Participants

The sample consisted of 121 women aged 18-65 years living in the metropolitan area of Buenos Aires. The mean age was 36.59 years old (SD = 12.59), 63.6% (n = 77) were under 40 years old, and 36.4% (n = 44) were over 40 years old. A minority (9.1% of the participants) had primary education (n = 11), 54.6% (n = 66) had secondary education, and 36.36% (n = 44) had university level education. The mean B.M.I. was 24.11 (SD = 5.76). Concerning occupation, 23.14% (n = 28) were self-employed, 43.80% (n = 53) were employed, 9.91% were housewives (n = 12), 0.83% were unemployed (n = 1), 5.79% were retired (n = 7), and 16.53% (n = 20) did not work. Regarding the level of physical activity, the majority (58.68%, n = 71) exercised less than 3 hours per week, followed by 31.41% (n = 38) who practiced 3 to 6 hours per week, 8.26% (n = 10) between 7 and 15 hours per week, and 1.65% (n = 2) more than 16 hours of physical activity per week. Participation on the study was voluntary and anonymous. Regarding the time spent on social media, 3.3% (n = 4) used them for less than 5 minutes, 5% (n = 6) for 5 to 30 minutes, 13.2% (n = 16) for 30 minutes to 1 hour, 14.9% (n = 18) for 1 to 2 hours, 12.3% (n = 15) for 2 to 3 hours, 24.8% (n = 30) for 3 to 4 hours, and 26.5% (n = 32) for more than 4 hours.

Instruments

Social Media Use Survey. A survey was designed to gather information about the type of social media use. Different social media platforms were listed and participants had to indicate whether or not they used them and how often. The general frequency of use of all social platforms and the type of activities carried out on them were also assessed. Participants were presented a list of 17 activities (see table 2) with 5 response options: never (1) to very often (5). Items were taken from activities usually associated to active and passive social media use (e. g., ^{Gerson et al., 2017}) but without referring to a specific platform (^{Berri & Góngora, 2021}; ^{Castro Solano & Lupano Perugini, 2019}).

Body Appreciation Scale -2 (BAS-2). It is a 10-item scale that assesses body appreciation through 5 response options: never (1) to always (5) (^{Tylka & Wood-Barcalow, 2015a}). Validation studies of the scale in Argentina (^{Góngora et al., 2020}) have demonstrated an adequate unidimensional factor structure, good convergent validity with measures of body dissatisfaction (r = -.72), pressure (r = -.49) and internalization of thin-ideals (r = .32 to r = -.61), as well as excellent levels of internal consistency (α = .96). In this sample, internal consistency was α = .94.

EDI-2 Eating Disorders Inventory. This 91-item instrument (^{Garner, 1991}) assesses characteristics associated with eating disorders. It has three disorder-specific subscales (Drive for Thinness, Bulimia, and Body Dissatisfaction) and the remaining eight scales correspond to associated traits. For this study, only the Body Dissatisfaction scale was used. The instrument has been validated in Argentina. The scale has shown evidence of validity to discriminate between clinical and general groups, men and women, and between different age groups. In the validation studies, the internal consistency of the specific scales were α > .80 (^{Casullo et al., 1996}; ^{Casullo et al., 2000}). In the sample of this study the internal consistency was α = .84.

Sociocultural Attitudes Towards Appearance Questionnaire (SATAQ-3). This instrument consists of 30 items assessing four dimensions of sociocultural influences on body image (^{Thompson et al., 2004}): Information, Pressure, General Internalization, and Athletic Internalization of thin-ideals. The items are answered on a 5-option Likert scale: from strongly disagree (1) to strongly agree (5). Validation studies conducted in Argentina have found adequate indicators of construct validity in women with and without an eating disorder, concurrent validity with the EDI-3 scale (r = .20 to r = .50), and internal consistency (α = .82 to α = .92) (^{Murawski et al., 2015}). For this study, the subscales Pressure, Internalization-General, and Internalization-Athletic were used. Internal consistency in this sample was α = .85 for Pressure, α = .79 for Internalization-General, and α = .73 for Internalization-Athletic.

Procedure

A cross-sectional design was used in this study. Instruments were completed in an online format through a link to Google Forms. Psychology students from the University of Palermo participated in the dissemination of the research through social media and contact with potential participants as part of their research practice. Participants were informed of the purposes of the research, the confidentiality of the data, and their right to refuse to participate and withdraw from the research at any time. They signed an informed consent form before completing the questionnaires. This research complies with the ethical standards of the American Psychological Association (2010) and the national standards for scientific research and from the University of Palermo. Given the characteristics of the research, voluntary and anonymous participation of consenting adults in non-intervention studies, it was tacitly approved by an internal ethics committee at the Department of Psychology, University of Palermo.

Data analysis

Data analysis was carried out using SPSS.24. The descriptive data, skewness and kurtosis of the measures used were analyzed. The values of skewness ranged between -.60 and .50, while the values of kurtosis ranged between -.63 and -.18. There is a fair consensus that values between -2 and +2 indicate a normal distribution of the variable (^{Field, 2017}), therefore, normality of the data was inferred.

