Polska adaptacja i walidacja skali the physical appearance comparison scale (PACS) - Analiza wśród młodzieży w okresie późnej adolescencji w kontekście otyłości

The data were derived from the study of problem eating behaviours (acronym SZAMKA) performed during the school year 2014/2015 by the Institute of Mother and Child

Internal project performed in the Institute of Mother and Child in 2014-2015 entitled: “Problem eating behaviours (PEB) and their correlation with changes in the body mass index (BMI) in the period from late adolescence to early adulthood. Prospective study”. ²Opinion of the Bioethical Committee of the Institute of Mother and Child in Warsaw, No. 29/2014 of 17.11.2014.

The study was performed by the survey using the internet-assisted method (CAWI – Computer Aided Web Interviewing). The young people completed the questionnaire in the school computer laboratory or at home after obtaining a link to the temporary study website, www.szamka.org All the participants signed the document of informed consent. The study procedure, the applied research tool and the consent form were approved by the Bioethical Committee of the Institute of Mother and Child².

An original tool for evaluating one’s appearance in comparison with others composed of 5 questions created by J.K. Thompson et al. was tested [14]. The tool was subjected to translation and back-translation: the Polish language version, reviewed by a team of specialists, was then translated back into English and compared with the original tool. after introducing the necessary corrections in the Polish version, the scale was tested in a pilot study on a sample of 128 second grade upper secondary school students. No objections to the wording of the individual items were raised in the pilot study, which led to using the same version in the main SZAMKA survey. Table I shows the original language version of the scale and its final translation into Polish.

The original scale was composed of five statements with a choice of five categories of answers scoring 0 to 4 points. The summary index of the 5-item scale ranges from 0 to 20. The Polish version of the scale recommended in the study consisted of 3 statements and its summary index ranged from 0 to 12 points. A higher score indicates a greater tendency to compare one’s appearance with others in social situations.

For the purpose of evaluating body image the Body Image Subscale (BIS) was used, which is an element of a tool used for examining the experience with one’s body (Body Investment Scale) developed by I. Orbach and M. Mikulincer [23]. The BIS scale is composed of six items with five responses, from definitely disagree to definitely agree. Each answer scored 0 – 4 points, which gives a summary score from 0 to 24 points. A higher score indicates a more positive approach to one’s body.

The young people were asked to answer the question: Do you think your body is….: much too thin, a a bit too thin, about the right size, a bit too fat, much too fat. Their perceived body size was examined in this way, so as to identify those participants who are dissatisfied with their body weight. In the form presented, the question has been included in HBSC protocols since 2001/02 and was tested numerous times in population studies [24]. During the analyses the question was treated as a quasi-continuous scale.

The Rosenberg Self−Esteem Scale (RSES) developed for investigating global self-esteem and self-acceptance among youth was used to evaluate the self-esteem of young people [25]. In our study we used the Polish adaptation developed by K. Doroszewicz. It was employed in student studies entitled The self-knowledge scale and obtained good psychometric features [26]. The scale consists of 10 statements about the respondent with four categories of answers, scoring 0 to 3 points. The summary index of the scale ranges between 0 and 30. A higher score indicates a higher level of self-esteem.

The Sense of Coherence Scale for Adolescents (SOC-11M) was used, the version being adapted for Polish studies by M. Zwoliński et al. [27]. Young people verified 11 statements which enabled a determination of their sense of coherence, resourcefulness and rationality in everyday life (e.g. interest in what goes on around them, feeling that one is unfairly treated, understanding one’s thoughts and feelings etc.). The coherence indicator was calculated by adding up the scores for each question. The scale ranged from 0 to 33 points, a high score indicated a higher level of coherence.

The Social acceptance scale, one of the modules included in the seven-item scale proposed by S. Harter was used [28]. The scale was adapted in the HBSC 2006 study. It enabled a description of the self-perception profile by adolescents. The module used in our study comprised 5 statements which concerned functioning in a peer environment. The answers scored from 0 to 3. The scale summary index ranged from 0 to 12 points, a higher score indicated a higher level of self-esteem in social relations.

The Three-Factor Eating Questionnaire (TFEQ-13) questionnaire as adapted by the HBSC 2010 studies was used to evaluate problematic eating behaviours [29]. The questions of the TFEQ-13 scale consist of three dimensions:

The TFEQ-13 questionnaire contains answers from definitely yes (3 points) to definitely no (0 points). The last question of the questionnaire described on an eight-degree scale (1-8) referred to the assessment of the ability to restrict eating and was re-encoded in a 4-point scale (0-3). The values were calculated separately for each sub-scale. A high overall score in the partial scale indicated serious problems in that area.

