Int J Adolesc Med Health 2015; 27(1): 31–39
Anders Raustorp* and Magnus Lindwall
Physical self-esteem – a ten-year follow-up study from early adolescence to early adulthood Abstract Introduction: One variable that has been consistently associated with adolescents’ physical activity is perceived activity competence. Perceived physical (or sport) competence is considered a sub-domain to the physical selfesteem or self-worth (i.e., a person’s valuation of what is good and worthy in their self-description). Objective: This study aimed to describe levels of and intercorrelations among physical self-esteem, physical activity, and body mass index in a longitudinal design spanning adolescence to early adulthood. Materials and methods: At mean ages of 12.7, 15.7, 17.7 and 22.7 years, we measured perceived physical self-esteem in 39 (22 boys) Swedish adolescents. Physical activity (steps/ day) for four consecutive schooldays, height, and weight were also measured. Results: No significant difference between the four time points for any variable of perceived physical self-esteem was seen, neither in boys nor girls. In general, all physical self-variables revealed non-linear trajectories across time, where the general trend was an increase during the younger ages followed by a decrease during older ages. At ages 12 and 15 years in boys and girls physical condition and physical strength as well as body attractiveness and physical strength, respectively, had the strongest correlations to physical self-esteem. At age 17 and 22 years sports competence had the strongest correlation to self-esteem in girls, while body attractiveness and physical strength had the strongest correlation to self-esteem in boys. Conclusion: An overall stability in physical self-esteem was found. However the impact of a sub-domain upon physical self-esteem vary during adolescence and early adulthood. Such information may be useful when creating physical activity programs that support and develop physical self-esteem. *Corresponding author: Anders Raustorp, Department of Sport Science, Linnaeus University, Kalmar, SE 391 82 Kalmar, Sweden, Phone: +46 480 446358, Fax +46 480 446354, E-mail: [email protected]; and Department of Food, Nutrition and Sports Science, University of Gothenburg, Sweden Magnus Lindwall: Department of Food, Nutrition and Sports Science, University of Gothenburg, Sweden; and Department of Psychology, University of Gothenburg, Sweden
Keywords: BMI; CY PSPP; longitudinal; pedometers; physical self esteem. DOI 10.1515/ijamh-2014-0001 Received January 28, 2014; accepted February 26, 2014; previously published online May 6, 2014
Introduction Levels of physical inactivity are rising in many countries. This phenomenon has major implications for the prevalence of non-communicable diseases (e.g., cardiovascular disease, diabetes, and cancer) and their risk factors (e.g., raised blood sugar and blood pressure and overweight) (1). Sweden is no exception from this trend. A change in body mass index (BMI) of Swedish children and adolescents has occurred during the last decades, resulting in a 2.5-fold increased prevalence of overweight/obesity between 1987 and 2001 (2, 3). More recent research in Sweden reports that, over the last two decades, there has been an increase of overweight/ obesity in 4- and 10-year-old children, especially in girls, but not among 16-year-olds (4). Understanding why people are physically active or inactive contributes to evidence-based planning of public health interventions. This is because effective programs can target factors known to cause inactivity and urgently. The World Health Organization (WHO) (5) calls for actions to monitor data and key influences on physical activity, including methods for evaluation of intervention programs. Research into correlates (i.e., factors associated with activity) or determinants (i.e., those with a causal relationship) examine biological, environmental, social and psychological factors, which influence physical activity and play important roles in promoting physical activity among the youth (6). One variable that is consistently associated with adolescents’ physical activity is perceived activity competence (7). Perceived physical (or sport) competence is considered a sub-domain to the physical selfesteem or self-worth, i.e., a person’s valuation of what is good and worthy in their self-description (8). Low global self-worth is further reported to be connected to overweight/obesity (9). Brought to you by | University of Exeter Authenticated Download Date | 10/8/16 8:43 PM
32 Raustorp and Lindwall: Follow-up study on physical self-esteem
Global self-worth
Physical selfesteem
Sport competence
Body attractiveness
Physical strength
Physical condition
Figure 1 Hierarchal model of the Physical Self-Perception Profile (Fox an Corbin, 1989).
