Gender-Related Attitudes toward Homosexuality in Greece

The purpose of this study is to investigate the relevance of the “Attitudes toward Lesbians and Gay Men” (ATLG) Scale developed by G. M. Herek to the Greek society. The study consists of two stages or sub-studies. At the first stage, the sample consisted of 186 undergraduate university students and at the second, 254 undergraduate university students, who studied at the Department of Physical Education and Sport Science, Democritus University of Thrace in Komotini, Greece. Methodologically speaking, our research relied on the methods of descriptive statistics, exploratory factor analysis, confirmatory factor analysis, reliability analysis (Cronbach’s α, composite reliability and average variance extracted) and t-test for independent sample. The results of the first stage showed that the two factors, men’s homophobia and women’s homophobia, accounted for 58% of the total variance. At the second stage, three confirmatory factor analyses were performed: men’s homophobia, women’s homophobia and total homophobia. We also found genderrelated differences in students’ attitudes to homosexuality, but only as far as male homosexuality is concerned.

Introduction 2016). There are even less studies dealing with this topic in relation to sports (Georgiou et al., 2018(Georgiou et al., , 2019Grigoropoulos, 2010;Grigoropoulos & Kordoutis, 2015). We believe, however, that this is an interesting topic which is worthy of attention since the majority of people in sports are young adults-the category that constitutes the core of the Greek society-and, therefore, their opinion is quite illustrative of the prevailing public attitudes to LGBT people. The contribution of this study to the existing field is that for the first time in Greece, it is based on the data from various sports. It also relies on a more thorough statistical analysis to examine the reliability and validity of the "Attitudes toward Lesbians and Gay Men" (ATLG) Scale (Herek, 1984) as an essential methodological tool. Thus, the resulting data provide a more comprehensive and complete picture of the situation in Greece related to the issue of homophobia in sports.
In most European countries, the majority of people are tolerant toward homosexuality (on average, about 70%: Spain 88%, Germany 87%, the Czech Republic 80%, France 77%, the UK 76%, Italy 74%). Poland is the only notable exception, where the percentage of people who reported being tolerant to homosexuality is 42% while 46% of people expressed negative attitudes towards homosexuality (Kohut, 2013). In Greece, the situation resembles that of Poland: only a half of the citizens consider themselves to be LGBT-friendly, with women being generally more tolerant to homosexuality than men. As for age differences, younger people tend to be more open-minded when it comes to sexual orientation, while older people seem to be more conservative (Kohut, 2013).
There have been a number of more specialized surveys focusing on specific population groups in Greece (see, for example, Drydakis, 2009;Grigoropoulos, 2010;Grigoropoulos & Kordoutis, 2015;Grigoropoulos et al., 2010;Karellou, 2003;V. Papadaki & E. Papadaki, 2011;Papadaki et al., 2015). Unfortunately, in comparison with other European countries, Greece ranks among the least tolerant countries in relation to homosexuality. There is an overall report, which reveals that the levels of homophobia and discrimination against the LGBT community are very high in Greece (Pavlou, 2009). Unfortunately, many homosexual Greeks find it very difficult to translate their sexual desires and behaviours into a political statement of homosexual identity even today (Zervoulis, 2016). This is partly due to the change in the geopolitical situation of the country compared to the past. Modern Greece is a Mediterranean country in the south-eastern part of Europe. It is located at the crossroads between the West and East, conveying the experiences of its past and being influenced by both Europe and the Middle East. Greece's war history with other great civilizations of the wider Middle East (such as the Egyptians, Assyrians and especially the Persians) and the Ottoman occupation, which lasted 400 years, and more in some areas, have severely affected the cultural background of the country (Zervoulis, 2016). Let us not forget that about 1.25 million Greeks were expelled from Turkey (Fygetakis, 1997). At the same time, 50,000 inhabitants lived in Athens, while today they are over 3 million. All these expelled people brought customs and beliefs with them, one of which certainly was the different types of their sexual attitude. These customs and beliefs were influenced due to the coexistence of these people with other cultural groups in the area (Zervoulis, 2016).

Homophobia and Gender
Gender is regarded as a critical factor that affects attitudes to homosexuality and homophobia. According to D'Augeili and Rose (1990), Wright et al. (1999), Warriner et al. (2013), Moral de la Rubia and Valle de la O (2014), Pavlica et al. (2016), men are homophobic to a greater degree than women. All of these studies considered young adults, in particular high school and college students. According to previous researchers (Moral de la Rubia & Valle de la O, 2014;Pavlica et al. (2016), the focus on this age is important, because it is the period when people form their social opinion. Therefore, the researchers of this study decided to use the same age group sample.

