Socio-cultural power of social media on orthorexia nervosa: An empirical investigation on the mediating role of thin-ideal and muscular internalization, appearance comparison, and body dissatisfaction

As social media continues to penetrate users ’ daily lives, concerns about its negative impact on body image issues and eating disorders are becoming focuses of research. It remains unclear whether social media should be held responsible for promoting orthorexia nervosa — a problematic and extreme fixation on healthy eating. The pre- sent study draws on socio-cultural theory and tests a social media-based model in the context of orthorexia nervosa to enhance knowledge of social media ’ s influence on body image issues and orthorectic eating ten- dencies. The socio-cultural model was tested using structural equation modeling based on responses from a German-speaking sample (n = 647). The results indicate that social media users ’ involvement with health and fitness accounts is associated with higher orthorectic eating tendencies. Thin-ideal and muscular internalizations mediated this relationship. Interestingly, body dissatisfaction and appearance comparison were not mediators, which can be attributed to the characteristics of orthorexia nervosa. Higher involvement with health and fitness accounts on social media was also related to higher appearance comparisons. The results highlight the powerful influence of social media on orthorexia nervosa and the relevance of socio-cultural models in investigating the mechanisms underlying this influence.


Introduction
Social media is gaining importance, especially concerning body image issues and eating disorders. Social media "are available for viewing, content-creating, and editing 24/7, on mobile devices, anywhere, anytime, allowing for exponentially more opportunities for social comparison and dysfunctional surveillance of pictures of disliked body parts than were ever available with the conventional mass media" (Perloff, 2014, p. 366). In addition, the lockdown due to the coronavirus disease pandemic has led to increased media pressure and higher internalizations of thin and low body fat ideals, a trend that might result in more disordered eating (Baceviciene & Jankauskiene, 2021).
Social media's impact on orthorexia is, up to date, under-researched (McComb & Mills, 2019). Derived from the term "ortho," it means straight, correct, and true. A fixation on eating proper food characterizes orthorexia nervosa (ON), and people suffering from orthorexia fixate on food quality (Bratman, 1997). Although healthy nutrition can benefit individuals, an overly strong fixation on a diet can eventually lead to physical, psychological, and social damage (Barthels, Meyer, & Pietrowsky, 2015). For example, nutritional deficiencies due to low and excessively restrictive food intake, anxiety about eating less healthy products, or spending less time with friends were mentioned as possible symptoms of ON, as most resources are devoted to food-related activities (Douma, Valente, & Syurina, 2021). Even though ON is not recognized as a disease by psychiatric boards (e.g., no inclusion of ON in Diagnostic Manuals like DSM-5) (Missbach & Barthels, 2017), more scientific research on ON is warranted (Greville-Harris, Smithson, & Karl, 2020;Vandereycken, 2011). One knowledge gap related to ON is the exploration of its potential risk factors (Bundros, Clifford, Silliman, & Neyman Morris, 2016). Future research projects should specifically apply socio-cultural models to the context of orthorexia to explore social media's influence on ON (White, Berry, Sharma, & Rodgers, 2021). Following this call, our study aims to contribute to this research gap by using a conceptualized socio-cultural model. We seek to determine whether a socio-cultural model focusing on social media can also be applied to the context of ON. Thus, our study aims to develop and empirically test a socio-cultural model for ON based on the Tripartite Model (Thompson, Heinberg, Altabe, & Tantleff-Dunn, 1999), which explicitly focuses on the socio-cultural influence of social media and whether social media directly influences ON.