Harter (10) presented a hierarchical self-esteem model composed of different aspects, such as social, cognitive, and physical. Fox and Corbin (11) developed the physical self-perception profile (PSPP) and identified four different sub-domains, namely, sport competence, body attractiveness, physical strength and physical condition, in order to build up one’s physical self-esteem (Figure 1) Based on PSPP, a scale for youth was later created, the Children and Youth Physical Self Perceptions Profile (CYPSPP) (8). Motivational theories, which state that individuals are drawn towards activities in which they can show a high degree of skill or competence (12), lead to the conclusion that physical self-esteem should be positively related to levels of physical activity. Systematic reviews of randomized controlled studies conclude that exercise may have, at the very least, short-term beneficial effects on global (13) and physical self-esteem (14) in children and adolescents. Few longitudinal studies of physical self-esteem using the CY-PSPP exist. Strauss (9) found in a 4-year followup study of 9- to 10-year-old children that self-esteem levels decreased in obese girls and that they demonstrated higher rates of sadness and loneliness. Stability in all the sub-domains over time was reported in a 3-year follow-up study of 12- to 15-year-olds in Sweden, and the level of physical self-esteem in girls was identified as the most important predictor for a healthy lifestyle com bination (i.e., high physical activity, normal weight, and high physical self-esteem) (15). The stability over time in both genders reported in that study and that reported in the 5-year follow-up study of 12- to 17-year-olds (16) are in line with previously reported findings on studies on Estonian (17) and American (18) youth. The 5-year follow-up study (16) further concluded a strong causal inverse link between BMI/body fat variables and perceived physical self-esteem. What caused the other is not known but if changing one variable, then the other will be affected (16). This result means that creating a situation [i.e., during Physical Education (PE) or sport participation]
that increases the perceived physical self-esteem can lead to lower BMI/body fat levels 5 years later and vice versa. Regression analysis indicated that BMI and body fat both increasingly counter-predicted physical self-esteem in girls with increased age (16). Obviously, the triad physical activity, body composition, and perceived physical selfesteem all influence one another. Identifying inactive youth early and creating a situation that supports the perceived physical self-esteem seems to be a priority for health care system administrators, teachers, researchers, and policy makers. However, studies on physical self-esteem influences on youth physical activity have mostly been done by cross-sectional design and with self-reported physical activity; such studies normally do not include body composition. Even though earlier studies have indicated stability over time in subdomains, it is of interest to explore whether the strengths in the correlations between subdomains and physical self-esteem vary across this time span. If, for example, condition and strength are considered important factors in the earlier adolescence, this might influence how we create PE programs or intervention programs for this age group. In a recent cross-sectional study by Velez et al. (19), they reported that a 12-week resistance training intervention in 16-year-old Hispanic adolescents increased perceived condition, body adequacy, and global self-esteem. To our knowledge, no other study has been published with a longitudinal design covering the time span from early adolescence to early adulthood. The aim of this study was to describe levels and intraindividual changes in perceived physical self-perceptions, physical activity as daily mean steps, and body composition using a longitudinal design from early adolescence to adulthood with four time points of measurements (mean ages of 12.7, 15.7, 17.7, and 22.7 years). It also aimed to assess the correlation between sub-domains, physical selfesteem, global self-worth, physical activity, and BMI at the four time points. Another goal was to explore predictors of global self-worth and physical self-esteem across ages. Brought to you by | University of Exeter Authenticated Download Date | 10/8/16 8:43 PM
Raustorp and Lindwall: Follow-up study on physical self-esteem 33
Materials and methods Subjects In October 2000, as part of an international study on schoolchildren, data on daily mean step, BMI, and perceived physical self-esteem were collected from 289 boys and girls aged 12–14 years (mean age 12.7 years) in Teacher Training schools in Kalmar, Oskarshamn, and Mörbylånga – all middle class Swedish communities (20). The schools in these communities were categorized as countryside schools, suburban schools, or city schools. Parental, child, and school consent was obtained in writing and formed the sample. Reasons for nonparticipation were lack of parental permission, missing the first day of the study, illness, and choosing not to participate. In October 2003, school adolescents in the same geographical area were asked to participate. A total of 295 adolescents aged 15–17 years (mean age 15.7 years) were included in the study. They were measured for the same variables and with the same methodology as those cited in a 2000 data collection (15). A total of 109 adolescents were identified as individuals who were measured for daily mean steps height and weight in 2000 as well as in 2003; and they were asked to participate for measures for a third time. In October 2005 and October 2010, 77 (36 boys and 41 girls, 70% of the possible 109) and 39 (36% of the possible 109) participants, respectively, were measured for the same variables with the same methodology for the third and fourth times. Included in the study are students who completed all four test occasions. Reasons cited for not participating a third and fourth time were lack of interest and inability to track the individual. Participants in the study were those whose data were collected at all four time points. The study was approved by the Research Ethics Committee at Linköping University (Dnr. M109-05) (Dnr.2010/100-32).