Methodological Tools to Measure Attitudes to Homosexuality in Sports
To measure tolerance toward homosexuality in sports, several questionnaires were used (Piedra et al., 2017). Among the most common ones are the "Attitudes towards Lesbian and Gays" (ATLG) (Herek, 1984), and the Modern Homophobia Scale (Raja & Stokes, 1998). These two methodological tools were designed to measure the attitudes towards sexual minorities in general and not specifically in sports. They came to be regarded as universal instruments and were applied in a range of social and cultural contexts (Cárdenas & Barrientos, 2008;Collier et al., 2015;Moral de la Rubia & Valle de la O, 2014;O'Brien et al., 2013;Rosik, 2007). As far as sports are concerned, the most commonly used tools are the "Perceptions of Homophobia" Rosik (2007) and "Heterosexism in Physical Education scale" (PHHPE) (Morrow & Gill, 2003), measuring the perception of homophobia and heterosexism (discrimination in favour of heterosexual and against homosexual people) among teachers (PHHPE-TS) and students (PHHPE-SS).

Research Objective
This study was considered necessary to address the perceived research gap due to the lack of studies on attitudes to homosexuality and homophobia in Greece and, most importantly, the lack of suitable methodological tools. The main aim of the study is to adapt Herek's "Attitudes toward Lesbians and Gay Men Scale" (ALTG) (Herek, 1984) to the Greek society. The research comprised two stages or sub-studies.

st stage
At this stage, the purpose was to investigate the structural validity and reliability of the ATLG Scale for the Greek context.

Sample
The participants were 186 undergraduate students of the School of Physical Education & Sport Science of the Democritus University of Thrace in Komotini, Greece (74, 39.8% men and 112, 60.2% women). The age of the respondents ranges between 18 and 23 years old. The sexual orientation of the sample remains unknown because researchers did not consider it necessary and did not intend to draw comparisons based on the sexual orientation of the respondents. This clarification was added to the guidance text given to the research participants.

The ATLG Scale
The majority of research on sexual prejudice in relation to sports had mostly used the methodology developed by Herek (1984)-the ATLG Scale (see, for example, A. R. Anderson & Mowatt, 2013;Ensign et al., 2011;Gill et al., 2006;O'Brien et al., 2013;Oswalt & Vargas, 2013;Roper & Halloran, 2007;Sartore & Cunningham, 2009). Another tool was called the "Modern Homophobia Scale" (Forbes et al., 2002), and it was used to measure the attitudes towards sexual minorities in sports. However, these scales were not designed to measure negative attitudes in a particular setting or environment, where there may be different types of stereotypes and prejudices associated with the traditional homophobic climate.
The ATLG Scale was originally elaborated in English (Herek, 1984) but it has been adjusted to other languages as well, such as Spanish (Cárdenas & Barrientos, 2008), and it has been used for research purposes in many countries such as the United States (LaMar & Kite, 1998), the UK (Hegarty, 2002), Canada (Mohipp & Morry, 2004), Chile (Cárdenas & Barrientos, 2008), Mexico (Moral de la Rubia & Valle de la O, 2013, 2014) and Ghana (Norman et al., 2016). The questionnaire consists of 20 questions. The first ten deal with the attitudes towards homosexual men (ATG subscale) and the other ten, with the attitudes towards homosexual women (ATL subscale). The respondents' choice of the degree of agreement or disagreement is expressed by using the 5-grade Likert-type scale (from "strongly disagree" to "strongly agree"). The highest scores, close to 5, refer to greater identification with the statement, while the lowest, close to 1, refer to greater rejection of the statement. Participants were also asked to provide information about their gender and age.

Statistical Analyses
The following statistical analyses were performed: an exploratory factor analysis (EFA) was used to examine the validity and reliability of the questionnaire. Factor reliability (Cronbach's α) was examined by conducting reliability analysis. Descriptive statistics were calculated to broadly examine the degree of homophobia.
The authors of this study proposed that if some of the factors turn out inappropriate, control of data and variables' suitability could be seen as the first stage of factorial analysis. The following statistical criteria were used to ensure this control: the partial correlation coefficient, which is controlled with the value of KMO (Kaiser-Meyer-Olkin) and Bartlett's test of Sphericity (Hair et al. 2009;Teixeira et al., 2020), and the Measure of Sampling Adequacy (MSA).
The KMO index takes values from 0 to 1. When the range of this value is between 0.8 and 1, the sampling is adequate. When this value is less than 0.6, the sampling is not adequate, and remedies should be taken. In some cases, limit is set at 0.5 (Glen, 2016). Regarding MSA values, rates .9 and .8 are the most appropriate, concerning that values from .6 to .7 are acceptable but not so reliable, and those which are less than or equal to .5 have to be deleted and not be taken into consideration at the analysis to come (Hair et al., 1998).