Social media-based socio-cultural model for ON tendencies
The Tripartite Influence Model of Body Image and Eating Disturbance is appropriate for our investigation. It was initially postulated that three socio-cultural influences-peers, parents, and media-cause comparisons and internalizations of ideals that may increase body dissatisfaction, which in turn contributes to eating dysfunction (e.g., restriction and bulimia) . The socio-cultural influence of media, in particular, on body dissatisfaction and eating disorders (Lovering, Rodgers, Edwards George, & Franko, 2018;Stefanile, Nerini, Matera, Schaefer, & Thompson, 2019) is mediated by internalization and comparison (DeCarvalho, dos Santos Alvarenga, & Ferreira, 2017;Girard, Chabrol, & Rodgers, 2018;Rodgers, Chabrol, & Paxton, 2011) and has been largely confirmed. The socio-cultural theory has also been often applied to study body image concerns and eating disorders emerging from and in relation to the Internet (Rodgers & Melioli, 2016). Scholars have highlighted that social forces (e.g., media) transfer ideals of attractiveness, which may further contribute to body dissatisfaction . Recent studies (Jarman, Marques, McLean, Slater, & Paxton, 2021;Roberts, Maheux, Hunt, Ladd, & Choukas-Bradley, 2022) indicate the importance of focusing on social media's influence on body image issues when investigating Tripartite Models.
In our social media-based socio-cultural model for ON tendencies, we investigated how social media users' involvement with health and fitness accounts relates to ON and how thin-ideal and muscular internalizations, appearance comparison, and body dissatisfaction mediate this relationship. Analyzing these mediators helps explain how social media influences ON and how body image variables are related to it. Thus, our model includes multiple factors that may influence ON and determine social media use (Prieler & Choi, 2014) to explore social media's effects on users' involvement with health and fitness accounts on social media. Fig. 1 illustrates the research model. In the following section, the model relationships are discussed in more detail.

Determining social media useinvolvement with health and fitness accounts on social media
Research has begun to investigate the impact of appearance-focused social media to explore its influence on body image. However, asking participants about the frequency of consuming such content might still be less indicative of body image outcomes compared to measures that also capture, besides others, the importance that respondents link to social media engagement and the types of content they consume . This study focuses on measuring people's involvement with health and fitness accounts on social media to capture the latter two aspects of the importance people attach to social media engagement and determine the type of content they consume. Involvement has been defined as "a person's perceived relevance of the object based on inherent needs, values, and interests" (Zaichkowsky, 1985, p. 342). Health and fitness accounts may also act as objects that have become relevant in people's lives, and some people may follow these accounts because they reflect their interest in health and fitness.
Once social media users participate in a community interested in health and fitness content, their interest in personal health is likely to increase, resulting in increased engagement in health-and fitnessrelated behaviors such as controlling which food types to eat (Dessart & Duclou, 2019). Unfortunately, the potential adverse effects may also intensify. Once individuals search for content that deals with the idealization of the body, similar content will be presented to them. The filtering function of social media exposes users to similar and problematic content all over again (Griffiths et al., 2018).
Thus, determining the involvement of health and fitness accounts seems essential, as adverse effects on body image variables have also been reported. For example, followers of health and fitness accounts were found to have a higher thin-ideal internalization and a higher drive for thinness, whereas appearance-neutral accounts yielded no such effect (Cohen, Newton-John, & Slater, 2017).
Exposure to idyllic images in the fitness, fashion, and food context also negatively influences young adults' body image and food choices (Rounsefell et al., 2020). Similarly, fitspiration videos have been reported to cause body appearance dissatisfaction among recipients (Ratwatte & Mattacola, 2021). One of the first studies (Turner & Lefevre, 2017) to find a positive correlation between higher Instagram use and higher ON tendencies was based on insights from a sample of participants following accounts focused on a healthy diet. Thus, it is expected that body image variables and orthorectic eating tendencies are significantly impacted once individuals show a higher involvement with health and fitness accounts on social media.

