Adolescents often inaccurately perceive their body weight, and this is a risk factor of negative self-perception(Reference Lucibello, Sabiston and O’Loughlin1), engagement in unhealthy weight-control behaviours and disordered eating behaviours(Reference Pasch, Klein and Laska2,Reference Haynes, Kersbergen and Sutin3) .
It has been shown that inaccurate body weight perception in adolescents persists to adulthood(Reference Aloufi, Najman and Mamun4). Despite the substantial percentage of adolescents who are objectively overweight or obese (OWOB), findings from Europe indicate that many do not perceive themselves as such(Reference Yang, Turk and Allison5). Overweight and obesity rates in Israel are high among adolescents; in the 2014 Health Behaviours in School-Aged Children (HBSC) multinational survey, overweight rates of children aged 11, 13 and 15 years were 14 % among girls at all ages and 26 %, 20 % and 25 %, respectively, among boys(Reference Inchley, Currie and Young6). Overweight adolescents with an accurate perception of their weight are more likely to attempt appropriate weight control behaviours, such as decreased energetic intake and increased physical activity(Reference Yang, Turk and Allison5,Reference Lenhart, Daly and Eichen7) . On the other hand, underestimation, such as regarding oneself as normal weight or thin while being OWOB, may lead to lower participation rates in weight control programmes due to the misconception(Reference Kim and So8,Reference Park9) .
Research has focused on the development of body image and its association with significant relationships. According to the ecologic model of health, interactions with family, teachers and peers provide the social context in which adolescents shape their health perceptions, attitudes and behaviours(Reference Alikasifoglu, Erginoz and Ercan10–Reference Garcia, Gatdula and Bonilla13). According to the Youth Resiliency Model(Reference Harel-Fisch, Bonnie and Tzimer14), adolescents are primarily influenced by the presence of a significant adult in their lives, positive daily school experiences, a sense of self-worth and positive social connectedness with peers. Adolescents who report communication interactions with their parents are more likely to rate their health positively, have fewer psychological complaints and report greater life satisfaction(Reference Vokáčová, Vašíčková and Hodačová15). Additionally, studies have shown that adolescents with better parent–adolescent relationships are less likely to experience body dissatisfaction(Reference Hosseini and Padhy16).
Positive relationships with non-family adult mentors (mainly teachers or guidance counselors) have been related to increased physical health, self-esteem, life-satisfaction, graduating from high school and decreased levels of risk behaviours(Reference Walsh, Harel-Fisch and Fogel-Grinvald17). However, higher perceived support from teachers was not found to moderate the inverse association between high BMI and quality of life measures among adolescents(Reference Lebacq, Dujeu and Méroc18).
During adolescence, perceived peer support plays a protective role against anxiety, depression and behavioural distress. Peer support was found to influence adolescent eating behaviours, possibly by supporting positive body satisfaction, which in itself has a protective role against maladaptive eating behaviours (e.g. restrictive dieting, binge eating and purging)(Reference Pace, D’Urso and Zappulla19). Peer support is also considered an indicator of social connectedness(Reference Pryce, Moutela and Bunker20).
While the influence of parents, teachers and peers on numerous health behaviours (e.g. smoking and suicidal behaviours) is well established(Reference Madjar, Walsh and Harel-Fisch21–Reference Xie, Palmer and Li23), information is limited on how contextual and social support patterns shape adolescent weight perception and in what way the accuracy of weight perception is related to actual body weight.
In the present study, we leveraged a country-wide, representative sample of adolescents with the objective of investigating how support from parents, teachers and peers interact to influence the accuracy of adolescent body image.