A series of t-student tests were first performed to compare the means of the scales used in the study between participants over 40 years old and those under 40 years old. Cohen’s d was calculated as a measure of effect size of differences. Following the guidelines established by the author, an effect size d = .20 was considered as small, d = .50 as medium, and d = .80 as large (^{Cohen, 1992}). Pearson correlations were calculated between types of social media activities and scores on the BAS-2, SATAQ-3, and the EDI-2 body dissatisfaction scale. Subsequently, the Macro Process (^{Hayes, 2017}) in SPSS-24 was used to perform a mediation analysis (model 4), a moderation analysis (model 1), and a moderate mediation analysis (model 58) with the study variables. A mediation analysis assumes that the relationship between a predictor variable, in this case type of social media activity, and the dependent variable, body dissatisfaction, is reduced by including another variable as a predictor (body appreciation, internalization of thin-ideals, and the pressure perceived from those ideals). Mediation would indicate that the relationship between these two variables is “explained” by a third variable (the mediator). Mediation is tested by assessing the indirect effect size and its confidence interval. If the confidence interval contains zero, it is assumed that there is no genuine mediation effect. If the confidence interval does not contain zero, it is concluded that mediation has occurred. In this study a 95% confidence interval was used and conducted the analysis with a bootstrap of 5000 samples to reduce bias (^{Field, 2017}). A moderation analysis occurs when the relationship between two variables changes as a function of a third variable. In this case, the relationship between social media activity and body dissatisfaction could change as a function of body appreciation values. Moderation occurs if there is a significant interaction between the predictor and the moderator (^{Hayes, 2017}). Moderate mediation assesses whether there is a significant moderator in the path between the predictor (social media activity) and the mediator (body appreciation) or in the path between the mediator and the outcome variable (body dissatisfaction). In this case the moderator assessed was age. Moderate mediation is considered to be present if in either of these two paths there is an interaction between the predictor and the moderator (path a), or between the mediator and the moderator (path b) (^{Hayes, 2017}).

Correlations between social media activities, body dissatisfaction, thin-ideal internalization and body appreciation

Means and standard deviations of the BAS-2, EDI-2 Body Dissatisfaction and SATAQ-3 subscales are presented in Table 1.

Given the wide age range of participants, scale means were compared between women over 40 years old and women below 40. It was found that there were no significant differences in Body Dissatisfaction (t = -.76, p = .44, d = -.14), Body Appreciation (t = -1.9 p = .07, d = -.36), Internalization- General (t = 1.8, p = .07, d = .34), and Internalization-Athletic (t = 1.6, p = .1, d = .30), and Pressure (t = 1.8, p = .07, d = .34) between the two age groups. There was also no difference in time (t = .58, p = .58, d = .11) spent on social media between the two groups. Given the sample size of the subgroups and the lack of significant differences in the measures between them, the whole group was analyzed together.

Pearson correlations were performed between the frequency of social media activities and the EDI-2 Body Dissatisfaction scale, the BAS-2 scale, and the SATAQ-3 subscales. The results are presented in Table 1. A mild to moderate positive association was found between the frequency of social media activities and body dissatisfaction, as well as a higher internalization of thin-ideals.

In relation to thin-ideal internalization, the most significant associations were linked to posting transient status updates, checking to see what my contacts are doing and tagging videos. Body dissatisfaction was significantly correlated with posting transient status updates and commenting (on statuses, posts, photos). Higher body appreciation was associated with lower frequency of social media activities, in particular, posting transient status updates and posting videos had the highest inverse correlations with this variable.

In summary, among the activities assessed, posting transitory status updates and checking to see what my contacts are doing were found to be the activities most associated with negative aspects of body image.

Mediation and moderation analysis

For a deeper understanding of the relationship between significant social media activities, body dissatisfaction, thin-ideal internalization, and body appreciation, a series of mediation and moderation analyses were conducted. Macro Process in SPSS-24 was used for this purpose. Transitory status updates was selected as the social media activity, as it was the activity previously found with the most significant associations with the study variables.

First, the relationship between transient state updates, body dissatisfaction, internalization of thin-ideals, and body appreciation was examined through Process model 4. This model allows assessing the mediating role of different variables in parallel. In this case, the mediating role of body appreciation and internalization of thin-ideals was assessed between social media activity and body dissatisfaction. The model is presented in Figure 1.

Figure 1. Mediation model of body appreciation, internalization and pressure of thin-ideals in the relationship between transitory state updates and body dissatisfaction 

Results indicated that transient state updates had no direct effect on body dissatisfaction (b = 0.07, p = .85), however, there was a significant indirect effect through the mediation of body appreciation (b = 1.01, 95% CI [ .18, 1.88]). Internalization of general thin-ideals (b = -0.004, 95% CI [ -.18, 19]), internalization of athletic-ideals (b = 0.0001, 95% CI [ -.18, .18]), and appearance related pressure (b = -0.009, 95% CI [ -.20, .16]) did not mediate the relationship between social media activity and body dissatisfaction. Body appreciation showed a protective effect against social media activity.

Given previous evidence regarding the moderating role of body appreciation between social activities and body dissatisfaction, this role was also assessed through the model 1 of Macro Process in SPSS-24. No interaction between social media activity and body appreciation was found in the studied sample in relation to body dissatisfaction (b = 0.01, 95% CI [-.07, .08], t = .13, p = .89). In this study, body appreciation plays a mediating, but not a moderating role with respect to body dissatisfaction.

Second, given the wide age range of the participants, the moderating effect of age was assessed in relation to the path between the frequency of transient state updates and body appreciation (b = 0.09, 95% CI [-.02, .21], t = 1.52, p = .21) and the path between body appreciation and body dissatisfaction (b = 0.001, 95% CI [-.006, .009], t = -.36, p = .71). Macro Process model 58 was used which tests for moderate mediation in which the moderator, in this case age, would have an effect on the path a and b of the mediation but not on path c with a direct effect on the dependent variable. No moderating effect of age was found in either of the two tested paths.