Identifying thin-ideal and muscular internalizations, appearance comparison, and body dissatisfaction as mediators
The internalization of societal ideals promoting a particular ideal of attractiveness is an important "individual difference variable" (Thompson & Heinberg, 1999, p. 172) that helps to explain why some individuals are more affected by media-mediated messages. Thin-ideal internalization is a central risk factor for body image and eating disturbances (Thompson & Stice, 2001;Wu, Harford, Petersen, & Prichard, 2022). Some studies have identified a positive link between thin-ideal internalization and ON (Eriksson, Baigi, Marklund, & Lindgren, 2008;Tóth-Király, Gajdos, Román, Vass, & Rigó, 2019;White, Berry, & Rodgers, 2020) and between muscular internalization and ON (Tóth-Király et al., 2019;White et al., 2020). Comparison is another relevant individual difference variable that helps explain how environmental influences (family, peers, and media) affect eating and weight-related behaviors (van den Berg, Thompson, Obremski-Brandon, & Coovert, 2002). Individuals regard others as social models about eating behavior and "use others' eating as a guide for what and how much to eat" (Cruwys, Bevelander, & Hermans, 2015, p. 3). According to the Social Comparison Theory (Festinger, 1954), people desire to evaluate their opinions and abilities. If objective means are not available, individuals will compare themselves with others. In addition, in the context of ON, reports of self-identified orthorectics (Greville-Harris et al., 2020) suggest that orthorectics compare their diet with those of others displayed on social media. Comparing one's body-related appearance to that of others has also been confirmed about disordered eating (Alcaraz-Ibáñez, Sicilia, Díez-Fernández, & Paterna, 2020), and there is evidence to assume that body-related appearance is also crucial about ON (Almeida, Borba, & Santos, 2018;Barnes & Caltabiano, 2017;White et al., 2021). The first evidence suggests that once individuals show higher body dissatisfaction, they also show higher ON tendencies (Barnes & Caltabiano, 2017).
The use of social media and its positive link to body dissatisfaction have recently been confirmed longitudinally (Marques, Paxton, McLean, Jarman, & Sibley, 2022). Previous research has identified body dissatisfaction as a mediator between internalization/comparison variables and disordered eating (Lovering et al., 2018;Stefanile et al., 2019;. Similarly, social and appearance comparisons and thin-ideal and muscular internalization have been found to mediate the relationship between social media intensity and body satisfaction (Jarman et al., 2021). However, the question remains if thin-ideal and muscular internalization, appearance comparison, and body dissatisfaction act as mediators between involvement with health and fitness accounts on social media and ON.

Hypotheses and derived research model
In light of the considerations outlined above, we have developed a comprehensive social media-based socio-cultural model for ON tendencies. We hypothesize that higher involvement with health and fitness accounts on social media will be directly associated with higher ON eating tendencies (H1). This relationship is mediated by thin-ideal (H2) and muscular internalization (H3), appearance comparison (H4), and body dissatisfaction (H5). Moreover, body dissatisfaction mediates the relationship between appearance comparison (H6), thin-ideal internalization (H7), muscular internalization (H8), and ON eating tendencies (see Fig. 1 for an overview of the derived research model).

Procedure
This study conducted a quantitative online survey to test the hypotheses outlined and the research model. Surveys are often used when investigating the influence of social media on body image-related variables . The survey was created using Lime Survey (https://www.limesurvey.org/). The research model was tested using a German-speaking sample (see section 3.2). As most measures in the questionnaire were formulated initially in English, we followed an extensive collaborative and iterative procedure for translating the items based on Douglas and Craig (2007). First, two persons, in parallel, translated the items into German. In the next step, a third person (a native speaker) acted as a judge and voted for the best translations that were then included in the final version of the questionnaire. The Institutional Review Board of Research Ethics at the University of Klagenfurt granted ethical approval for the study (ER-AAU) (2021-051).
A qualitative pretest (n = 6 participants, 50% female, 50% male; M age = 25.7, SD age = 3.3) and a quantitative pretest (n = 20 participants, 50% female, 50% male; M age = 26.9, SD age = 3.2) ensured the comprehensibility of the questionnaire. For the quantitative pretest and primary survey, participants were recruited via the online panel, Clickworker (https://www.clickworker.de/clickworker). Participants received a compensation of €2.00 in return for their participation. Pretests were conducted from October 2021 to February 2022, and data were collected for the primary survey.
We also followed the suggestions and procedural methods outlined by MacKenzie and Podsakoff (2012). The objective was to counteract common method bias by showing appreciation for people's participation, assuring them of the confidentiality and anonymity of their data, explaining why their responses are essential, and emphasizing that there are no right or wrong answers since we are interested in their very personal opinions. In addition, we labeled all questions' response options, avoided double-barrelled questions, and ensured that the formulations of the questions were clear and concise. In addition, to ensure high data quality, we included an attention check question at the midpoint of the questionnaire with the following item: "There now follows a brief check of your attention, and it is vital to complete this survey with care; hence, please select the option strongly agree." Participants who failed the attention check were excluded from the data analysis.
In the briefing, participants were also informed that there were no health-related risks linked to their participation in the study and that participation was voluntary. A contact person was indicated, and at the end of the briefing, participants were informed that by clicking the "Continue button," they would agree to participate in the study. The first filter question ensured that participants were social media users. The survey ended with a debriefing.