Methods
Participants
The 2014 Israeli HBSC study surveyed 7563 students (49·6 % male, 50·4 % female) in grades 6 to 12 (ages 11 to 18 years)(Reference Madjar, Walsh and Harel-Fisch21). In brief, the HBSC is a school-based survey of adolescent health behaviours and socio-economic determinants, conducted in a representative sample of school children, using an international standardised methodological protocol(Reference Currie, Inchley and Molcho24,Reference Roberts, Freeman and Samdal25) . A clustered sample of students in grades 6 to 12 (303 schools) was selected through weighted probability methods to obtain a balanced representation of school population characteristics. School characteristics used for stratification included the language of instruction (Hebrew or Arabic) and type of school (secular public school or religious public school). Students in private schools, schools for children with special needs, schools for incarcerated youth and schools serving ultra-orthodox Jewish populations were excluded. Response rates were 303/337 (90·0 %) at the school level. The most common reasons for nonparticipation were failure to return the informed consent form, failure to receive parental consent and absence on the day of survey administration. All students in sampled classrooms present on the day of the survey were included (> 95 % pupil response). The current study focused on Hebrew speakers only.
Measures
Socio-demographic measures
Age group students were divided into two groups according to school type: middle school v. high school.
Family affluence
The Family Affluence Scale, a composite measure of six questions, was used as an indicator of socio-economic status(Reference Voráčová, Sigmund and Sigmundová26,Reference Dalmasso, Borraccino and Lazzeri27) . Answers were ranged on a scale from 0 to 13 and later categorised as low (0–6), medium (7–9) or high (10–13) family affluence(Reference Voráčová, Sigmund and Sigmundová26,Reference Dalmasso, Borraccino and Lazzeri27) .
Peer and family support
Was measured using part of the Multidimensional Perceived Social Support Scale, with a higher score indicating higher social support. The revised version of the scale, which is a twelve-item questionnaire developed by Zimet et al. (Reference Zimet, Dahlem and Zimet28), measures perceived support from family (4 items), friends (4 items) and significant others (4 items), with the sum score ranging from 4 to 28 for each subscale. In the original Israeli 2014 HBSC survey, only two scales were included: perceived support from family and friends. Family support was measured by the following questions: (1) whether they felt that their family really tries to help them; (2) whether they could get emotional support from their family when they needed it; (3) whether they can talk to their family about problems and (4) whether their family is prepared to help them make decisions. Peer support was measured by the following questions: (1) whether respondents perceived that their friends really try to help them; (2) whether they could count on their friends when things go wrong; (3) if they have friends with whom they can share their sorrows and joys and (4) whether they can talk to their friends about their problems. This instrument provides response options ranging from 0 to 7 (very strongly disagree to very strongly agree)(Reference Zimet, Dahlem and Zimet28).
Teacher support
Three items in the questionnaire measured teacher support: ‘“I feel my teacher accepts me as I am”’, ‘“I feel that my teachers care about me as a person”’ and ‘“I feel a lot of trust in my teachers”’(Reference Walsh, Harel-Fisch and Fogel-Grinvald17).
Dimension reduction
Dimension reduction was performed using the principal component analysis (PCA) to create psychosocial indices for inclusion in statistical models(Reference Kaiser29). PCA procedures were applied to all variables of peer support, family support and teacher support as indicated above. Variables were included based on the condition that correlations were greater than r = 0·35. PCA was performed using the eight peer and family support questions. The first factor accounted for 78·38 % of the total variance. Factor loadings are presented in supplemental Table S1.
Later, quartile cut points were later calculated (Q1–Q4) from the continuous factors, such that Q1 indicated strong disagreement, and Q4 indicated strong agreement.
Communication with parents
Participants was asked about how easy it is for them to talk to their fathers, stepfathers, mothers and stepmothers about things that really bother them. Two separated variables were generated: one for communication with the father and the other for communication with the mother. Five response choices were provided (0 do not have or see this person; 1 = very difficult; 2 = difficult; 3 = easy and 4 = very easy). The resulting father and mother communication variables were dichotomised and recoded as ‘very easy/easy’ and ‘difficult/very difficult’(Reference Vokáčová, Vašíčková and Hodačová15).