Assessments Collection of physical activity data At all four occasions (2000, 2003, 2005, and 2010), daily step counts were measured by cable-tie sealed Yamax SW-200 pedometers (Tokyo, Japan) during four consecutive October weekdays. The pedometers were attached to the waistband and placed in line with the midpoint of the right knee. PE teachers were trained to collect data. Each day in a 24-h interval, the researcher collected the pedometers, documented the number of steps, resealed the devices, and returned them to the participants. To ensure that the pedometer was carried according to given instructions during the entire day, a questionnaire was completed by all the participants. In 2005, 10 of the individuals had graduated from school, and from them self-reported data collected on mailed pedometers were recorded. In 2010, data collected from mailed pedometers were recorded from all the 39 participants.
Collection of physical self-esteem data Physical self-esteem was assessed with the Children and Youth Physical Self-Perception Profile (CY-PSPP) (8). The CY-PSPP has been
found valid and reliable for youth as young as 9 years of age (21). A Swedish version has been tested for reliability and validity (20). This version consists of 36 items about children’s physical self-perception. The items represent six domains of self-perception, including global self-worth, physical self-esteem, sport competence, body attractiveness, physical strength, and physical condition. Each item consists of two statements and a four-point structured alternative format. First, the participants must decide which of the two statements best describes them, after which they must mark (with X) whether the statement is “kind of” true or “really” true for them. The assistant carefully described the procedure and read both statements for each item. At ages 15 and 17 years, the assistant carefully described the procedure but did not pre-read the items. At age 22 years, questionnaires were mailed to the participants accompanied by detailed instructions and a return envelope. Every item gives a score of one to four, where a score of four means highest self-perception on that item. The maximum score on all six items of a subdomain is 24 points and the minimum is six points; consequently, the maximum and minimum scores on the CY-PSPP are 144 points and 36 points, respectively. Error of measurement in a sub-domain is reported, due to Bland-Altman weighted κ, to range from 0.06 to 1.06 and 0.88 in the aspect of physical self-esteem (20).
Collection of BMI data Measures of height and weight were performed for each participant to determine their BMI. Height was measured on a tape attached to a wall (Friedrich Richter; Kirchenlaibach, Germany) and rounded down to the nearest centimeter. Weight was measured on step-up scales (EKS International, Wittiesheim, France and HEFA Digital AB Halmstad, Sweden) and rounded up to the nearest kilogram. Overweight and obesity were determined based on age- and sex-specific reference values developed by the International Obesity Task Force (22). At age 22 years height and weight were self-reported.
Statistical analysis Data were analyzed using SPSS 19.0 for Windows. All data, parametric, and descriptive mean standard deviations were reported. Independent t-test (daily mean steps, BMI) and Mann-Whitney U-test (self-perceptions) were used to analyze the representativeness of the follow-up sample. Independent t-tests were used to analyze the differences between boys’ and girls’ daily mean steps. To calculate the differences among time points 12, 15, 17, and 22 years in the followup group, repeated measures ANOVA for daily steps were conducted. The correlations among physical self-perceptions (Spearman’s rank correlation), BMI, and daily mean steps (Pearson product-moment correlation) were calculated. Latent growth curve models (LGCM), which are a specific case of mixed or random effects models (23, 24) were used to analyze intraindividual changes in global self-worth, physical self-esteem, BMI, and steps per day. In LGCM within the structural equation modeling framework, observed repeated measures across time are used as indicators of unobserved (latent) underlying true growth trajectory factors (25). In the present study, LGCM was fit to data using Mplus (version 5.21, Muthen and Muthen, 1998–2009) with
Brought to you by | University of Exeter Authenticated Download Date | 10/8/16 8:43 PM
34 Raustorp and Lindwall: Follow-up study on physical self-esteem a robust maximum likelihood estimator. Collinearity diagnostics was performed given the risk of multicollinearity in the regression analyses due to high correlations between the CY-PSPP subfactors. The tolerance statistics was clearly above 0.2 for all analyses, and the average variance inflation factor (VIF) was not substantially over 1 for any analyses. Therefore, no apparent risk for multicollinearity was detected. In addition, a 0.1% level of confidence was maintained unless otherwise stated.
Results Descriptive analysis Descriptive data regarding height in centimeter, weight in kilogram, calculated BMI, standard deviations, and perceived physical self-esteem at baseline and at 10 year follow-up are presented in Table 1. The attrition analysis showed a significant difference regarding BMI (p = 0.009) between the follow-up group and the remaining 250 of the comparator group of 289 individuals from the year 2000 baseline measurement. No differences were observed in terms of physical activity level (as daily mean steps) or in any of the domains of physical self-perception between the baseline group and the follow-up group. BMI at baseline showed that 14.8% of the boys and 14.5% of the girls were overweight and that 2.2% of the boys and 2.0% of the girls were obese. In general, boys’ self-perception scores were higher than those of girls. No significant difference between the four time points for any variable of perceived physical self-esteem was observed, neither in boys nor girls. Values for perceived self-perceptions, BMI, daily mean steps at all four time points, and gender differences found are presented in Tables 2 and 3 for boys and girls, respectively.