Suitability of Data and Variables
The results showed that the statistical criterion of Kaiser-Meyer-Olkin is remarkably high (.939). Furthermore, the Bartlett's Test of Sphericity rejects the zero hypothesis that the correlation's table is the only one (the value of control function 2358.520, degrees of freedom 190, and p = .000). Data analysis indicated that the survey's data are suitable for a factorial analysis.
In order to check if all the indicators are appropriate for this model, the value of the "Measure of Sampling Adequacy" (MSA) has been considered. According to the results, all the indicators are within the limits of the MSA criteria with the index ranging between .879 and .963 (Hair et al., 1998).

Exploratory Factor Analysis
One of the objectives of the paper was to test the psychometric properties of the scale. A Principal Component Analysis (Exploratory Factor Analysis) with varimax rotation was performed using SPSS 26 to test the factor structure of the scale in the Greek context. As shown in Table 1, the two factors that emerged from the analysis accounted for 58% of the total variance. These factors corresponded to the two dilemmas (Table 1).

Reliability Analysis
The values for Cronbach's α were calculated to assess the internal consistency reliabilities of "Men's Homophobia" (.93), and "Women's Homophobia" (.89). The results indicated that both scales showed acceptable internal consistency since Cronbach's α was higher than .88.

Discussion
The purpose of the first stage was to examine the structural validity and internal consistency of the "Attitudes Toward Lesbians and Gay Men" Scale in the Greek environment.
Concerning this tool's structural validity, the exploratory factor analysis of the questionnaire has brought to light the two factors proposed by Herek (1988). As can be expected from these foreseeable results, ours contradict those of earlier studies, such as those of LaMar and Kite (1998), who found 4 factors, Cárdenas and Barrientos (2008) in Chile, who found 5, Moral de la Rubia and Valle de la O (2013) in Mexico, who found 3. Concerning the differences resulting from the analyses, we believe that this difference may be due to the cultural and geographical differences between the countries surveyed. Other factors may have contributed to this differentiation of results. For instance, in Cárdenas and Barrientos (2008), a 6-grade Likert-type scale is used while we use the traditional 5-grade scale. Furthermore, since the study of LaMar and Kite (1998) was conducted quite long ago, this may also have caused the differences in the results.
We found that both scales, for men and women, showed acceptable high internal consistency. As for the degree of homophobia, the results indicated that Greek men are friendlier towards lesbians than male homosexuals. This result can be based on men's erotic fantasies about the relationship between two women (Louderback & Whitley, 1997). We believe that this explanation may also apply to people in Greece and to the social stereotypes that prevail in Greek society as the idea of sexual encounters between women is more widespread and accepted than between two men. In contrast with men's behaviour, women's attitude to male and female homosexuality is more conciliatory. Greek women appear to be more openminded than men, a hypothesis that works cooperatively with the available research evidence (Cárdenas & Barrientos, 2008; Moral de la Rubia & Valle de la O, 2014). Most importantly, the outcomes of the study have shown the suitability and validity of the ATLG scale in the Greek context and we renamed the adjusted scale the "Range of Homophobia Scale" (RHS).

Research Goals
The purpose of this stage was to confirm the factor structure and reliability of the Greek version (RHS) of the ATLG Scale (Herek, 1984).

Sample
The participants were 254 undergraduate students of the School of Physical Education & Sport Science of the Democritus University of Thrace in Komotini, Greece (102, 40.2% men and 152, 59.8% women). The age of the respondents ranges between 18 and 23 years old. The sexual orientation of the respondents remains unknown because we did not intend to draw any comparisons based on their sexual orientation. This clarification was added to the guidance text given to the respondents. Another reason why we chose not to ask the respondents about their sexual orientation was that the requirement to disclose information about their sexuality might have affected their answers.

Tools
The instrument which was used was the adjusted ATLG Scale renamed as the RHS questionnaire

Statistical Analyses
The following statistical analyses were performed. Confirmatory factor analysis (CFA) was used to examine the validity and reliability of the questionnaire. Factor reliability (Composite Reliability and Average Variance Extracted) was examined by conducting a reliability analysis. Descriptive statistics were calculated to broadly examine the degree of satisfaction.
Confirmatory factor analysis: the method it was used for estimation was the Maximum Likelihood (ML). The adjustment (fit) indices that have been taken into account and their acceptable values are the following: namely minimum discrepancy (CMIN or χ 2 ), degrees of freedom (df), minimum discrepancy divided by the degrees of freedom (χ 2 /df) less than 5, Root Mean Square Error of Approximation (RMSEA) less than .08, Standardized Root Mean Square Residual (SRMR) less than .05, incremental indices Comparative Fit Index (CFI) greater than .90, Normed Fit Index (NFI) (Bentler, 1990;McDonald, & Marsh, 1990) and Parsimony Normed Fit Index (PNFI) values equal to or greater than .60. The control of the internal consistency of the factors was performed using the index composite reliability (Aguirre-Urreta et al., 2013;Hair et al., 2019). The reliability of the factor is acceptable when the index takes values greater than or equal to .70. To assess the discriminant and convergent validity, the Average Variance Extracted (AVE) index was considered. Values of greater than or equal to .50 are acceptable.