Anthropometric measures
Participants reported on their current height (in metres) and weight (in kilograms (kg)). BMI was calculated as weight in kilograms divided by height in metres squared (weight (kg)/height (m2)). We used the WHO growth reference for children aged 5–19 years to compare the prevalence of weight status in our sample, using AnthroPlus version V1.0.2 software.2(30). According to WHO recommendations, underweight, normal weight, overweight and obese ranges were defined as BMI-Z < -2 sd, −2 sd < BMI Z-score < 1 sd, 1 sd < BMI-Z > 2 sd and BMI-Z > 2 sd, respectively(Reference De Onis, Onyango and Borghi31). Due to low obesity prevalence in the study population (2·1 % of the girls and 3·4 % of the boys), overweight and obesity were combined into a single category. Missing and extreme values exceeding the possible limits for age and sex subgroups were excluded from the analysis (i.e. values (±3 sd) of BMI Z-scores)(32).
Perceived weight status
Perceived body size was assessed by the question: ‘Do you think your body is…’ Response options included: ‘much too thin,’ ‘a bit too thin,’ ‘about the right size,’ ‘a bit too fat’ and ‘much too fat.’ Subjects who responded ‘too thin’ or ‘a bit too thin’ were classed as perceived underweight; subjects who responded ‘about the right size’ were classed as perceived normal weight; and subjects who answered, ‘a bit too fat’ or ‘too fat’) were classed as perceived overweight(Reference Marques, Naia and Branquinho33).
Agreement between BMI category and perception of weight category
The perceived weight category was compared to the actual weight status for each participant, as calculated by BMI Z-scores. Students were categorised into one of the three resulting groups: underestimation, correct estimation or overestimation of body weight status.
Statistical methods
All statistical analyses were performed using SPSS version 25.0 (IBM). Descriptive statistics of socio-demographic characteristics of the study participants according to age groups were compared using the χ 2 test. A weighted κ coefficient was applied to assess the accuracy of perceptions of body weight: agreement between perceived weight status and actual BMI category. Additionally, the proportions of correct estimation, underestimation and overestimation according to age groups were also presented for each existing BMI category (underweight, normal and overweight/obese). All tests were considered significant at P < 0·05.
Associations between SSP to under and overestimation of body weight were examined using multivariable stepwise conditional logistic regression models, with adjustment for selected socio-economic factors. OR and 95 % CI are presented.
Results
In total, 7811 students completed the questionnaires. Of them, ninety-one students had missing socio-economic data, and 157 students reported an extreme BMI Z-score (values greater or smaller than 3 sd of BMI-Z-scores). Those observations were excluded from the analysis. The final sample included 7563 students, equally divided by sex and school type (Table 1). The majority of students were from middle and upper-middle-class families. Most of the students were from secular schools. Most of the students were categorised as normal weight.
* Other – Russia, Europe, Ethiopia, South America, English-speaking country or other.
** FAS – family affluence scale. This socio-economic status scale has been developed for the HBSC study group.
Table 2 presents body weight categories and the correctness of weight perception by age group. Overall, middle school students were classified as underweight three times as often as high-school students (9·7 % v. 3·3 %), and 16·1 % of the boys and 10·7 % of the girls were OWOB. Among high-school students, an increased rate of OWOB status was documented among 17·6 % of the boys and 11·7 % of the girls. The majority of the boys and girls perceived their body weight status correctly (59·2 % and 57·2 %, respectively), while 25·6 % of the boys and 15·1 % of the girls underestimated their weight status, and 15·2 % of the boys and 27·7 % of the girls overestimated their weight status.
OWOB, overweight or obese.
* BMI was calculated as weight (kg) divided by the square of their height in meters (m2). Underweight was defined as BMI Z-score <-2 sd. Normal weight was defined as: −2 sd < BMI Z-score < 1 sd. Overweight or obese was defined as BMI Z-score >1 sd.
** Weight perception was self-reported by the students.
Table 3 presents the agreement between BMI classification and self-perception of body weight status. The weighted Kappa coefficients indicated fair agreement between perceived weight status and actual BMI categories in both age groups and sexes.
OWOB, overweight or obese.
Underweight was defined as: BMI Z-score <-2 sd; normal weight was defined as (−2) sd < BMI Z-score < 1 sd; overweight or obese was defined as BMI Z-score >1 sd.
