Peer Aggression and Sexual Harassment among Young Adolescents in a School Context: A Comparative Study between Finland and Turkey

The study investigates peer aggression and sexual harassment among young adolescents in Finland and Turkey. Sex differences and the interaction effect between country of residence and sex are also examined. A questionnaire was completed by 1,747 adolescents (1, 268 from Finland, 479 from Turkey, Mage = 14.1). Six different forms of aggression (physical, verbal, indirect, cyber, verbal sexual harassment, physical sexual harassment) were examined. More adolescents from Turkey, and more boys, were found to be involved in aggression as both victims and perpetrators compared to adolescents from Finland and girls. The interaction effect was significant between country of residence and sex with being a boy from Turkey was related to having the highest involvement in cyber aggression, verbal sexual harassment, and physical sexual harassment, as both victim and perpetrator. Regarding victimization from indirect aggression, girls from Finland scored higher than Turkish girls, while boys from Turkey scored higher than Finnish boys.

nvolvement in aggression is a risk factor for the psychosocial adjustment of adolescents (Moore et al., 2013;Nansel et al., 2004;Özdemir & Stattin, 2011). Victims of aggression report poorer emotional adjustment, poorer relationships with peers, and higher depression and anxiety (Nansel et al., 2004;Undheim & Sund, 2010), while perpetrators of aggression report poorer school adjustment and higher levels of depression than non-involvers (Moore et al., 2013;Nansel et al., 2004). The studies on relationship between perpetration of aggression and depression have varying results. Some found that perpetrators had lower depression than non-involvers (Juvonen et al., 2003), while others found higher depression rates among perpetrators (Chang et al., 2013;Yen et al., 2014). Despite of varying results regarding depression, perpetration of aggression is related to low anxiety (e.g. Juvonen et al., 2003;Yen et al., 2014). Aggression is also associated with negative social relationships among the adolescents. In a study conducted in 25 countries, victims and perpetrator/victims in all countries were found to have poorer peer relationships than others, and the same was the case for perpetrators of aggression in 15 countries (Nansel et al., 2004). Involvement in aggression and sexual harassment has been found to be associated with poor attachment to peers, teachers, and parents (Doty et al., 2017;Gruber & Fineran, 2016). However, high peer connectedness has also been found to be associated with perpetration of sexual harassment (Doty et al., 2017).
Victims of sexual harassment have been found to have higher scores on depression, anxiety, substance abuse, and self-harm, and lower school satisfaction and academic engagement than non-victim youth (Bucchianeri et al., 2014;Gruber & Fineran, 2016). Furthermore, victimization from sexual harassment has been found to have a greater effect than bullying on adolescents' poor school outcomes (Gruber & Fineran, 2016). Adolescents who perpetrated sexual harassment were found to have higher levels of depression, alcohol and drug abuse, and aggressive behavior than adolescents who were not involved in sexual harassment in any way (Clear et al., 2014;Lacasse & Mendelson, 2007;Rinehart et al., 2017). Nevertheless, victims of sexual harassment report higher levels of alcohol and drug abuse than others (Clear et al., 2014;Lacasse & Mendelson, 2007). In addition to that, adolescents are more likely to perpetrate sexual harassment if they themselves are victimized from it (Moyano et al., 2017;Munoz-Rivas et al., 2009). 203 management, and acceptance of aggression were found to predict involvement in aggression (Chaux et al., 2009).
Nationwide anti-bullying programs and how they are implemented might reflect different outcomes regarding adolescents' aggression. All types of aggression and sexual harassment were found to decrease in schools that were part of the KiVa anti-bullying program in Finland (Salmivalli et al., 2011). A study compared two nationwide anti-bullying programs that had similar structure and methodology, one in Norway and the other in Ireland (Midthassel et al., 2009). Norwegian authorities were found to implement the program more actively than Irish authorities, where the program was carried out by psychologists, researchers, and school authorities, all receiving salary for their work, while the Irish program was carried out by teachers who volunteered for the program. Thus, the participating adolescents from Norway were found to score lower on aggression after the program than the adolescents from Ireland.

