Development and psychometric assessment of the Sociocultural Influences on Exercise Behaviours in Adolescents Questionnaire

Abstract

Objectives

This research aimed to construct and psychometrically test a measure of multiple sociocultural dimensions (i.e. family, peers, media) theoretically associated with exercise behaviours/attitudes in adolescents; the Sociocultural Influences on Exercise Behaviours in Adolescents Questionnaire (SIEBAQ).

Methods and measurement

Part 1 of this study focused on measure construction and psychometric testing, involving item generation and exploratory factor analysis (EFA) to refine the item pool, with 905 adolescents (Mage 13.66 years (SD = 0.94); girls = 442). Part 2 sought to explore the convergent validity of the SIEBAQ (n = 846; n = 414 girls).

Results

EFA resulted in a 47-item measure with a nine-factor structure (including social media modelling, parent exercise expectations, peer co-participation; α = 0.72-0.92). Correlations revealed weak-moderate significant relationships between the SIEBAQ and related constructs (e.g. compulsive exercise, sociocultural attitudes towards appearance). Regression analyses with the SIEBAQ identified social media modelling of exercise as a significant predictor of compulsive exercise in boys and girls. Proving exercise ability to significant others also significantly predicted compulsive exercise outcomes.

Conclusion

This newly developed measure holds promise. Further psychometric testing and validation of the SIEBAQ is the recommended next step to confirm the measure’s nine-factor structure identified through EFA.

Keywords:

• adolescent
• young people
• teenager
• compulsive exercise
• sociocultural influence
• social media

Introduction

Adolescence, defined by the World Health Organization (World Health Organization, 2023) as the period between 10-19 years of age, is a key developmental phase for identity formation and social change (e.g. Krayer et al., 2008). Adolescents are particularly susceptible to social cues, which can motivate and influence certain behaviours at this age (Blakemore & Mills, 2014). Of particular interest are maladaptive eating and exercise behaviours, which can be triggered by body-related comparisons with society-endorsed ‘ideal’ body types, due to the significant changes in body size and shape which typically occur during this developmental period (e.g. Reel et al., 2015). While there are many benefits to exercise, such as decreased risk of morbidity and increased self-esteem (Lubans et al., 2009), adolescents often engage in exercise behaviours as an appearance-change strategy rather than for health-related purposes (e.g. Patton et al., 2016; Zach et al., 2013). Understanding the social cues which drive exercise behaviours at this age is therefore of critical importance.

Three well-documented sources of sociocultural influence on exercise behaviours in adolescents, as outlined by the tripartite influence model (Thompson et al., 1999), are parents, peers and traditional media (e.g. television and magazines; Goodwin et al., 2014; White & Halliwell, 2010). Adolescents are more likely to engage in exercise if their parents and/or peers are also physically active (Mackey & La Greca, 2007; Rodrigues et al., 2018). However, receipt of negative body-related messages from parents and peers has been significantly associated with elevated levels of compulsive exercise behaviours at this age (e.g. Neumark-Sztainer et al., 2006; Reynolds et al., 2022). Compulsive exercise is characterised as a severe and intense drive to exercise, despite illness or injury (Taranis et al., 2011), and as a way to manage negative emotions (Goodwin et al., 2012). However, additional influences on exercise beyond these known sources have yet to be explored in detail. For example, adolescents frequently co-participate in exercise with their siblings (e.g. Lisinskiene & Juskeliene, 2019), yet less is known about the potential influence that siblings may have on exercise psychopathology in adolescents; a research area that warrants further exploration (Blazo & Smith, 2017). As most adolescents typically have extensive contact with their siblings, both in the presence and absence of parent supervision (e.g. McHale et al., 2012), and parent influence is significantly associated with compulsive exercise in adolescents, it would be of interest for research to explore sibling influence as a predictor of compulsive exercise. Research with young adult women has identified the preventative nature of positive body-related messages from siblings on reduced compulsive exercise (Patterson & Goodson, 2018), but this has yet to be considered among adolescents.

In addition to sibling influence as a potential addition to the tripartite influence model, there is growing evidence to suggest that the model must now be adapted to include social media as a source of influence on body change strategies in adolescent populations (Roberts et al., 2022). The majority of UK adolescents aged 12-15 years use social media (87%; Ofcom, 2020), and often report using it to inform their exercise behaviours and perceptions of ‘ideal’ body types (Reynolds et al., 2022). As body-related messages and pressure to conform to ‘ideal’ body types portrayed in traditional media sources (e.g. television and magazines) have also been associated with increased compulsive exercise behaviours in adolescent boys and girls (e.g. Jankauskiene et al., 2019), it is therefore essential to capture social media as a source of influence on adolescent exercise behaviours and explore its potential significance on compulsive exercise at this age.