Participants
The present study's survey was conducted with a German-speaking sample of young men and women (18-30 years). A focus on the age group of 18 to 30 years was considered especially relevant, as according to Rounsefell et al. (2020), this age group is prone to experience negative social media influences on dietary behavior and body image. Moreover, focusing on both men and women is essential as it has been emphasized that ON affects both men and women (McComb & Mills, 2019). Furthermore, weight and shape concerns predict restrained eating across genders (Hoffmann & Warschburger, 2019). In total, 788 participants participated in the primary survey. From these, 140 participants were excluded from the analysis as they 1) failed the attention check, 2) did not indicate the correct age, 3) were of a different nationality, 4) did not use social media, or 5) answered the questionnaire only partially. Thus, the total and final sample for this study and the analysis was n = 647. Of these 647 participants, 64.9% identified themselves as German and 35.1% as Austrian (see Table 1 for a detailed overview of participants' sociodemographic characteristics).

Measures
All questions were answered on 7-point Likert scales ranging from strongly disagree (1) to strongly agree (7). The one-dimensionality of the constructs was tested for all variables. All scales showed good internal consistency. Table 2 provides an overview of all the measures and items applied in this study, together with the means, standard deviations, and Cronbach's alpha values.

Involvement with health and fitness accounts on social media
Before the posed questions, we explained in the questionnaire that social media health-and fitness-related content relates to healthy food and exercise practices (Dessart & Duclou, 2019). Participants' involvement with health and fitness accounts on social media was then measured using three items based on Schaefer, Terlutter, & Diehl (2019) and adapted to the social media context.

Thin-ideal and muscular internalization
In the present study, four items of the subscale "Internalization: Thin/Low Body Fat" from the SATAQ-4R (Schaefer, Harriger, Heinberg, Soderberg, & Thompson, 2017) were used to assess thin-ideal internalization. Another four items of the subscale "Internalization: Muscular" from the SATAQ-4R (Schaefer et al., 2017) were used to measure muscular internalization.

Appearance comparison
One item was used to measure the frequency of comparisons based on Lee (2014), which was adapted to the social media context ("I think I often compare myself with others on social media"), as also used by Jarman et al. (2021). In addition, five items measured "Upward Physical Appearance Comparisons" (O'Brien et al., 2009). The wording "on social media" was added at the beginning of each item to link the comparisons to the social media context, as practiced by Jarman et al. (2021), and to capture comparisons in the social media context more fully . We focused on measuring upward appearance comparisons because most appearance comparisons made through social media were reported to be upward. Upward appearance comparisons on social media also negatively impact diets (Fardouly, Pinkus, & Vartanian, 2017), thus fitting the present study's focus on exploring appearance   (2009). Moreover, we ensured that our applied measures are reliable measures among both women and men as also reported in the respective papers (Barthels et al., 2015;Jarman et al., 2021;Schaefer et al., 2017Schaefer et al., , 2019Thiel et al., 1997). Items in italics also belong to the original measures and were assessed in the questionnaire, however, items in italics were not used for the analysis in the present study as a) the factor loading was too low (results EFA: 0.45/0.48) and loadings of 0.5 or even higher are desirable (Hair et al., 2006), b) items had high cross-loadings on two factors (Hair et al., 2006) or c) items formed own factors and had lower Alpha values compared to the retained items/factors. *marked items were reverse coded. Validated German items were available for measuring comparisons on social media related to orthorectic eating tendencies. The items were combined to form one factor, and the check for one-dimensionality was satisfactory.

Body dissatisfaction
Five items of the subscale "Body Dissatisfaction" subscale of the German version of the Eating Disorder Inventory (EDI-2) (Thiel et al., 1997) were used to assess body dissatisfaction.

Orthorectic eating tendencies
We measured people's orthorectic eating tendencies with the "Duesseldorf Orthorexia Scale" (DOS) (German version) (Barthels et al., 2015). The DOS has been recommended for future research projects to measure ON (Meule et al., 2020). Six items remained in the analysis, resulting in a shortened version of the DOS (see Table 2).