Confirmatory Factor Analyses
Three confirmatory factor analyses were performed using LISREL 8.80. The method of estimating the parameters was that of maximum likelihood (Bentler, 1990;Platsidou, 2001).

Men's Homophobia
The hypothesized model consists of one latent variable, namely "Men's homophobia". The results of the confirmatory factor analysis demonstrated that the hypothesized model produced a significant chi-square, χ 2 (254) = 126.87, χ 2 /df = 3.639, p < .05. The NFI and CFI were found to be .94 and .95 respectively. The RMSEA was also used to assess the degree of fit of the model. The RMSEA value for the hypothesized model was found to be .065 and SRMR = .038.

Women's Homophobia
The hypothesized model consists of one latent variable, namely "women's homophobia". The results of the confirmatory factor analysis demonstrated that the hypothesized model produced a significant chi-square, χ 2 (254) = 115.36, χ 2 /df = 3.30, p < .05. The NFI and CFI were found to be 0.96 and 0.97 respectively. The RMSEA was also considered to assess the degree of fit of the model. The RMSEA value for the hypothesized model was found to be .063 and SRMR = .041.

Total Homophobia
The confirmatory analysis was performed to check whether the 20 questions could be a factor in examining total homophobia. The theoretical model consists of one latent variable, namely "Total homophobia". The results of the confirmatory factor analysis demonstrated that the hypothesized model produced a significant chi-square, χ 2 (254) = 826.96, χ 2 /df = 4.86, p < .05. The NFI and CFI were found to be .94 and .96 respectively. The RMSEA was also used to assess the degree of fit of the model. The RMSEA value for the hypothesized model was found to be .062 and SRMR = .044.

Composite Reliability and Average Variance Extracted
As shown in Table 3, all factors showed particularly good reliability since the CR index takes values .939 for men's homophobia and .912 for women's homophobia. In terms of discriminant and convergent validity, the AVE index showed a satisfactory value for men's homophobia (.608). In contrast, the factor "women's homophobia" showed marginally acceptable values (.514) ( Table 3).
As shown in Table 4, men are more homophobic than women when it comes to male homosexuality. On the contrary, we found no gender-related differences in the level of homophobia towards lesbians.
Regarding total homophobia, no statistically significant gender-related differences are observed (see Table 5).

Discussion
The purpose of the study was to confirm the factor structure and reliability of the Greek version (1 st stage) of the Scale of the "Attitudes towards Lesbians and Gay Men", which means that it is relevant for the survey in Greece.
The theoretical model of confirmatory factor analysis included investigation of the three factors: the degree of homophobia against male homosexuality, the degree of homophobia against female homosexuality, and homophobia regardless of the gender.
Confirmatory factor analysis has shown the validity of the proposed theoretical model. The scale proposed by Herek (1984) and the one we developed at the first stage can also be used as a single scale measuring homophobia regardless of the gender. The internal consistency of both the overall scale and the two factors was extremely high, considering the results that were similar to those of the first stage.   Both the stability of the internal reliability of the two factors and the scale as a whole, as well as the stability of the averages presented in the second survey, show that the "Attitudes toward Lesbians and Gay Men" Scale is a reliable measure of homophobia in Greece. Additionally, the validated new scale was renamed into the "Range of Homophobia Scale" (RHS) based on the ATLG scale and intended for the Greek context.

Conclusion
This study has shown the reliability of the ATLG scale and subscales in the Greek context, thus addressing the perceived gap in the current research on this topic (Papadopoulos, 2019). This tool was used to measure the attitudes of Greek people towards homosexual men and women by focusing on the students majoring in physical education and sports. Taking into account the fact that only 22% of the citizens in Greece feel comfortable about watching a public kiss (or even a handshake) between two people of the same gender (Papaioannou, 2019), the research on attitudes to homosexuality is more relevant than ever.
We found that the adapted version of the ATLG scale and the "Range of Homophobia Scale" (RHS) we developed are suitable for use in Greece. We believe that these methodological tools hold a lot of potential for further research on this topic.
In this research, the age of the respondents in our sample is known but not their sexual orientation. The gender distribution of the sample corresponds to the percentages of people of each gender in the Greek society according to the data from the population census (ELSTAT, 2011). It is suggested that the questionnaire should be used by other researchers in the following surveys focusing on other social groups in Greece, in order to expand the results and have more comprehensive research data.