A recent systematic review of sociocultural influences on compulsive exercise in adolescents identified the three key influences of parents, peers and media (Reynolds et al., 2022), but found limited research to date exploring the role of siblings and social media. Furthermore, established measures such as the Perceived Sociocultural Pressure Scale (PSPS; Stice & Bearman, 2001) and the Sociocultural Attitudes Towards Appearance Questionnaire (SATAQ; Schaefer et al., 2015) have been used extensively in research, identifying positive relationships between the influences of parents, peers and traditional media with maladaptive exercise behaviours, such as compulsive exercise, in young people (e.g. Goodwin et al., 2011; Jankauskiene et al., 2019). As noted in the Reynolds et al. review, however, these measures are designed to capture the sociocultural influences on appearance and weight/shape concerns through body-related messages from parents, peers and traditional media sources. These measures, therefore, do not capture the sociocultural impact on exercise behaviours and attitudes more specifically.

Another established measure which captures peer influence on body image perceptions is the Body, Eating and Exercise Comparison Orientation Measure (BEECOM; Fitzsimmons-Craft et al., 2012). In contrast to the PSPS and SATAQ, the BEECOM also includes social comparison dimensions associated with eating and exercise psychopathology. While the BEECOM captures sociocultural influences on exercise behaviours and attitudes, the item pool is limited to peer influence and does not measure a multitude of sociocultural influences (as per the tripartite influence model). The BEECOM is also yet to be used in adolescent populations. As adult exercise behaviours are often established in adolescence (Patton et al., 2016) and adolescents are particularly susceptible to social influence (Blakemore & Mills, 2014), it is important that a comprehensive measure of sociocultural influences on exercise behaviours is developed for use with adolescent populations. Having a measure that captures the full scope of sociocultural influences on exercise behaviours will allow for a more accurate assessment of the sociocultural influences on exercise psychopathology. In addition, a measure of this nature will serve to facilitate a clearer understanding of the motives and barriers associated with physical activity engagement in adolescents over time and help to understand how intervention strategies can prevent maladaptive exercise behaviours prior to consolidation in adulthood.

Research aims

The aims of this research were twofold. The first aim of this study (Part 1) was to construct and psychometrically test a self-report tool capable of measuring a multitude of sociocultural influences that are theoretically associated with exercise behaviours in adolescents: the Sociocultural Influences on Exercise Behaviours in Adolescents Questionnaire (SIEBAQ). Development of the SIEBAQ involved several components and followed guidelines by Boateng et al. (2018) for scale development in health, social and behavioural research. Part 1 also aimed to use exploratory factor analysis (EFA) to identify the factor structure of the SIEBAQ and to refine the item pool. Inter-item reliability of the SIEBAQ was also assessed.

The second aim of this research (Part 2) was to use the newly developed measure (SIEBAQ) in cross-sectional research to explore convergent validity with related constructs (e.g. sociocultural attitudes towards appearance) and to identify sociocultural predictors of compulsive exercise in adolescents. It was hypothesised that, after accounting for related constructs (e.g. disordered eating, mental wellbeing), higher levels of sociocultural influences on exercise would significantly predict higher compulsive exercise outcomes in adolescents.

Part 1: Development of the SIEBAQ and item refinement via exploratory factor analysis

Method

Initial development of the Sociocultural Influences on Exercise Behaviours in Adolescents Questionnaire (SIEBAQ)

The SIEBAQ was designed to assess multiple sociocultural influences on exercise behaviours and attitudes in adolescents. Items were initially generated from key components identified in focus group discussions with adolescent boys and girls aged 12-16 years (see Reynolds et al., 2022). A comprehensive appraisal of literature exploring sociocultural influences on exercise behaviours and attitudes in young people, particularly in adolescent populations, was also used to inform item generation. Existing validated measures related to sociocultural influences on body change strategies, such as eating behaviours, were also consulted to assist with item development. Included items aimed to address five key sociocultural influences on exercise behaviours and attitudes in adolescents, as identified within the literature: (i) parents; (ii) siblings; (iii) peers; (iv) social media; (v) traditional media. Eighty-seven items were initially included in the measure, and, for each item, respondents were asked to rate how true each statement was for them on a Likert scale, with the anchors 1 (never true) to 5 (always true). The initial item pool was designed to be more comprehensive and broader than the target construct, erring on the side of over-inclusiveness to begin with (Clark & Watson, 2019).