Correlation analysis At age 12 years, physical condition in boys (rs = 0.741) and body attractiveness in girls (rs = 0.840) had the strongest correlations to physical self-esteem. Three years later, physical strength was most important for both boys (rs = 0.852) and girls (rs = 0.699). At age 17 years, sports competence had the strongest correlation to physical selfesteem in girls (rs = 0.840), in boys it was body attractiveness (rs = 0.661). Finally at age 22 years, sports competence in girls (rs = 0.902) still had the strongest correlation to physical self-esteem, while in boys it was physical strength (rs = 0.627). At all four time points from ages 12 to 22 years, body attractiveness had the strongest correlation to global self-worth in both boys (rs = 0.760, 0.714, 0.652 and 0.631, respectively) and girls (rs = 0.893, 0.690, 0.766 and 0.669, respectively). Physical activity showed poor correlation to all the sub-domains and to physical self-esteem in both boys and girls.
Intraindividual changes across time The fixed estimates for the latent growth curve analyses are described in Table 4. In general, all physical self-variables, including global self-worth, revealed non-linear trajectories across time, where the general trend was an increase during the younger ages followed by a decrease (or an decelerated increased) during older ages. Mirroring this trend, the slope for global self-worth was positive and significant, whereas the quadratic term was negative and significant. The same pattern was found for physical condition and body attractiveness, with the exception that the slopes for both variables, and the quadratic term for body attractiveness were
Table 1 Descriptive data over height in centimeter, weight in kilogram, calculated body mass index standard deviations, and perceived physical self-esteem at baseline and at ten-year follow-up.
Baseline n = 250
Age Height Weight BMI Mean steps Sport competence Physical condition Body attractiveness Physical strength Physical self to esteem Global self-worth
13.0 162 cm 52.8 19.9 14,318 15.7 15.8 15.2 15.5 16.4 17.5
SD alt range
FOU group n = 39
SD alt range
ns ns p = 0.014 p = 0.014 ns p = 0.025 ns ns ns ns ns
0.8 8.6 12.6 3.7 4038 7–24 7–24 6–24 6–24 7–24 6–24
12.7 160 cm 48.7 18.9 13,758 16.9 16.2 16.0 15.9 17.1 18.5
0.7 7.1 8.4 2.4 3008 7–23 8–23 8–22 7–23 6–24 9–24
Brought to you by | University of Exeter Authenticated Download Date | 10/8/16 8:43 PM
Raustorp and Lindwall: Follow-up study on physical self-esteem 35 Table 2 Values for perceived self-perceptions, BMI, daily mean steps at all four time points for boys. n = 22
Time 1 Baseline 2000
Age BMI Mean steps Sport competence Physical condition Body attractiveness Physical strength Physical self-esteem Global self-worth
12.5 18.6 14,696 16.9 16.4 16.1 15.6 17.4 18.9
SD Time 2 2003 altrange
0.7 2.3 3057 7–23 8–22 8–22 7–21 6–21 11–24
15.5 20.1 11,432 17.4 16.9 18.0 16.9 18.5 21.0
SD Time 3 2005 altrange
0.7 1.9 2688 8–24 7–24 12–24 12–24 12–24 14–24
17.5 21.2 10,356 17.7 17.8 18.2 16.9 18.2 20.4
SD Time 4 2010 altrange
0.7 2.5 2314 8–24 6–24 11–24 10–24 11–24 13–24
22.5 22.9 10,542 17.5 15.9 16.2 16.1 17.2 19.1
SD altrange
0.7 2.7 3571 9–24 2.7 6–24 6–24 12–23 12–24
Time 4 2010
SD altrange
Table 3 Values for perceived self-perceptions, BMI, daily mean steps at all four time points for girls. Age BMI Mean steps Sport competence Physical condition Body attractiveness Physical strength Physical self-esteem Global self-worth
Time 1 Baseline 2000 12.9 18.9 13,053 17.3 16.2 16.3 15.8 17.4 19.0
SD Time 2 2003 altrange 0.8 2.8 2638 12–23 10–23 9–20 9–20 14–24 13–24
15.9 21.3 13,421 18.1 18.5 15.9 15.7 17.9 19.9
non-significant trends (p
Anders Raustorp* and Magnus Lindwall
