*First Phase*

The initial analysis of the data showed that women ( *N* = 168) had a range of age between 21 and 22, 96.7% of them being Caucasian ethnicity. Moreover, the body image that they had was in range between 3 and 4, which may imply a normal weight. The perception that they had of themselves was fatter (3.56 ± 1.2) when compared to the desired body image (2.99 ± 0.83) (Table 1). Additionally, the most common description of body satisfaction showed low and medium-high levels of body satisfaction (48.7%). In this sense, the di fference among the group with lower and higher levels of body satisfaction was related to the body image given by the women ( χ2 = 113.64, *p* < 0.001).


**Table 1.** Mean, standard deviation and confidence intervals.

Moreover, the results from the data showed that almost 93% of the women desired to change at least three zones of their body using at least two di fferent methods (1.98 ± 0.82). The methods most used were physical activity (92%), diet (48%), surgery (24%) and beauty or alimentary products (23%). Among the zones to be modified by a surgical procedure 68% of the women indicated breast implants.

The analysis of the results from the EAT-26 test showed that most of the women had a medium probability of having disordered eating attitudes (18.34 ± 10.7). Figure 2 reflects the frequency of the scores from the EAT-26 related to body satisfaction.

The figure displays a higher frequency of scores over 20 points in disordered eating behaviours in the lower points of the body satisfaction scale. This figure implies that there were more values over 20 points when women su ffered higher levels of body dissatisfaction. Additionally, the analysis between the score in the disordered eating behaviour test and level of body satisfaction showed significant di fferences among individuals with low and high levels of body satisfaction and scores over 20 points in the EAT-26 ( χ2 = 375.34, *p* < 0.001). Moreover, a more in-depth analysis of the data, based on women with more than 20 points in the EAT-26, 48 out of 168 women showed that 40.81% had food oral control, 38.77% presented bulimia and food preoccupation and 20.5% dieting.

Further study of the data was carried out in order to address the possible correlations between the body image that women perceived of themselves and the other variables analysed. In Table 2, the correlations between the body image and the di fferent variables have shown significant value with numerous factors, including disordered eating attitudes, self-esteem, desired body image or number of methods. These correlations were positive for a fatter body image in higher scores in the EAT-26 and more methods used to modify the body image and the current body image. Moreover, negative correlations were found for a curvier description that the women gave about their body and higher desires for a thinner body image, higher body dissatisfaction and lower levels of self-esteem.

**Figure 2.** Results from the EAT-26 related to body satisfaction.


**Table 2.** Correlations with body image that women perceived of themselves.

Another variable that determines a "fatter" body image is the level of prenatal testosterone, measured by the 2D:4D ratio. This result displayed a positive relationship implying that a higher 2D:4D ratio, lower levels of intrauterine testosterone, may lead to a fatter body image.

On the other hand, Table 3 exposed the analysis of correlations between the score obtained in EAT-26 for disordered eating attitudes and the other factors analysed. This test displayed a negative correlation between having a higher score in the test and having lower levels of body satisfaction, self-esteem, the desired of having a thinner body image and worse perception of their own body.

Moreover, the positive correlations were obtained for numerous factors studied. The most highlighting positive correlations were reflected for a higher score in the SNS addiction test, a fatter body image and a higher difference in the 2D:4D ratio. These results implied that a higher 2D:4D ratio or fatter body image may lead to a higher score in the EAT-26.

The logistic regression model was used to define a disordered eating behaviour related to having lower levels of body satisfaction, the desired to achieve a thinner body image, lower levels of self-esteem, higher score in the SNS addiction test, higher duration of connection to this media and higher difference between the second and fourth finger (Table 4).


**Table 3.** Correlations with having higher scores in the disordered eating attitudes test.

**Table 4.** Logistic regression for disordered eating attitudes.


From the analysis based on levels of self-esteem and social networks, the results showed that most women have high levels of self-esteem (31.1 ± 4.7) and low levels of addictive behaviour to social network sites (14.69 ± 10.37). Furthermore, the results of the social network sites presented a high dispersion of the results. In this sense, the confidence intervals (95%) were focused on medium levels regarding addictive behaviour to SNS (13.11–16.26).

Based on this, the correlations for the score in the SNS addition test were studied. The results indicated positive significance for the number of methods used to change their body image (<0.001), higher desired of a thinner body (*p* < 0.001), lower levels of self-esteem (*p* < 0.001), greater number of social media accounts (*p* < 0.001), longer duration of the connections (*p* < 0.001) and the importance given to the social networks (*p* < 0.001). Nevertheless, the difference between the second and fourth phalange (2D:4D ratio) showed no significance with scores in the social network addiction test.

Finally, based on the results from the logistic regression, a probabilistic model was obtained. This model could diagnose 42.9% of the population with disordered eating attitudes (R<sup>2</sup> Cox and Snell 0.429) by knowing if the person had scored high in the SNS addiction test, body image, body dissatisfaction and high desire of having a thinner body. The specificity (90.3), sensibility (68.9) and valid index (84.6) results were optimal. Finally, the curve of the model was analysed (Figure 3) obtaining an acceptable probabilistic high risk of a disordered eating attitudes (area = 0.94, *p* < 0.001, CI 0.88–0.97).

**Figure 3.** ROC curve from the logistic model for disordered eating.