The initial 87-item SIEBAQ was pilot tested with a sample of adolescent boys and girls (n = 60; mean age = 13.80, SD = 0.86; n = 32 females) to gather feedback on comprehension and relevance of the questionnaire items to the target population. Participants provided feedback via semi-structured interviews, focus group discussions, or open-ended and closed feedback questions after completion of the measure. Responses were analysed using content analysis and percentages for multiple choice (closed) questions were calculated to analyse frequencies of responses. Feedback on the clarity of the measure was largely positive. For example, most adolescents reported that they understood the questionnaire items and were totally or fairly sure what to do when following the questionnaire instructions. Participant responses resulted in small changes being made to the Likert scale anchors for questions such as “exercising with my friends is fun” and “exercise is important to my parent(s)” to include additional ‘not applicable’ and ‘don’t know’ anchors, respectively, for adolescents who did not exercise with their parent(s), peers or sibling(s), and/or for those who did not know how important exercise was to their significant others. These additional anchors are scored as missing data. Three additional items were also added to the SIEBAQ in response to feedback from the pilot study, which were “I exercise with my parent(s)/sibling(s)/peers” (respectively) to explore the level of co-participation in exercise behaviours with significant others, making the updated version of the SIEBAQ 90-items in length. Higher scores indicate a greater influence of the sociocultural sources on exercise behaviours and attitudes. The SIEBAQ demonstrated a Flesch (1948) Readability score of 72.1, which is considered ‘fairly easy’ and equivalent to the reading ability of US Grade 7 - i.e. 12-13 years, therefore indicating that the items were appropriately accessible to adolescent participants.

Participants and recruitment

Using purposive sampling, adolescent boys and girls (aged 12–15 years) were recruited from five secondary schools in the East Midlands, UK, via an initial email to the headteacher. Adolescents between the ages of 12–15 years were recruited to coincide with the onset of puberty as defined by the World Health Organization (e.g. 12–13 years; World Health Organization, 2014). Furthermore, maladaptive health behaviours (e.g. disordered eating) and sociocultural stressors such as peer pressure appear to increase between the ages of 12–15 years (e.g. Breton et al., 2022). Schools were recruited from diverse socioeconomic status areas, based on their postcode. Deprivation indices range from 1 to 10 with 1 indicating the most deprived areas to 10 being the least deprived (Ministry of Housing et al., 2019). Deprivation ranking of the areas in which the schools were recruited ranged between 2 and 9. The headteacher was informed of the required age range of participants and notified adolescents and their parents from at least one or all of the following year groups about the opportunity to participate in the study: year 8 (12–13 years), year 9 (13–14 years) and/or year 10 (14–15 years).

Procedure

Following institutional ethical review board approval [2022-4960-8593], each secondary school was provided with a parent information sheet, which was distributed to parents by the school, either electronically or via letter. Parents were given the opportunity to opt-out their child from completing the study. Preferences for paper-based or online questionnaire completion were indicated by each school and distributed accordingly. Participants completed the questionnaire during school time, in their classroom with their teacher. Following informed consent, participants were instructed by their teacher to complete the measure independently, without discussing their responses with others, and then paper questionnaires were returned to the teacher after completion and collected from each school by the researcher. In addition to the SIEBAQ measure, and as part of the broader scope of this two-part study, participants completed measures of exercise and eating psychopathology, mental wellbeing, physical activity levels and sociocultural influences on body image. These measures are detailed in Part 2. Participants also provided demographic information (gender, age and ethnicity). Questionnaires took approximately 30-45 min to complete.

Results

Descriptive statistics

A total of 905 participants were recruited for this study (boys n = 432, girls n = 442, prefer not to say n = 31). Participants’ mean age and percentages of those with a social media account and/or at least one sibling, are displayed and separated by gender in Table 1. The mean age of the participants was 13.66 years (SD: 0.94, range 12-15 years). Over half of the participants (54.9%, n = 497) identified their ethnicity as White British, 19.4% (n = 176) identified as Asian British, 8.8% (n = 80) Asian other, 5.7% (n = 52) ‘Other’, 4.9% (n = 44) White other, 2.0% (n = 18) Black other, 1.7% (n = 15) Black British, and 0.3% (n = 3) Chinese. Ethnicity was not reported by 20 (2.2%) participants. The majority of the participants reported having at least one sibling (92%) and almost all of the adolescents in the study reported using social media (92%). Chi-square analyses identified no significant differences in characteristics (siblings, social media use) by gender (X²(2) ≥ .411, φc ≥ .03; p > .05). In addition, a Kruskal-Wallis test identified no significant differences in age between genders (H(2) = 2.40, ε² < .01; p > .05).

Table 1. Descriptive statistics of the 905 participants separated by gender.