Physical self-esteem – a ten-year follow-up study from early adolescence to early adulthood Abstract Introduction: One variable that has been consistently associated with adolescents’ physical activity is perceived activity competence. Perceived physical (or sport) competence is considered a sub-domain to the physical selfesteem or self-worth (i.e., a person’s valuation of what is good and worthy in their self-description). Objective: This study aimed to describe levels of and intercorrelations among physical self-esteem, physical activity, and body mass index in a longitudinal design spanning adolescence to early adulthood. Materials and methods: At mean ages of 12.7, 15.7, 17.7 and 22.7 years, we measured perceived physical self-esteem in 39 (22 boys) Swedish adolescents. Physical activity (steps/ day) for four consecutive schooldays, height, and weight were also measured. Results: No significant difference between the four time points for any variable of perceived physical self-esteem was seen, neither in boys nor girls. In general, all physical self-variables revealed non-linear trajectories across time, where the general trend was an increase during the younger ages followed by a decrease during older ages. At ages 12 and 15 years in boys and girls physical condition and physical strength as well as body attractiveness and physical strength, respectively, had the strongest correlations to physical self-esteem. At age 17 and 22 years sports competence had the strongest correlation to self-esteem in girls, while body attractiveness and physical strength had the strongest correlation to self-esteem in boys. Conclusion: An overall stability in physical self-esteem was found. However the impact of a sub-domain upon physical self-esteem vary during adolescence and early adulthood. Such information may be useful when creating physical activity programs that support and develop physical self-esteem. *Corresponding author: Anders Raustorp, Department of Sport Science, Linnaeus University, Kalmar, SE 391 82 Kalmar, Sweden, Phone: +46 480 446358, Fax +46 480 446354, E-mail: [email protected]; and Department of Food, Nutrition and Sports Science, University of Gothenburg, Sweden Magnus Lindwall: Department of Food, Nutrition and Sports Science, University of Gothenburg, Sweden; and Department of Psychology, University of Gothenburg, Sweden
Keywords: BMI; CY PSPP; longitudinal; pedometers; physical self esteem. DOI 10.1515/ijamh-2014-0001 Received January 28, 2014; accepted February 26, 2014; previously published online May 6, 2014
Introduction Levels of physical inactivity are rising in many countries. This phenomenon has major implications for the prevalence of non-communicable diseases (e.g., cardiovascular disease, diabetes, and cancer) and their risk factors (e.g., raised blood sugar and blood pressure and overweight) (1). Sweden is no exception from this trend. A change in body mass index (BMI) of Swedish children and adolescents has occurred during the last decades, resulting in a 2.5-fold increased prevalence of overweight/obesity between 1987 and 2001 (2, 3). More recent research in Sweden reports that, over the last two decades, there has been an increase of overweight/ obesity in 4- and 10-year-old children, especially in girls, but not among 16-year-olds (4). Understanding why people are physically active or inactive contributes to evidence-based planning of public health interventions. This is because effective programs can target factors known to cause inactivity and urgently. The World Health Organization (WHO) (5) calls for actions to monitor data and key influences on physical activity, including methods for evaluation of intervention programs. Research into correlates (i.e., factors associated with activity) or determinants (i.e., those with a causal relationship) examine biological, environmental, social and psychological factors, which influence physical activity and play important roles in promoting physical activity among the youth (6). One variable that is consistently associated with adolescents’ physical activity is perceived activity competence (7). Perceived physical (or sport) competence is considered a sub-domain to the physical selfesteem or self-worth, i.e., a person’s valuation of what is good and worthy in their self-description (8). Low global self-worth is further reported to be connected to overweight/obesity (9). Brought to you by | University of Exeter Authenticated Download Date | 10/8/16 8:43 PM
32 Raustorp and Lindwall: Follow-up study on physical self-esteem
Global self-worth
Physical selfesteem
Sport competence
Body attractiveness
Physical strength
Physical condition
Figure 1 Hierarchal model of the Physical Self-Perception Profile (Fox an Corbin, 1989).
