we compared the 2008 scores with the 1999 scores on the same question for the 33 countries that had available scores on both surveys. The average difference of the 2008 minus the 1999 score was 0.19 (indicating a slight improvement in national attitudes over time), but this is statistically non-significant. The correlation between the 1999 and the 2008 scores was r = .93, p < .001. This reveals that national attitudes seem to be rather stable, suggesting that current attitudes should not be very different from those reported in 2008.

Analytic approach

Because our data comprise a multilevel structure (i.e. participants nested within countries), we used MlwiN to conduct multilevel regression analysis. Life satisfaction, victimisation and internalised homonegativity were all within-country variables (i.e. at the within-level), while national tolerance of homosexuality was a between-country variable (i.e. at the between-level, namely, comprising the same value for all respondents within the same country). All predictor variables were centred to the grand mean (Hox 2002). Because older gay men tend to adapt better to life challenges (Fingerhut, Peplau, and Gable 2010), we used age as a control variable in the analysis. The regression analysis (predicting life satisfaction) was built on the basis of three nested models comprising successively the intercept (Model 0); age, the within-country victimisation dummy-coded variables, the within-country internalised homonegativity and the between-country tolerance of homosexuality (Model 1); and all interaction terms between victimisation, on the one hand, and internalised homonegativity and national tolerance of homosexuality, on the other hand (Model 2). Because certain countries were overrepresented in the data (e.g. Germany), we conducted parametric Bootstrapping in MlwiN with 5 sets of 300 replicates.

Results

Table 1 shows how each country scored in relation to the victimisation variables, internalised homonegativity, national tolerance of homosexuality and life satisfaction, and reveals considerable variation among the examined countries. Table 2 shows the means, standard deviations and intercorrelations for all the study variables.

Table 2. Means, standard deviations and intercorrelations for study variables (N = 85,301 individuals, 44 countries).

	Within-level		1	2	3	4	5	6	7	8	Between-level
	M	SD									M	SD
1. Age	38.15	12.39	–	−.75**	−.70**	−.68**	−.49**	−.90**	.89**	.81**	35.61	4.82
2. Verbal insults	.17	.37	−.18**	–	.81**	.78**	.60**	.82**	−.83**	−.67**	.19	.07
3. Threats of violence	.05	.23	−.10**	.45**	–	.88**	.68**	.80**	−.76**	−.70**	.07	.05
4. Minor physical assaults	.03	.17	−.07**	.30**	.51**	–	.84^**	.82**	−.73**	−.73**	.05	.03
5. Major physical assaults	.01	.11	−.03**	.17**	.31**	.46**	–	.61**	−.56**	−.59**	.02	.02
6. Internalised homonegativity [0–6]	1.93	1.52	−.11**	.04**	.03**	.04**	.02**	–	−.92**	−.85**	2.27	.62
7. National tolerance of homosexuality [1–10]	4.89	1.53	.27**	−.06**	−.07**	−.06**	−.05**	−.28**	–	.85**	3.88	2.13
8. Life satisfaction [0–6]	3.74	1.60	.18**	−.14**	−.11**	−.09**	−.06**	−.32**	.25**	–	3.51	.59

Note:

Correlations below diagonal refer to the within-level (N = 85,301 individuals) and correlations above the diagonal refer to the between-level (N = 44 countries). All victimisation variables (variables 2–5) are coded with 1 = experienced within the last year and 0 = experienced more than a year ago or never; descriptive statistics may differ from Berg, Lemke, and Ross (2017) because of different methods of dealing with missing data.

* p < .05

** p < .01.

Table 3 presents the results of the multilevel regression analysis. As can be seen in Model 1, all four victimisation variables were positively and significantly related to life satisfaction, providing support to Hypotheses 1a–1d. Similarly, internalised homonegativity negatively related to life satisfaction, providing support to Hypothesis 2. Although not hypothesised, national tolerance levels of homosexuality positively related to life satisfaction.

Table 3. Unstandardised estimates, standard errors and Bootstrap estimates for models with life satisfaction as outcome variable (N = 85,301 individuals, 44 countries).