Sex Differences in Aggression and its Concomitants
A large number of studies investigating sex differences in adolescents' involvement in various forms of aggression have found that boys overall perpetrate more aggression and sexual harassment than girls Espelage et al., 2012;Gruber & Fineran, 2016). Boys have also been found to be both victims and perpetrators of direct aggression more often than girls (Barzilay et al., 2017;Lundh et al., 2014;Mehari et al., 2019). Studies have found varying results regarding sex differences on indirect aggression. According to a recent study among adolescents from ten European countries, girls reported more victimization from indirect aggression (Barzilay et al., 2017). On the other hand, some studies found that boys were more likely to become victims and perpetrators of indirect aggression than girls (Salmivalli & Kaukiainen, 2004;Wang et al., 2015). Still, extremely indirectly aggressive groups of pupils consisted of girls only (Salmivalli & Kaukiainen, 2004). In a review (Björkqvist, 2018), it was concluded that in proportional terms (proportional scores of total aggression scores), girls score always higher than boys on indirect aggression, as indirect aggression is their preferential style of aggression. Since the total aggression scores of boys are higher than those of girls, their scores on indirect aggression may be higher than those of girls, although it is not their preferential style. In a meta-analysis, it was found that 204 girls were more likely to perpetrate cyber aggression than boys during early and mid-adolescence, while it was found to be the other way around during late adolescence (Barlett & Coyne, 2014). Rejection by peers was found to be associated with the perpetration of physical and verbal aggression among girls (Cheng, 2009;Salmivalli et al., 2000), while it was associated with only verbal aggression among boys (Cheng, 2009;Salmivalli et al., 2000). On the other hand, the perpetration of indirect aggression was found to increase social acceptance among boys (Salmivalli et al., 2000).
Studies have found varying results regarding sex differences in sexual harassment. In Finland, girls reported more victimization from physical and verbal sexual harassment (Kaltiala-Heino et al., 2018), while in the Netherlands, boys reported more victimization from verbal sexual harassment (Timmerman, 2003). A study from Canada did not find any sex differences regarding victimization from sexual harassment, while boys were found to perpetrate more sexual harassment than girls (McMaster et al., 2002).
The aim of the current study was to compare the levels of both victimization from and perpetration of peer aggression and sexual harassment among a sample of Finnish and Turkish adolescents in a school context. Sex differences in aggression were also investigated. Sexual harassment is rarely studied in school contexts, and comparative studies with adolescent samples are limited in this area. The two countries were selected partly for reasons of convenience, but also because Finland has effective nationwide anti-bullying programs, and sex education is a part of the curriculum, while Turkey falls behind in terms of such implementations on reducing aggression among schoolchildren. For this reason, adolescents in Turkey were expected to show more involvement in aggression and sexual harassment compared to Finnish adolescents. Specifically, boys in Turkey were expected to become victimized from and to perpetrate more aggression than others.

Method Sample
The sample consisted of young adolescents (N = 1,747, 928 girls, 746 boys and 73 other/preferred not to say) from Finland and Turkey. The participants from Finland were 1,268, while from Turkey, 479 adolescents participated in the study. The age range was 9 to 16 years (M = 14.1, SD = 1.55, MFinland= 14.71, SD = 1.23, MTurkey = 12.57, SD = 1.19). The age difference between 205 Finland and Turkey was significant [t(1747) = 32.57, p< .001, d =.002]. The participants were secondary school students, including the 6 th , 7 th and 8 th grades in both countries, but also 5 th graders in Turkey; the age difference might be due to this circumstance.

Measures
The instrument used for measuring various types of aggression was Mini-DIA-R (Österman & Björkqvist, 2015) but some items about sexual harassment were added. Six items measured victimization from aggression and sexual harassment, and six measured the perpetration of the same. The following forms of aggression were measured: physical aggression, verbal aggression, indirect aggression, cyber aggression, verbal sexual harassment, and physical sexual harassment. Examples of each aggression form were provided in the instruction to the pupils in order to avoid misperceptions. Translations of the questionnaire into Turkish and Finnish were made by translators who were fluent in the source language, and native speakers of the target language (Beaton et al., 2000). In addition, the questionnaire was backtranslated into the source language in order to eliminate possible inconsistencies or conceptual errors. The choices of responses were on a fivepoint scale ranging from 0 = never to 4 = very often. The different types of aggression can be analyzed separately or added together as a total score of victimization or perpetration of aggression. In the present report, they were analyzed separately. If added together, the reliability score for victimization from aggression and sexual harassment was α = .84, and for perpetration of aggression and sexual harassment it was α = .82. Since the sample was large and the percentage of missing data was low (<5%), missing data were handled with list-wise deletion.

Procedure
The data were collected with an online questionnaire in Finland, and with both an online and a paper-and-pencil questionnaire in Turkey. The data from Finland were collected from the Ostrobothnian region in Western Finland, which is bilingual, as both Finnish and Swedish are spoken in the area. The questionnaire was provided in both spoken languages. The data from Turkey were collected from four different regions, namely the Aegean, Central Anatolia, Marmara, and Southeast Anatolia.
With respect to statistical analyses, the data were foremost analyzed with multiple analysis of variance (MANOVA) and multiple analysis of covariance (MANCOVA). Pearson's correlations were also used. Due to doubtful homogeneity of variance of some variables, the data were in addition analyzed with a nonparametric test, Mann-Whitney's U.