Statistical analysis
Data were analyzed using IBM SPSS Version 28 and IBM SPSS AMOS 28. First, confirmatory factor analysis (CFA) was conducted to support the measurement model. Internal consistency reliability, convergent validity, and discriminant validity were used to assess the measurement model (Hair, Matthews, et al., 2017) (see Table 3). Second, the research model and outlined hypotheses ( Fig. 1)

Evaluation of the research model and hypotheses testing
The model depicted in Fig. 1 did not yield a satisfactory fit to the data. Modification indices were applied to improve the model structure and fit. One additional path from thin-ideal internalization to appearance comparison was included in the model, which was also in line with theory (Jarman et al., 2021). This new model yielded a good fit to the data, χ 2 (330) = 999.67, p 0.000, CFI = 0.96, TLI = 0.95, RMSEA = 0.06, SRMR = 0.07. The final model is illustrated in Fig. 2.
Standardized path coefficients were examined to test the hypotheses. Involvement in health and fitness accounts on social media was positively and directly associated with ON tendencies (ɣ = 0.204, p < 0.001), supporting H1. Moreover, involvement with health and fitness accounts on social media was also positively related to thin-ideal internalization (ɣ = 0.170, p < 0.001), and thin-ideal internalization was also positively related to ON tendencies (β = 0.439, p < 0.001), confirming H2 and the mediating influence of thin-ideal internalization. The analysis further revealed a positive relationship between involvement and muscular internalization (ɣ = 0.367, p < 0.001), and muscular internalization was also positively related to ON tendencies (β = 0.188, p < 0.001), supporting H3 and muscular internalization as mediators. Interestingly, involvement was also positively associated with appearance comparison (ɣ = 0.237, p < 0.001); however, appearance comparison was not positively related to ON tendencies (β = 0.032, p = 0.525). Therefore, appearance comparison did not mediate this relationship; thus, H4 is only partially supported. Surprisingly, involvement was not positively associated with body dissatisfaction (ɣ = − 0.060, p = 0.171), and body dissatisfaction was also not positively related to ON (β = − 0.085, p = 0.080); thus, H5 was not supported.
The final three hypotheses predicted a mediating role of body dissatisfaction in appearance comparison, thin-ideal internalization, muscular internalization, and ON. However, as already outlined for H5, no significant influence of body dissatisfaction on ON tendencies was found; therefore, body dissatisfaction did not mediate in these relationships. Only two direct influences on body dissatisfaction turned body dissatisfaction (Thiel et al., 1997) and orthorectic eating tendencies (Barthels et al., 2015), which were translated for publication. correlations of construct with other constructs (below the diagonal) and heterotrait-monotrait ratio of correlations (HTMT) <0.85 (above the diagonal in italics). Our chosen approaches to assess convergent and discriminant validity are also in line with suggestions by Swami and Barron (2019) to establish construct validity. ***p < 0.001. out to be significant, lending partial support to H6 and H7. Appearance comparison was positively associated with body dissatisfaction (β = 0.196, p < 0.001), partially supporting H6. Thin-ideal internalization was positively related to body dissatisfaction (β = 0.410, p < 0.001), partially supporting H7. Muscular internalization was not significantly related to body dissatisfaction (β = − 0.062, p = 0.138); thus, H8 was rejected.
We aimed to test our model using a general population sample. This sample generally showed relatively low ON tendencies (M = 2.7, SD = 1.4). We also tested the model (Fig. 2) among participants who showed higher ON tendencies using multigroup structural equation modeling. 89 participants in our study had ON scores of ≥ 5 (M = 5.2, SD = 0.6). The model yielded a good fit to the data (χ 2 (695) = 1488.50, p 0.000, CFI = 0.95, TLI = 0.94, RMSEA = 0.04, SRMR = 0.07). The same significant paths emerged in this model, as shown in Fig. 2. The model for the participants with higher ON tendencies is shown in Fig. 3.