Harter (10) presented a hierarchical self-esteem model composed of different aspects, such as social, cognitive, and physical. Fox and Corbin (11) developed the physical self-perception profile (PSPP) and identified four different sub-domains, namely, sport competence, body attractiveness, physical strength and physical condition, in order to build up one’s physical self-esteem (Figure 1) Based on PSPP, a scale for youth was later created, the Children and Youth Physical Self Perceptions Profile (CYPSPP) (8). Motivational theories, which state that individuals are drawn towards activities in which they can show a high degree of skill or competence (12), lead to the conclusion that physical self-esteem should be positively related to levels of physical activity. Systematic reviews of randomized controlled studies conclude that exercise may have, at the very least, short-term beneficial effects on global (13) and physical self-esteem (14) in children and adolescents. Few longitudinal studies of physical self-esteem using the CY-PSPP exist. Strauss (9) found in a 4-year followup study of 9- to 10-year-old children that self-esteem levels decreased in obese girls and that they demonstrated higher rates of sadness and loneliness. Stability in all the sub-domains over time was reported in a 3-year follow-up study of 12- to 15-year-olds in Sweden, and the level of physical self-esteem in girls was identified as the most important predictor for a healthy lifestyle com bination (i.e., high physical activity, normal weight, and high physical self-esteem) (15). The stability over time in both genders reported in that study and that reported in the 5-year follow-up study of 12- to 17-year-olds (16) are in line with previously reported findings on studies on Estonian (17) and American (18) youth. The 5-year follow-up study (16) further concluded a strong causal inverse link between BMI/body fat variables and perceived physical self-esteem. What caused the other is not known but if changing one variable, then the other will be affected (16). This result means that creating a situation [i.e., during Physical Education (PE) or sport participation]
that increases the perceived physical self-esteem can lead to lower BMI/body fat levels 5 years later and vice versa. Regression analysis indicated that BMI and body fat both increasingly counter-predicted physical self-esteem in girls with increased age (16). Obviously, the triad physical activity, body composition, and perceived physical selfesteem all influence one another. Identifying inactive youth early and creating a situation that supports the perceived physical self-esteem seems to be a priority for health care system administrators, teachers, researchers, and policy makers. However, studies on physical self-esteem influences on youth physical activity have mostly been done by cross-sectional design and with self-reported physical activity; such studies normally do not include body composition. Even though earlier studies have indicated stability over time in subdomains, it is of interest to explore whether the strengths in the correlations between subdomains and physical self-esteem vary across this time span. If, for example, condition and strength are considered important factors in the earlier adolescence, this might influence how we create PE programs or intervention programs for this age group. In a recent cross-sectional study by Velez et al. (19), they reported that a 12-week resistance training intervention in 16-year-old Hispanic adolescents increased perceived condition, body adequacy, and global self-esteem. To our knowledge, no other study has been published with a longitudinal design covering the time span from early adolescence to early adulthood. The aim of this study was to describe levels and intraindividual changes in perceived physical self-perceptions, physical activity as daily mean steps, and body composition using a longitudinal design from early adolescence to adulthood with four time points of measurements (mean ages of 12.7, 15.7, 17.7, and 22.7 years). It also aimed to assess the correlation between sub-domains, physical selfesteem, global self-worth, physical activity, and BMI at the four time points. Another goal was to explore predictors of global self-worth and physical self-esteem across ages. Brought to you by | University of Exeter Authenticated Download Date | 10/8/16 8:43 PM
Raustorp and Lindwall: Follow-up study on physical self-esteem 33
Materials and methods Subjects In October 2000, as part of an international study on schoolchildren, data on daily mean step, BMI, and perceived physical self-esteem were collected from 289 boys and girls aged 12–14 years (mean age 12.7 years) in Teacher Training schools in Kalmar, Oskarshamn, and Mörbylånga – all middle class Swedish communities (20). The schools in these communities were categorized as countryside schools, suburban schools, or city schools. Parental, child, and school consent was obtained in writing and formed the sample. Reasons for nonparticipation were lack of parental permission, missing the first day of the study, illness, and choosing not to participate. In October 2003, school adolescents in the same geographical area were asked to participate. A total of 295 adolescents aged 15–17 years (mean age 15.7 years) were included in the study. They were measured for the same variables and with the same methodology as those cited in a 2000 data collection (15). A total of 109 adolescents were identified as individuals who were measured for daily mean steps height and weight in 2000 as well as in 2003; and they were asked to participate for measures for a third time. In October 2005 and October 2010, 77 (36 boys and 41 girls, 70% of the possible 109) and 39 (36% of the possible 109) participants, respectively, were measured for the same variables with the same methodology for the third and fourth times. Included in the study are students who completed all four test occasions. Reasons cited for not participating a third and fourth time were lack of interest and inability to track the individual. Participants in the study were those whose data were collected at all four time points. The study was approved by the Research Ethics Committee at Linköping University (Dnr. M109-05) (Dnr.2010/100-32).