Model variables	M1			M2
	B	SE B	Bootstrap estimates	B	SE B	Bootstrap estimates
Intercept	3.86**	.05	3.76/3.96	3.85**	.05	3.76/3.96
Age	.01**	.00^a	.01/.01	.01**	.00^a	.01/.01
Verbal insults	−.39**	.02	−.41/−.36	−.38**	.02	−.41/–.35
Threats of violence	−.23**	.03	−.32/−.20	−.24**	.03	−.29/–.19
Minor physical assaults	−.21**	.04	−.29/−.13	−.24**	.04	−.32/–.15
Major physical assaults	−.11*	.05	−.22/−.01	−.12*	.06	−.25/–.01
Internalised homonegativity	−.28**	.00^a	−.28/−.27	−.27**	.00^a	−.28/–.26
National tolerance of homosexuality	.12**	.02	.08/.16	.13**	.02	.09/.17
Verbal × homonegativity				−.04**	.01	−.06/–.02
Verbal × tolerance				−.03*	.01	−.05/–.01
Threat × homonegativity				.03	.02	−.01/.06
Threat × tolerance				.00	.02	−.03/.04
Minor × homonegativity				.01	.03	−.06/.10
Minor × tolerance				−.06*	.02	−.10/–.02
Major × homonegativity				−.07*	.03	−.13/–.00^a
Major × tolerance				−.03	.03	−.08/.02
−2 × log		305535.26			305485.46
Δ −2 × log		9111.63**			49.80**
Δdf		7			8
		R^2			R^2
Within-level variance		10%			10%
Between-level variance		75%			75%

Note

All victimisation variables are coded with 1 = experienced within the last year and 0 = experienced more than a year ago or never.

^a The zero is a result of rounding.

* p < .05

** p < .01.

As can be seen in Table 3, overall Model 2 had significantly better fit to the data, compared to Model 1, Δχ²(8) = 49.80, p < .001. Four out of the eight hypothesised interaction effects were found to be significant. Those included the interactions between: (1) verbal insults and internalised homonegativity; (2) verbal insults and national tolerance of homosexuality; (3) minor assaults and national tolerance; and (4) major assaults and internalised homonegativity.

To interpret these interactions, we conducted four simple slope tests that revealed the following results. First, the negative link between verbal insults and life satisfaction was stronger when homonegativity was 1 SD above the mean (estimate = –.45, S.E. = .05, p < .001) than when it was 1 SD below the mean (estimate = –.32, S.E. = .05, p < .001; see Figure 2). Second, the link between major assaults and life satisfaction was negative and significant when homonegativity was 1 SD above the mean (estimate = –.23, S.E. = .07, p < .05), but it was non-significant when it was 1 SD below the mean (estimate = –.02, S.E. = .08, p = .81; see Figure 3). Taken together, these findings provide support to Hypotheses 3a and 3d respectively.

Figure 2. The relationship between verbal insults and life satisfaction moderated by internalised homonegativity.

Figure 3. The relationship between major physical assaults and life satisfaction moderated by internalised homonegativity.

Third, the link between verbal insults and life satisfaction was weaker when national tolerance was 1 SD below the mean (estimate = –.34, S.E. = .02, p < .001) than when it was 1 SD above the mean (estimate = –.43, S.E. = .02, p < .001; see Figure 4). Fourth, the link between minor physical assaults and life satisfaction was weaker when national tolerance was 1 SD below the mean (estimate = –.16, S.E. = .05, p < .001) than when it was 1 SD above the mean (estimate = –.33, S.E. = .06, p < .001; see Figure 5). Taken together, these findings provide support to Hypotheses 4a and 4c.

Figure 4. The relationship between verbal insults and life satisfaction moderated by national tolerance of homosexuality.

Figure 5. The relationship between minor physical assaults and life satisfaction moderated by national tolerance of homosexuality.

All in all, it seems that the link between victimisation and life satisfaction is stronger when homonegativity is high and weaker when national tolerance of homosexuality is low. It should be noted that the link is always negative and significant at both extremes (high and low) of national tolerance, while this is not the case for internalised homonegativity (i.e. at low levels of homonegativity the link between major physical assaults and life satisfaction becomes non-significant).

Discussion

In this study, we expected anti-gay victimisation and internalised homonegativity to relate negatively to gay individuals’ life satisfaction levels. We also expected the negative link between victimisation and life satisfaction to be stronger when internalised homonegativity is high and weaker when national tolerance of homosexuality is low. The results partly supported our hypotheses: anti-gay victimisation and internalised homonegativity were negatively related to life satisfaction. The expected moderation by internalised homonegativity occurred for instances of verbal insults and major physical assaults, while the expected moderation by national tolerance of homosexuality occurred for verbal insults and minor physical assaults.

The negative links that we found between anti-gay victimisation and internalised homonegativity, on the one hand, and life satisfaction, on the other, are in line both with previous empirical findings (Berg, Munthe-Kaas, and Ross 2016; D’Augelli and Grossman 2001) and theoretical propositions such as the minority stress model (Meyer 2003). The interaction effects also provide support to both the minority stress model (Meyer 2003) and theoretical views on the self-protective properties of social stigma (e.g. Crocker and Major 1989). These frameworks suggest, respectively, that discriminating events are more detrimental when they target minorities with self-stigma (i.e. who are vulnerable) and they are less detrimental when they occur within intolerant environments (i.e. because in such environments the targets expect, rationalise and attribute such events to social norms rather than to themselves).¹ These findings highlight that victimisation is processed by the victims not only on the basis of their own stable beliefs but also on the basis of their context, suggesting that victims often choose to cope with the victimisation by attributing it to wider societal norms and attitudes (Bogart et al. 2016). This is in line with insights from fairness (Folger and Cropanzano 2001) and social processing theories (Salancik and Pfeffer 1978), suggesting that individuals do not simply use internal standards to judge whether events are fair and should be tolerated but they interpret such events through the lenses of their context and its norms. For example, employee motivation has been found to be more severely affected by social undermining behaviours at work when those occur within work groups where such behaviours are common compared to work groups where such behaviours are uncommon (Duffy et al. 2006).