Discussion
The present study aimed to investigate whether social media may lead to higher ON tendencies and whether thin-ideal internalization, muscular internalization, appearance comparison, and body dissatisfaction mediate this relationship. For this purpose, we developed and empirically tested a comprehensive socio-cultural model for ON tendencies, focusing on the socio-cultural power of social media.
The developed model was largely confirmed using a general population sample and among participants with higher ON tendencies. We confirmed that users' involvement with health and fitness accounts on social media is positively related to ON tendencies. The results align with those of the first qualitative studies Greville--Harris et al., 2020;White et al., 2021) and of the first quantitative study (Turner & Lefevre, 2017), the latter that only analyzed the correlation between Instagram use and ON. We add robustness to the qualitative studies' findings, given our study's quantitative nature and relatively large sample size. As our study revealed, individuals who showed greater interest in health and fitness topics also showed higher ON tendencies than those who showed less interest in such content.
In addition, we gained essential findings on how body image variables mediate the relationship between social media and ON and the direct influence of body image variables and ON. As expected, thin-ideal and muscular internalizations mediated the relationship between involvement with health and fitness accounts on social media and ON tendencies. Individuals who indicated greater involvement with health and fitness accounts on social media also showed higher thin-ideal and muscular Note. ***p < 0.001. Dashed lines indicate non-significant paths. The path between thin-ideal internalization and appearance comparison was added during data analysis, thus no hypothesis was assigned. H = Hypothesis. internalizations, which further contributed to higher ON tendencies. This finding is in line with those of prior studies that explored the influence of media and social media on thin-ideal internalizations (Jarman et al., 2021;Lovering et al., 2018) and muscular internalizations in other contexts (Jarman et al., 2021;Schaefer, Rodgers, Thompson, & Griffiths, 2021) and with prior research that identified a positive relationship between thin-ideal and muscular internalizations (Tóth-Király et al., 2019;White et al., 2020) and ON. Our results suggest that both internalization variables are essential to understand how social media, especially content related to health and fitness, can potentially increase ON tendencies.
Interestingly, body image variable appearance comparisons did not mediate the relationship between involvement with health and fitness accounts on social media and ON tendencies. However, we found that more significant involvement with health and fitness accounts on social media significantly influenced higher appearance comparisons. In our study, which focused on upward comparisons, appearance comparisons were unrelated to the ON tendencies. Barnes and Caltabiano (2017) found a relationship between appearance orientation and ON. However, they did not explore any direction of comparison (upward or downward). Our study revealed that upward appearance comparisons are not related to ON tendencies. One possible explanation for why upward appearance comparisons were not related to ON could be the different direction of comparison. Regarding eating choices, Greville-Harris et al. (2020) reported that orthorectics engage in downward comparisons with people who lead less healthy lifestyles and make less healthy eating choices, making them feel superior. Future research could explore whether these findings can be extended to appearance comparisons and whether downward comparisons with appearance are related to ON.
However, as previously stated, higher involvement with health and fitness accounts on social media caused more appearance comparisons, which may lead to higher body dissatisfaction, confirming the mediating role of appearance comparisons for these variables. This result corroborates previous findings that have also detected a mediating role of comparison in tripartite models linked to media's influence on body dissatisfaction (DeCarvalho et al., 2017;Rodgers et al., 2011;. Individuals may feel inferior compared to others; consequently, negative self-evaluations are likely to follow (Halliwell, 2012). The results of our study suggest that once individuals compare their appearance with the appearance of individuals depicted in health-and fitness-related content on social media, their body dissatisfaction is likely to increase. Thus, health and fitness accounts on social media may also trigger dangerous comparisons, making it an important type of content to consider in future research.
Our study paints an unexpected picture of body dissatisfaction as a mediator. Besides what was expected, body dissatisfaction did not significantly mediate the relationship between involvement with health and fitness accounts on social media and ON tendencies. We found that involvement with health and fitness accounts on social media does not significantly lead to higher body dissatisfaction, and body dissatisfaction, in turn, also does not significantly lead to higher ON tendencies. These findings are contrary to prior research, which has suggested a direct positive connection between the use of social media leading to higher body dissatisfaction (Marques et al., 2022) and disordered eating (Holland & Tiggemann, 2016;Rodgers & Melioli, 2016). They also contradict prior research that found a positive relationship between body dissatisfaction and ON (Barnes & Caltabiano, 2017;Barthels, Kisser, & Pietrowsky, 2021).
One possible explanation for why involvement with health and fitness accounts on social media was not related to body dissatisfaction in our study could be people's media literacy level. Rodgers, Paxton, and Wertheim (2021) stated that research on media literacy, which might protect some individuals from the negative impact of social media images on body image, is still limited to date. Participants with high media literacy might know that the images displayed on health and fitness accounts might further dissatisfaction with their bodies. Additional research could investigate media literacy as a potential moderator in this relationship.
A possible explanation for why higher body dissatisfaction did not lead to higher ON tendencies in our study could be related to the different components of negative body image (e.g., over-evaluation of weight and shape) that influence ON to different degrees (Messer et al., 2022). Future research could replicate our study by investigating the different components of body image in a social media-based socio-cultural model for ON tendencies. Another possible explanation might be that studies that have reported a positive link between body dissatisfaction and ON focused on women (Barthels et al., 2021) or, although focusing on men and women, used a different measurement for ON (Barnes & Caltabiano, 2017) than that used in our study. Our study included men and women, and our mixed sample did not confirm the positive relationship between body dissatisfaction and ON tendencies. Future studies should further investigate the link between body dissatisfaction and ON in mixed samples to support our findings.
Our study makes significant contributions to existing research. First, previous studies have primarily focused only on body image as a dependent variable when exploring the influence of social media as an independent variable (Holland & Tiggemann, 2016). We explored the influence of social media on body image variables (body dissatisfaction, internalizations, appearance comparisons) and ON tendencies. Second, we extend the insights of the first quantitative study that found a positive link between social media use and ON based on correlation analysis (Turner & Lefevre, 2017). We move beyond simple correlation analysis and use structural equation modeling to test a comprehensive socio-cultural model focusing on social media in the ON context. Hence, our findings are consistent with the hypothesis that social media, specifically involvement with health and fitness accounts, is a risk factor for ON (McComb & Mills, 2019).