Assessments Collection of physical activity data At all four occasions (2000, 2003, 2005, and 2010), daily step counts were measured by cable-tie sealed Yamax SW-200 pedometers (Tokyo, Japan) during four consecutive October weekdays. The pedometers were attached to the waistband and placed in line with the midpoint of the right knee. PE teachers were trained to collect data. Each day in a 24-h interval, the researcher collected the pedometers, documented the number of steps, resealed the devices, and returned them to the participants. To ensure that the pedometer was carried according to given instructions during the entire day, a questionnaire was completed by all the participants. In 2005, 10 of the individuals had graduated from school, and from them self-reported data collected on mailed pedometers were recorded. In 2010, data collected from mailed pedometers were recorded from all the 39 participants.
Collection of physical self-esteem data Physical self-esteem was assessed with the Children and Youth Physical Self-Perception Profile (CY-PSPP) (8). The CY-PSPP has been
found valid and reliable for youth as young as 9 years of age (21). A Swedish version has been tested for reliability and validity (20). This version consists of 36 items about children’s physical self-perception. The items represent six domains of self-perception, including global self-worth, physical self-esteem, sport competence, body attractiveness, physical strength, and physical condition. Each item consists of two statements and a four-point structured alternative format. First, the participants must decide which of the two statements best describes them, after which they must mark (with X) whether the statement is “kind of” true or “really” true for them. The assistant carefully described the procedure and read both statements for each item. At ages 15 and 17 years, the assistant carefully described the procedure but did not pre-read the items. At age 22 years, questionnaires were mailed to the participants accompanied by detailed instructions and a return envelope. Every item gives a score of one to four, where a score of four means highest self-perception on that item. The maximum score on all six items of a subdomain is 24 points and the minimum is six points; consequently, the maximum and minimum scores on the CY-PSPP are 144 points and 36 points, respectively. Error of measurement in a sub-domain is reported, due to Bland-Altman weighted κ, to range from 0.06 to 1.06 and 0.88 in the aspect of physical self-esteem (20).
Collection of BMI data Measures of height and weight were performed for each participant to determine their BMI. Height was measured on a tape attached to a wall (Friedrich Richter; Kirchenlaibach, Germany) and rounded down to the nearest centimeter. Weight was measured on step-up scales (EKS International, Wittiesheim, France and HEFA Digital AB Halmstad, Sweden) and rounded up to the nearest kilogram. Overweight and obesity were determined based on age- and sex-specific reference values developed by the International Obesity Task Force (22). At age 22 years height and weight were self-reported.
Statistical analysis Data were analyzed using SPSS 19.0 for Windows. All data, parametric, and descriptive mean standard deviations were reported. Independent t-test (daily mean steps, BMI) and Mann-Whitney U-test (self-perceptions) were used to analyze the representativeness of the follow-up sample. Independent t-tests were used to analyze the differences between boys’ and girls’ daily mean steps. To calculate the differences among time points 12, 15, 17, and 22 years in the followup group, repeated measures ANOVA for daily steps were conducted. The correlations among physical self-perceptions (Spearman’s rank correlation), BMI, and daily mean steps (Pearson product-moment correlation) were calculated. Latent growth curve models (LGCM), which are a specific case of mixed or random effects models (23, 24) were used to analyze intraindividual changes in global self-worth, physical self-esteem, BMI, and steps per day. In LGCM within the structural equation modeling framework, observed repeated measures across time are used as indicators of unobserved (latent) underlying true growth trajectory factors (25). In the present study, LGCM was fit to data using Mplus (version 5.21, Muthen and Muthen, 1998–2009) with
Brought to you by | University of Exeter Authenticated Download Date | 10/8/16 8:43 PM
34 Raustorp and Lindwall: Follow-up study on physical self-esteem a robust maximum likelihood estimator. Collinearity diagnostics was performed given the risk of multicollinearity in the regression analyses due to high correlations between the CY-PSPP subfactors. The tolerance statistics was clearly above 0.2 for all analyses, and the average variance inflation factor (VIF) was not substantially over 1 for any analyses. Therefore, no apparent risk for multicollinearity was detected. In addition, a 0.1% level of confidence was maintained unless otherwise stated.
Results Descriptive analysis Descriptive data regarding height in centimeter, weight in kilogram, calculated BMI, standard deviations, and perceived physical self-esteem at baseline and at 10 year follow-up are presented in Table 1. The attrition analysis showed a significant difference regarding BMI (p = 0.009) between the follow-up group and the remaining 250 of the comparator group of 289 individuals from the year 2000 baseline measurement. No differences were observed in terms of physical activity level (as daily mean steps) or in any of the domains of physical self-perception between the baseline group and the follow-up group. BMI at baseline showed that 14.8% of the boys and 14.5% of the girls were overweight and that 2.2% of the boys and 2.0% of the girls were obese. In general, boys’ self-perception scores were higher than those of girls. No significant difference between the four time points for any variable of perceived physical self-esteem was observed, neither in boys nor girls. Values for perceived self-perceptions, BMI, daily mean steps at all four time points, and gender differences found are presented in Tables 2 and 3 for boys and girls, respectively.