Ethical considerations
The study followed the guidelines of the Finnish Advisory Board on Research Integrity (2012) for responsible conduct of research. Required permissions were obtained from school officials or municipalities, as well as consent from parents/legal guardians and the adolescents themselves. Anonymity, confidentiality and voluntary attendance were principles observed in the conduction of the study. Some school officials in Turkey who were approached for participation in the study rejected the implementation of the questionnaire due to the existing items about sexual harassment.

Results
Correlations between the dependent variables of the study are presented in Table 1. As the table shows, all of the variables (scales) correlated significantly with each other. This was the case for both variables measuring victimization and perpetration of aggression and sexual harassment.  (1) 207

The Effect of Country of Residence on Victimization from and Perpetration of Aggression
A two-way MANCOVA was performed with age as covariate, country of residence and sex as independent variables, and victimization from aggression and sexual harassment as dependent variables. The results are presented in Table 2 and Figures 1 and 2. As the table indicates, the effect of country of residence on four variables of victimization (physical aggression, cyber aggression, verbal sexual harassment, and physical sexual harassment) was significant, with adolescents from Turkey scoring higher on all four (Fig. 1). The effect of sex was also significant, on the same four variables (Fig. 2). There were interaction effects between country of residence and sex on five variables (physical, indirect, and cyber aggression; and both physical and verbal sexual harassment). The boys from Turkey stood out with their high scores. On indirect aggression, girls from Finland had higher scores than girls from Turkey.  A two-way MANOVA was performed with country of residence and sex as independent variables, and the measurements of perpetration of aggression and sexual harassment as dependent variables. Since age was not found to have a significant effect on perpetration of aggression, it was not included as a covariate in this case. The results are presented in Table 3 and Figures 3 and 4. As the table indicates, the effect of country was significant on all variables of perpetration, with adolescents from Turkey scoring higher on all of them (Fig. 3). There was a significant interaction effect on four variables of victimization with boys from Turkey scoring highest in cyber aggression, verbal sexual harassment, and physical sexual harassment, while the perpetration of indirect aggression was found to be higher among girls from Finland, and among boys from Turkey. The interaction effect was significant on all variables of 209 perpetration from aggression, with boys from Turkey scoring the highest in all types of perpetration of aggression.

Figure 1
Mean scores for victimization from six types of aggression and sexual harassment, according to country of residence (N = 1,597). For significant differences, cf.

Figure 2
Mean scores for victimization from six types of aggression and sexual harassment, according to sex (N = 1,597). For significant differences, cf. Table  2.

Figure 3
Mean scores for perpetration of six types of aggression and sexual harassment, according to country of residence (N = 1,597). For significant differences, cf. Table 3.

Figure 4
Mean scores of perpetration of six types of aggression and sexual harassment, according to sex (N = 1,597). For significant differences, cf.  Psychology,10(1) 213 Because a Levene's test for the measurement of the homogeneity of variance indicated that the variables did not fully satisfy the criteria of homogeneity between them, analyses were also conducted with a non-parametric test, Mann-Whitney's U for two independent samples, which is not sensitive to this problem.

IJEP -International Journal of Educational
According to the Mann-Whitney's U with the significance level set at p < .05, there was a significant difference between the Finnish and the Turkish adolescents on all variables of both victimization from and perpetration of aggression and sexual harassment. Note that according to the MANCOVA, there was not a significant difference between the countries on two variables, victimization from verbal aggression and indirect aggression.
When sex differences were analyzed with Mann-Whitney's U, a significant difference was observed on three of the victimization variables: physical aggression, indirect aggression, and physical sexual harassment, and five of the perpetration variables: physical aggression, verbal aggression, cyber aggression, verbal sexual harassment, and physical sexual harassment.
Whether one should trust the results from the parametric or the nonparametric tests more is a moot point. When the criteria for the homogeneity of variance are not fulfilled, and when large sample variances are associated with smaller group sizes, the F-statistic may be biased and the significance level may be underestimated (Field, 2013). On the other hand, Mann-Whitney's U does not correct for multiple analyses, which MANOVA does, and is therefore more likely to make so called type 1 errors. The present authors are of the opinion that the MANOVA and MANCOVA results in this case may perhaps present a slightly truer picture of the reality. However, it should be noted that the results of the parametric and the non-parametric tests are very much in concert with each other.