The young people answered the question: Do you think that your physical development and sexual maturation occurs(ed) earlier or later in comparison with most girls (boys) of your age? There were five categories of answers: much earlier, a bit earlier, similarly as in most girls (boys) of my age, a bit later, much later. The question was re-encoded into 3 categories by combining extreme answers: early starters, normal, late starters.

On the basis of the data obtained from young people their body mass index was calculated (BMI = body mass in kg/body height in cm²). The classification into BMI categories was conducted on the basis of the International Obesity Task Force (IOTF) criteria [30]. Four categories of body mass were distinguished: underweight, normal body weight, overweight and obesity. The BMI variable was analyzed as continuous and also divided into the above four categories. The mean BMI was 21.5 (SD=3,27).

In the first step of the analyses the 5-item version of the scale was subject to exploratory factor analysis using the principle component analysis, without a specific number of distinguished factors. The analysis indicated that the scale consisting of five items creates a univariate structure. The values of the factor loadings of the 4 statements ranged from 0.695 to 0.869. The factor loading of statement No. 4 assumed a very low value – 0.259. Moreover, confirmatory factor analysis showed that the 5-item model of PACS was not well fitted to the data (tab. III). The 5-item model explained 56.15% of the total variability of the scale.

An analysis of the distribution of all PACS items indicated significant deviation from normality. Four items were positively skewed and one was negatively skewed. The floor effect ranged from 16.5% to 45.5% (question No. 2), and the ceiling effect was from 1.8% (question No. 2) to 21.8% (tab. II).

Due to the inadmissible value of the factor loading of one of the statements as a next step, an exploratory factor analysis of a 4-item scale was carried out, excluding question No. 4. The shortened version of the scale omitting the same question was also tested in French and American studies.

The scale thus constructed formed a univariate structure and the factor loadings of successive questions assumed values in accordance with the recommended statistical criteria (from 0.708 to 0.879). The 4-item model explained 69.09% of the total scale variability. The reliability coefficient (α-Cronbach’s) for the shortened, 4-item scale amounted to 0.851.

As a further step a confirmatory factor analysis of the above scale was performed. Due to the indicators obtained, which revealed the 4-item model to be unfit, a decision was made to remove yet another statement from the scale (question No.2), which according to previous analyses revealed a relatively high floor effect (45.5%).

The parameters of the 3-item model show that the model was well fitted and factor loadings of all the questionnaire items were appropriate, which made it acceptable. The value of fit indices of the models tested (5, 4 and 3 items) are presented in Table 3. The diagrams display the values of standardized regression weights and squared multiple correlations (fig. 1).

Fig. 1

The Polish version of the scale (PACS-PL) used to evaluate the degree of comparison of one’s appearance with others, shortened as a result of the analyses, consists of 3 items (questions 1, 3 and 5 of the original scale) and has a univariate structure. The value of the KMO index is 0.736, and Bartlett’s sphericity test is statistically significant (p<0.001). All the items in the scale are significantly correlated. The size of the total variation explained was 79.1% and reliability was at α-Cronbach’s=0.868 level.

The summary, 3-item PACS-PL scale ranges from 0 to 12 points. A high score indicates a strong tendency to compare one’s appearance with others. The scale is not normally distributed, showing a relatively small positive asymmetry (SKE=0.411, SE_SKE=0.068) and negative kurtosis in comparison to the normal distribution (K=-0.551, SE_K=0.136).

The average PACS-PL index was 4.29 (SD=3.18). An analysis by the U Mann-Whitney rank test indicated that it is significantly higher in the group of girls (M=5.23, SD=3.15) than in boys (M=3.27, SD=2,88), U= 131.87; p<0.001.

The mean PACS-PL index was higher among girls who underwent puberty later or earlier than in the group of their friends who matured at approximately the same time as their peers (p<0.001). The pubertal timing in boys did not reveal a relationship with PACS-PL (tab. IV).

The correlation between PACS-PL and psycho-social and behavioural factors and the BMI was investigated. On the basis of the analysis conducted on a full sample of young people, Table V presents the values of correlation ratios between variables including significance levels (p). PACS-PL showed significant correlations with all the indicators. In line with expectations a negative correlation was recorded between PACS-PL and the body image (BIS), sense of coherence (SOC-11M), self-esteem (RSES) and self-esteem in social relations (SPPA-SSE). PACS-PL also had a negative correlation with the BMI, although the value of the r-Pearson’s ratio was very low (r=-0.13). The analyses conducted separately in the group of boys and girls showed the absence of significant correlation between these variables. Self-perception of body weight had a positive correlation with PACS-PL (only among girls), as well as all dimensions of the TFEQ-13 questionnaire (for both genders). Stronger correlations were observed in the girls’ group. The value of r-Pearson’s ratios were higher in the group of girls than boys with the exception of the relationship between PACS-PL and self-esteem (scale 4), where the values of the ratios were similar for both genders.