The PCA was conducted to capture distinct SSP. Three separate SSP were generated, accounting for a total of 81·9 % of the variance. The first social support pattern accounted for 30·7 % of the variance, the second accounted for 28·5 % of the variance and the third accounted for 22·7 % of the variance. To generate unique factors, the variables that loaded the highest in each factor were entered into a second PCA. Factor loadings of the separate SSP are presented in Supplemental 1. Each continuous pattern was then divided into quartiles, such that Q1 represented the lowest factor load and Q4, the highest. The patterns were entered into a multivariable logistic regression model.
Compared to high-school students, middle school students were significantly less likely to overestimate their body weight status (OR = 0·77; 95 % CI (0·68, 0·88); P<·001; Table 4). Girls had more than twice as likely as boys to overestimate their weight status (OR = 2·03; 95 % CI (1·76, 2·33); P<·001). Boys were more likely to underestimate their weight status than girls (OR = 0·63; 95 % CI (0·55, 0·72); P<·001; Table 5). In addition, those at the highest quartile of family support (v. Q1–Q3) were less likely to underestimate their body weight (OR = 0·83; 95 % CI (0·72, 0·96); P < 0·05; Table 5).
Students in the highest v. the lowest socio-economic status group were more likely to overestimate their weight status (OR = 1·19; 95 % CI (1·01, 1·41); P<·04). Students in the top quartile (Q4 v. Q1–Q3) of family support were less likely to overestimate or underestimate their body weight status. Students who had good communication (v. bad) with a parent were less likely to overestimate their body weight status. Students who had higher teacher support (Q4 v. Q1–Q3) were less likely to overestimate their body weight.
Discussion
This study examined the associations between support patterns on adolescents’ perceptions of their body image.
The present study, based on a representative sample of 7563 Israeli students, found that the majority of the girls and boys in both middle and high school were normal weight; however, a significant inaccurate perception of body weight, overestimation and underestimation were detected. In a multivariate logistic regression analysis, we found that overestimation of weight status was associated with the following variables: high-school attendance; female sex; high socio-economic status; lower family, peer and teacher support, and more insufficient communication with a parent. Underestimation of weight status was associated with male sex and higher family support.
In our research, high-school attendance was associated with higher odds of overestimation of body weight. Although we have not found similar research findings of differences in weight misperception, which are age-related, Inaccurate body weight among high-school students has been documented in studies(Reference Maximova, McGrath and Barnett34,Reference Opie, Glenister and Wright35) among adolescents and adults. About 10 % of US high-school students that participated in the 2009 National Youth Risk Behaviour Survey (YRBS) overestimated their body weight and an additional 27·8 % underestimated their body weight(Reference Ibrahim, El-Kamary and Bailey36). Among adolescents in the UK, high percentages of high-school students demonstrated misperceptions of body weight(Reference Viner, Haines and Taylor37).
Overestimation of weight status was more commonly reported among girls than boys(Reference Kim and So8,Reference Steinsbekk, Klöckner and Fildes38,Reference Jackson, Johnson and Croker39) . Several studies have examined rates of overestimation of body size among adolescents, and results varied across countries. In a representative survey conducted among adolescents in the UK, only 7 % of normal-weight adolescents perceived themselves to be too heavy, with higher percentages among girls(Reference Jackson, Johnson and Croker39). Among adolescents in Japan, overestimation was documented among girls only, with 23·2 % of underweight girls and 45 % of low-normal weight girls who perceived themselves as a bit too fat(Reference Nishida, Foo and Shimodera40). Overestimation of body weight was prevalent among 16·6 % of the participants in the Youth Risk Behaviour Surveillance Surveys, with higher prevalence rates among sexual minorities(Reference Mantey, Yockey and Barroso41). In our study, 15·2 % of boys and 34·2 % of girls overestimated their weight status. Underestimation was found among 25·6 % of boys and 15·1 % of the girls. Body dissatisfaction and weight concerns may begin in pre-adolescence and increase after puberty, particularly among females. Some studies have found that weight issues, dietary restraint and attempts to lose weight are more prevalent among females of higher socio-economic status(Reference O’Dea42).