Correlation analysis At age 12 years, physical condition in boys (rs = 0.741) and body attractiveness in girls (rs = 0.840) had the strongest correlations to physical self-esteem. Three years later, physical strength was most important for both boys (rs = 0.852) and girls (rs = 0.699). At age 17 years, sports competence had the strongest correlation to physical selfesteem in girls (rs = 0.840), in boys it was body attractiveness (rs = 0.661). Finally at age 22 years, sports competence in girls (rs = 0.902) still had the strongest correlation to physical self-esteem, while in boys it was physical strength (rs = 0.627). At all four time points from ages 12 to 22 years, body attractiveness had the strongest correlation to global self-worth in both boys (rs = 0.760, 0.714, 0.652 and 0.631, respectively) and girls (rs = 0.893, 0.690, 0.766 and 0.669, respectively). Physical activity showed poor correlation to all the sub-domains and to physical self-esteem in both boys and girls.
Intraindividual changes across time The fixed estimates for the latent growth curve analyses are described in Table 4. In general, all physical self-variables, including global self-worth, revealed non-linear trajectories across time, where the general trend was an increase during the younger ages followed by a decrease (or an decelerated increased) during older ages. Mirroring this trend, the slope for global self-worth was positive and significant, whereas the quadratic term was negative and significant. The same pattern was found for physical condition and body attractiveness, with the exception that the slopes for both variables, and the quadratic term for body attractiveness were
Table 1 Descriptive data over height in centimeter, weight in kilogram, calculated body mass index standard deviations, and perceived physical self-esteem at baseline and at ten-year follow-up.
Baseline n = 250
Age Height Weight BMI Mean steps Sport competence Physical condition Body attractiveness Physical strength Physical self to esteem Global self-worth
13.0 162 cm 52.8 19.9 14,318 15.7 15.8 15.2 15.5 16.4 17.5
SD alt range
FOU group n = 39
SD alt range
ns ns p = 0.014 p = 0.014 ns p = 0.025 ns ns ns ns ns
0.8 8.6 12.6 3.7 4038 7–24 7–24 6–24 6–24 7–24 6–24
12.7 160 cm 48.7 18.9 13,758 16.9 16.2 16.0 15.9 17.1 18.5
0.7 7.1 8.4 2.4 3008 7–23 8–23 8–22 7–23 6–24 9–24
Brought to you by | University of Exeter Authenticated Download Date | 10/8/16 8:43 PM
Raustorp and Lindwall: Follow-up study on physical self-esteem 35 Table 2 Values for perceived self-perceptions, BMI, daily mean steps at all four time points for boys. n = 22
Time 1 Baseline 2000
Age BMI Mean steps Sport competence Physical condition Body attractiveness Physical strength Physical self-esteem Global self-worth
12.5 18.6 14,696 16.9 16.4 16.1 15.6 17.4 18.9
SD Time 2 2003 altrange
0.7 2.3 3057 7–23 8–22 8–22 7–21 6–21 11–24
15.5 20.1 11,432 17.4 16.9 18.0 16.9 18.5 21.0
SD Time 3 2005 altrange
0.7 1.9 2688 8–24 7–24 12–24 12–24 12–24 14–24
17.5 21.2 10,356 17.7 17.8 18.2 16.9 18.2 20.4
SD Time 4 2010 altrange
0.7 2.5 2314 8–24 6–24 11–24 10–24 11–24 13–24
22.5 22.9 10,542 17.5 15.9 16.2 16.1 17.2 19.1
SD altrange
0.7 2.7 3571 9–24 2.7 6–24 6–24 12–23 12–24
Time 4 2010
SD altrange
Table 3 Values for perceived self-perceptions, BMI, daily mean steps at all four time points for girls. Age BMI Mean steps Sport competence Physical condition Body attractiveness Physical strength Physical self-esteem Global self-worth
Time 1 Baseline 2000 12.9 18.9 13,053 17.3 16.2 16.3 15.8 17.4 19.0
SD Time 2 2003 altrange 0.8 2.8 2638 12–23 10–23 9–20 9–20 14–24 13–24
15.9 21.3 13,421 18.1 18.5 15.9 15.7 17.9 19.9
non-significant trends (p