Univariate linear regression models specific for the BMI categories (underweight, norm, overweight, obesity) were estimated in order to determine the impact of appearance in social comparisons (PACS-PL) on body image (BIS) and self-esteem (RSSE) (tab. VI). A higher score on the PASC-PL scale was associated with significant deterioration of body image in all the groups distinguished. However, in the group of obese teenagers, comparison of appearance explained the highest variability of the BIS index (R²=21%). PASC-PL also proved to be a negative predictor of self-esteem in groups of adolescents with normal body weight, overweight and obese persons. The highest percentage of RSES variation explained by PACS-PL was also recorded in the obese group (R²=23%).

In the course of the analyses two questions were removed from the original version of the tool (number four (PACS_4) and subsequently number two (PACS_2)). The shortened 4-item version of the PACS scale was also used in French and American studies. The reason for omitting question 4 was its low correlation with other components of the scale (squared multiple correlation <0.05) [31] or low internal correlation factor (<0,2) [22]. Both the foreign versions of the 4-item scale were characterized by good reliability, as was the model tested in this study. The main factor explained the 56% variance in the Polish sample, which is similar to the results obtained in the abovementioned studies by French and American authors. In the Polish version of the 4-item scale attention was paid to question No. 2 (PACS_2), which showed unacceptable skewness of distribution. Similarly, the results of the confirmation factor analysis unequivocally indicated that the 4-item model was unsuitable and the best one is a 3-item model, obtained after omitting question number 2 (PACS_2). In the future however, the 4-item model is also worth taking into account and a comparison of the parameters of both models in other populations ought to be conducted.

It was found that girls show a much higher tendency to make social comparisons of their appearance than boys. According to S. Shon, among other factors the differences associated with gender may result from a different approach to the body by boys and girls [32]. Girls are more often dissatisfied with their body and more likely than boys to experience body image concerns, which is expressed e.g. by an inadequate assessment of their weight [33]. The appearance of girls and young women is more often the object of interest for the media and peer conversations than the appearance of boys [33]. Girls are more likely to experience negative comments associated with their appearance than boys, and are more often under pressure to look according to the expected standards [34]. Another important factor is the type of social comparison dependent on gender and the motives which accompany it. Girls mainly compare their appearance with others [35], and their purpose is to assess themselves. [32]. Boys usually make social comparisons in order to evaluate their abilities against the background of other persons (e.g. in sport), and the result of the comparison is inspiring for them and motivates them to improve their skills or is a source of self-enhancement [35]. Greater concentration on their body among girls is expressed by an increased tendency to seek feedback about their appearance and attractiveness by comparison with other persons. Another result of this study which confirms this thesis is the connection between appearance social comparisons (PACS-PL) and the level of puberty observed among girls. Puberty that is too early or delayed (i.e. being temporarily “different” from others) tends to increase the tendency to the self-evaluation of one’s appearance on the basis of observations of other people in one’s surroundings.

It was demonstrated that there is a connection of PACS-PL with psycho-social and behavioural factors, which on the one hand confirms the validity of the tool adapted and on the other sets the directions for future analyses. Intensified social comparisons of appearance among both genders are accompanied by a more negative approach to one’s own body, lower self-esteem, lower perception of social acceptability, lower sense of coherence, more frequent diet restrictions, increased tendency to emotional eating and loss of control over eating. In addition an increased tendency to make social comparisons is associated with a more negative assessment of own body weight among girls.

The analyses did not confirm a direct relationship between the tendency to compare one’s appearance and body weight measured by the BMI. However, on the basis of the results obtained, it is to be concluded that comparing oneself with others may indirectly affect the obesity risk. PACS-PL significantly correlated with indicators associated with increases in BMI (applying diet restrictions, self-assessment of body weight). A similarly perceived level of puberty which affects PACS is also a recognized obesity risk factor [36].

In addition, after stratification into body weight categories, it was demonstrated that the body mass index is a moderator of the association between the tendency to compare one’s appearance in social situations and satisfaction with one’s body and self-esteem. Among obese adolescents the impact of PACS on the above dependent variables was stronger than among their peers who are overweight, normal or underweight.

The absence of a direct connection between the PACS scale results and the body mass index was also described in Australian studies concerning adolescents aged 12-16 years. By applying structural modeling Webb et al. showed that BMI and social comparisons are independent predictors of dissatisfaction with one’s appearance but do not correlate with each other. In addition, it was demonstrated that a social comparison of appearance is a mediator of the correlations between peer pressure to be attractive and dissatisfaction with one’s appearance. Conversely, dissatisfaction with one’s appearance directly affected the sense of social rejection, which was the main object of the study [37]. As such, social comparison of appearance may be treated as a factor which affects health and social functioning, for instance a sense of rejection, which, in the light of other studies, frequently applied to persons with above-normal body weight [38].