Discussion
Comparative studies on aggression and sexual harassment enable the identification of cross-national similarities and differences among youth. In the current study, the aim was to compare adolescents' aggression and sexual harassment according to country of residence and sex, and the interaction effect between these independent variables was also investigated.
Country of residence was significantly related to victimization from aggression and sexual harassment, and differences between the two countries regarding victimization were found. Adolescents in Turkish schools were found to become victimized more from most types of aggression (according to the MANCOVA) and from all types of aggression (according to the Mann-Whitney's U) than adolescents in Finnish schools. The difference might be due to the quality of relationship with peers and respectful/fair school environment, since these have been found to predict adolescent aggression (Cook et al., 2010). Peer competition is more prevalent in Turkey than in Finland (p.121, OECD, 2019), a fact which might reduce the quality of peer relations. From another point of view, the difference might also reflect varying cultural values between the two countries (Bergeron & Schneider, 2005).
Scores of victimization from sexual harassment were also higher among adolescents from Turkey. The results suggest that in Turkey, an effective nation-wide prevention program against bullying should have its focus on victimization from physical and cyber aggression, and on sexual harassment. Moreover, the results might reflect inadequate communication patterns in the families (Gruber & Fineran, 2016). Victimization from sexual harassment might even in some cases be attempted to be covered up for the sake of family reputation, which makes the situation not only more traumatic for the victims, but it might also make them reluctant to talk about similar incidents to their parents and teachers in the future. In Finland, parents are provided information about adolescents' sex education by the Ministry of Health, which is perceived positively (Lottes & Kontula, 2000). On the other hand, in Turkey, sexual education is not included in the school curriculum. However, an increasing number of people in Turkey are becoming aware of the existence of sexual harassment towards children, and several cases have been reported in media, encouraging child victims to report such incidents (Bulut, 2016). In Finland, children receive sex education already in primary school (Kontula, 2010), and the education includes also information about sexual harassment. Such implementations are not likely to occur in Turkey in a foreseeable future.
The perpetration of aggression and sexual harassment was found to be related to country of residence, and the differences were significant in all types of perpetration. To our knowledge, this is the first study that compared sexual harassment among adolescents in Finland and Turkey. The results on perpetration of aggression were in line with the study by Craig and colleagues (2009). Community indicators such as neighborhood, lower socio-economic factors (Cook et al., 2010), higher income inequality (Due et al., 2009;Elgar et al., 2009), or less family/peer support (Elgar et al., 2009) might explain the higher perpetration rates among adolescents from Turkey.
The findings on sex differences in victimization from physical aggression support previous findings (Barzilay et al., 2017;Lundh et al., 2014). Boys were found to experience more victimization from verbal and cyber aggression, which is in line with some previous studies (Aricak et al., 2008;Barzilay et al., 2017;Lundh et al., 2014). Boys also reported more victimization from verbal and physical sexual harassment, which is partly congruent with previous findings (Timmerman, 2003). Similarly, sex differences on victimization from indirect aggression were also insignificant. Scores on victimization from indirect and cyber aggression, as well as verbal and physical sexual harassment, were higher among boys from Turkey, but among girls from Finland. The mean difference between the mentioned aggression types was particularly high among boys and girls in Turkey.
Boys were overall found to perpetrate more aggression and sexual harassment than girls, as in line with previous studies (Barlett & Coyne, 2014;Craig et al., 2009;Espelage et al., 2012;Gruber & Fineran, 2016;Timmerman, 2003). There was also a significant sex difference on almost all types of perpetration. No sex difference was found on the perpetration of indirect aggression. It is known from previous research that sex differences in indirect aggression might vary cross-nationally (Björkqvist, 2018). Regardless of country of residence, boys show higher externalizing behavior, which has been found to predict perpetration of aggression (Cook et al., 2010). Further research is needed to address the association between sex and the aforementioned variables.
Further research might also investigate the effect of respectful/fair school environment and peer relations on victimization and perpetration in the two countries. Similarly, whether sex education has a mitigating effect on sexual harassment could be investigated. Moreover, future research might focus on differing cultural norms about peer aggression in the countries being studied.
Some limitations of the study should be mentioned. First, some authors recommend that translations are made by native translators of the target language that are not the authors themselves (Borsa et al., 2012). In the current study, the translation of the measurement tool to Turkish was made by the first author who is a native speaker, due to lack of resources. The authors tried to limit this possibility by adapting the current questionnaire after a synthesis and evaluation of several translations of similar measurement tools.
Second, it is possible that an underreporting of sexual harassment might have occurred, especially among the adolescents from Turkey, because sexual harassment is a taboo and a topic which is not spoken freely about in the country. The researchers tried to overcome this problem as much as possible by emphasizing the anonymity and confidentiality of the study during the data collection. Another limitation was that the Turkish data were collected from five different regions, while Finnish data all were from one region; this fact might affect the representability of the Finnish sample. On the other hand, unlike in Turkey, regional disparities in terms of school contexts are very low in Finland.
Overall, the current study compares adolescents' involvement in aggression and sexual harassment in two countries with standard methodology and measures. It draws attention to similarities and disparities in different forms of aggression and sexual harassment in Finland and Turkey. The results support to a large extent previous research on sex differences, but they also stress the interaction effect between the two independent variables (country and sex) in question. Moreover, the findings reflect cross-cultural aspects of the nature of adolescent aggression. Adolescents' involvement in aggression and sexual harassment is a serious risk factor for their mental well-being, and the issue should be addressed cross-nationally.