In consistency with our results, it has been reported that boys tend to underestimate their weight, while girls tend to overestimate their weight(Reference Kim and So8,Reference Yan, Zhang and Wang43,Reference Page, Johnson and Simonek44) .
Body weight perception is complex and influenced by cultural and societal factors, as well as individual characters. Higher satisfaction from life has been reported among adolescents who reported higher family support(Reference Schnettler, Miranda-Zapata and Lobos45). Parents can play an essential role in reinforcing positive stimuli and filtering out negative influences on their children(Reference Al Sabbah, Vereecken and Elgar46). Adolescents who report easy communication with their parents are more likely to rate their health positively, mention fewer psychological complaints and report greater life satisfaction(Reference Vokáčová, Vašíčková and Hodačová15).
Positive relationships with non-family adult mentors (mainly teachers or guidance counselors) are related to increased physical health, self-esteem, life-satisfaction and high-school completion, and decreased levels of risk behaviours(Reference Walsh, Harel-Fisch and Fogel-Grinvald17). However, the role of the school in the context of adolescent weight perception remains unknown. The present study demonstrates that teacher support confers protection from the overestimation of body weight.
In our models, peer support was not a significant predictor of accuracy of body weight. This may be caused by the fact that perceptions of body weight are developed at early life stages(Reference Hosseini and Padhy16). However, during adolescence, peer support plays a protective role against anxiety, depression and behavioural distress. Peer support may influence adolescent eating behaviours, suggesting that perhaps by supporting body satisfaction. This may play a protective role against maladaptive eating behaviours such as restrictive dieting, binge eating and purging(Reference Pace, D’Urso and Zappulla19). Having a supportive group of friends is an essential part of healthy adolescent development and for overweight adolescents, it could be a protective factor against body dissatisfaction(Reference Dunkley, Wertheim and Paxton47). Body weight dissatisfaction and fear of being overweight in early adolescence are important risk factors for disordered eating. Overestimation of body weight may increase weight loss attempts, perhaps leading to disordered eating(Reference Dunkley, Wertheim and Paxton47–Reference Jankauskiene and Baceviciene51). Little or insufficient support from family and friends may cause a negative self-image. Conversely, a sense of support from the immediate social environment may serve as a protective factor from the social pressures that are hypothesised to foster body dissatisfaction(Reference Chen, Gao and Jackson52).
Our study had several limitations. We present a cross-sectional analysis, precluding the determination of causality. BMI status and classification of weight were based on self-reported height and weight. Adolescents have been found to overreport their height and underreport their weight. Therefore, it is possible that some students who were OWOB were misclassified as normal weight. Additionally, we have used the WHO BMI-Z scores cut-off values for weight categories which is accepted by the Israeli Ministry of Health and not the International Obesity Task Force which makes it harder to compare our results to the standard HBSC prevalence rates of weight categories. Our research is based on the 2014 survey data so social and weight patterns may have changed; however, those do not change rapidly and there was no national programme which can justify such change. The main strength of the study was the use of a large and representative Israeli sample, which allowed for comparison between the different variables,
In conclusion, the present study identifies that female sex, high-school attendance and upper socio-economic status are associated with overestimation of weight status, while support from significant adults and peers reduces the risk for this perception.
Acknowledgements
Acknowledgements: We would like to acknowledge the study participants for their valuable contributions to the survey. Financial support: This study received no specific grant from any funding agency, commercial or not-for-profit sectors. Conflict of interest: There are no conflicts of interest. Authorship: S.B.Y. was responsible for managing data, conducting the statistical analysis, ensuring accuracy of results and drafting the manuscript. V.K.S. contributed to managing data, conducting the statistical analysis, interpretation of the results, and critical review and revisions of the manuscript. M.B. provided analytical guidance and contributed to manuscript review and revisions. All authors have read and agreed to the final version of the manuscript. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the Institutional Ethics Committee (Helsinki Committee), Ariel University, Israel. Written informed consent was obtained from all subjects. Verbal consent was witnessed and formally recorded.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980021002676