One notable pattern in our results was that the only victimisation type that was moderated both by internalised homonegativity and by national tolerance was verbal insults. One explanation for this could be that the more intimidating victimisation becomes (e.g. on a continuum from verbal to physical), the more difficult it is for its effects to be moderated. That, however, does not explain why the effects of threats of violence were not moderated (while the effects of minor and major physical victimisation were). Existing typologies of anti-gay victimisation (e.g. D’Augelli, Hershberger, and Pilkington 1998) do not always explicitly distinguish between verbal victimisation and threats of violence, and it could be argued that the two victimisation types overlap with one another. Therefore, it could be that since the interaction effects generally seem to be smaller than the main effects in our study, the moderation of the effects of threats of violence is already partly explained or reflected by the moderation of the effects of verbal insults.

Last but not least, while minor physical victimisation was moderated by national tolerance, major physical victimisation was moderated by internalised homonegativity. Even more importantly, a remarkable interaction pattern occurred in our study only for major physical assaults: while the effect on life satisfaction was negative for high internalised homonegativity, the effect became non-significant for low internalised homonegativity. This implies that gay individuals with a healthy sexual identity (i.e. low internalised homonegativity) have the power to mitigate the effects of victimisation events as aggressive as major physical assaults. A possible interpretation could be that a form of victimisation that is as distressing and confronting as major physical assaults can only be moderated when the victim is resilient and has a particularly positive self-image (i.e. low internalised homonegativity). A healthy self-image and a high level of self-efficacy has been suggested and found to be a protecting factor against several types of serious traumatic events because it empowers victims to engage in proactive coping and to avoid ruminating painfully about the events (Benight and Bandura 2004).

Limitations and future research

One limitation of our study was that our measured variables were not temporally separated, therefore causal links cannot be inferred. Second, our model proposed and found that victimisation impacts upon a rather global outcome, i.e. life satisfaction, which is, of course, also determined by several other factors. This could explain why many of the effects we found are rather small. Future studies should control and account for multiple other factors that shape the life satisfaction of lesbian, gay, bisexual and transsexual individuals. Finally, our results cannot be generalised to non-European contexts.

Implications for practice

The most straightforward advice that can be given on the basis of our findings is that societies should educate people from a very early age (e.g. at schools, sports or leisure associations and through parents) so as to promote a healthy sexual identity among the lesbian, gay, bisexual and transsexual population. Teaching lesbian, gay, bisexual and transsexual youth how to deal with their identity in a healthy and non-judgemental way is one of the most important tools they can have when dealing with future life challenges. This is because a positive internalised attitude towards homosexuality not only enhances life satisfaction, but also has the potential to act as a buffer against victimisation or other discriminating acts and to even fully mitigate the effects of major physical assaults.

Additionally, our results suggest that tolerant environments may, in fact, exacerbate the effects of homophobic events. Curiously, such tolerant environments where people tend to believe that homophobic events ‘would never happen’ should be the most attentive to victimisation. This is because victimisation in such contexts is particularly unexpected and, therefore, shocking, and may have the power to disorient and intimidate victims in a very powerful way. The main step towards being attentive to victimisation is defining what victimisation exactly means and urging social institutions to be alert towards it, monitor and report it. This will help ensure that victimisation, even those forms commonly characterised as ‘benign’ will no longer be justified or tolerated.

Disclosure statement

No potential conflict of interest was reported by the authors.

Acknowledgements

Grateful acknowledgement is owed to Joran Jongerling (Erasmus University Rotterdam) for statistical advice.

Notes

1. To clarify this point and to ensure that any wrong conclusions are avoided, this does not indicate that an intolerant environment is actually a better environment for gay men to live in. Both individual experiences with victimisation as well as a lower national tolerance of homosexuality have a negative impact on life satisfaction levels. Moderation analysis shows that in environments with a low national tolerance of homosexuality, which in itself is associated with low levels of life satisfaction among gay men, victimisation experiences have a smaller (further) effect on their satisfaction with life. In contrast, in environments with high national tolerance of homosexuality, which is associated with higher life satisfaction among gay men, victimisation experiences have a greater negative impact on life satisfaction levels among gay men.