Practical implications
The results of our study highlight the importance of creating awareness among users of social media and its influence on orthorexia nervosa and related body-image issues. Awareness campaigns could be used to explicitly warn users about the potential adverse effects of health and fitness accounts. Popular social media influencers can share these campaigns with their followers, particularly in the health and food sectors. Organizations behind social media platforms could also implement warning messages as soon as users are exposed to health-and fitness-related content, especially as companies are aware of the harmful effects of algorithm-driven content on users' well-being (Harriger, Evans, Thompson, & Tylka, 2022). We also urge practitioners and public health organizations to develop interventions to create awareness of social media's influence on ON and related body-image issues.

Limitations and directions for future research
This study has limitations that offer essential opportunities for future research. The methodological approach of this study followed a quantitatively oriented survey-based design. Future research might use mixed-methods approaches to corroborate the study's results and include experiments to examine which health and fitness content, in particular, leads to higher ON tendencies (e.g., food content vs. content depicting individuals). Our model is based on a well-established and researched theoretical model (Jarman et al., 2021;Roberts et al., 2022;, and our derived relationships between the variables are theoretically sound. However, this study used cross-sectional data. Hence, longitudinal studies are required to confirm our findings. It would also be interesting to focus on different social media channels and capture the effects of different social media channels on body image issues and orthorectic eating tendencies. Furthermore, including upward and downward comparisons and analyzing the role of social media literacy may yield exciting results. We included the full version of the DOS with ten items (Barthels et al., 2015) to measure orthorectic eating tendencies; however, we could only use a shortened version of the DOS scale for our analysis. Hence, future research might use different scales to measure ON. Finally, it would be interesting to analyze sex-specific differences and extend our research to other cultures.

Conclusion
This study provides the first evidence supporting the importance of focusing on socio-cultural models in the context of orthorexia and social media (White et al., 2021). Our social media-based socio-cultural model of orthorectic eating tendencies was largely confirmed. As such, this model enhances existing knowledge on how social media influence orthorectic eating tendencies via body image variables. The results provide the first evidence for considering social media, especially involvement with health and fitness accounts, as a risk factor for ON. Thus, the findings highlight the importance of protecting users from the potential negative consequences linked to health and fitness content on social media.

Funding
This work was supported by the Faculty of Humanities of the University of Klagenfurt (financial support provided for the data collectioncompensation for study participants).

Ethical statement
The Institutional Review Board of Research Ethics at the University of Klagenfurt granted ethical approval for the study (ER-AAU) (2021-051).

Informed consent
Informed consent was obtained from all participants included in the study.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.