	Male (n = 432)	Female (n = 442)	Prefer not to say (n = 31)	Total sample (n = 905)
Mean age	13.69 (SD: 0.92)	13.64 (SD: 0.96)	13.42 (SD: 0.89)	13.66 (SD: 0.94)
One or more sibling(s)	399 (92.5%)	405 (91.6%)	29 (94.1%)	833 (92.0%)
Social media use	388 (89.8%)	418 (94.6%)	28 (91.2%)	834 (92.1%)

Exploratory factor analysis

To explore the factor structure of the SIEBAQ (90 items), an exploratory factor analysis (EFA) was conducted. EFA is considered a suitable approach for the preliminary stages of measure development (e.g. Flora & Flake, 2017), particularly when the number of factors and allocation of items to factors are unknown (Fokkema & Greiff, 2017). Using IBM SPSS Statistics 27, principal axis factoring was employed using a promax rotation. Promax is an oblique rotation which first assumes that factors are unrelated and then relaxes the rotation to allow for correlations (Russell, 2002). While the SIEBAQ assesses potentially unrelated influences on exercise behaviours in adolescents (e.g. parents and social media), a Promax rotation was deemed most appropriate to ensure that any underlying relationships between influences could be captured.

The determinant was above 0.00001 for the SIEBAQ and therefore multicollinearity in the dataset was not a cause for concern. Bartlett’s Test of Sphericity was also significant (p<.001), suggesting that correlations between variables were significantly different from zero and the variables were adequately related to find clusters within the dataset. The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy scored .926, suggesting a ‘marvellous’ sample size (Kaiser & Rice, 1974). KMO scores above .50 are considered sufficient for EFA (e.g. Field, 2018).

Item retention and elimination

Item retention and elimination were conducted in accordance with several criteria. As factor loadings of .40 and above are considered substantial (e.g. Field, 2018), items with factor loadings less than .40 were removed. It is also advised that all factors must have a minimum of three items, all of which are non-cross loading (e.g. Knekta et al., 2019). Items were therefore removed where there were fewer than three items loaded onto a factor, or where cross-loading was evident between two or more factors; an approach considered plausible where there are several adequate to strong loadings (.50 or higher) on each factor (Costello & Osborne, 2005). These item elimination criteria resulted in 41 items being removed from the SIEBAQ, leaving 47 items remaining within the final factor structure.

Analysis of the remaining items

Principal axis factoring was conducted a second time with the remaining 47 items (Promax rotation). Kaiser’s (1960) criteria suggests that factors with an eigenvalue greater than 1 should be retained. Nine factors with an eigenvalue greater than 1 were identified, which explained 63.75% of the total variance (see Table 2). In contrast, the scree plot was ambiguous and showed inflections that would satisfy an eight and ten-factor structure (Cattell, 1966). As scree plot analysis is often subjective, leading to disagreements in the number of factors between researchers (e.g. Norman & Streiner, 2014), to reach a plausible factor solution, it is suggested that the final factor structure is developed from the exploration of a combination of methods (e.g. Izquierdo et al., 2014; Lloret et al., 2017). The eigenvalues and scree plot were therefore analysed in turn and then combined. Based on these analyses, all nine factors were retained and explored. All item loadings were greater than .40 and all items conceptually fitted within the factor that they had been placed. Factor loadings for the nine-factor structure are presented in Table 2.

Table 2. Factor loadings, eigenvalues and percentage of variance each factor explains for the final 47 items of the SIEBAQ (n = 905).

	Factor
Item	1	2	3	4
Eigenvalue	11.62	5.28	3.31	2.46
Variance explained (%)	24.72	11.24	7.04	5.24
Cronbach’s alpha	.91	.92	.83	.84
Factor 1: Peer exercise pressures (9 items)
(11) I worry about exercising in front of my friends	.86
(9) I feel that I am not as good at exercising as my friends	.76
(12) I exercise because I want to have a similar body to my friends	.76
(19) I worry what my body looks like when I exercise with my friends	.75
(14) I avoid exercises that I am not as good at as my friends	.69
(13) I exercise so my friends think my body looks good	.68
(18) When I exercise with my friends, I compare my body to theirs	.66
(3) I feel pressure from my friends to exercise	.54
(5) I compare how much exercise I do with how much exercise my friends do	.54
Factor 2: Traditional media influence (6 items)
(4) When I see images of toned or muscular body types on TV, in films or in magazines, it makes me want to exercise		.88
(1) I exercise because I want to have a similar body to the people I see on the TV shows, in films and in magazines		.85
(7) When I exercise, I compare my body to the body types I see on the TV shows, in films and in magazines		.85
(3) I feel pressure to exercise from what I see on the TV, in films and in magazines		.76
(2) Exercise content that I see on the TV, in films and in magazines makes me do more exercise		.76
(8) When I see others exercising on TV shows, in films and in magazines, it encourages me to exercise		.72
Factor 3: Sibling co-participation (6 items)
(4) I do similar types of exercise to my sibling(s)			.78
(2) When my sibling(s) exercise, I exercise too			.77
(1) Exercise is important to my sibling(s)			.64
(17) Exercising with my sibling(s) is fun			.63
(16) I avoid exercises that I am not as good at as my sibling(s)			.57
(15) I exercise with my sibling(s)			.54
Factor 4: Peer co-participation (4 items)
(16) I exercise with my friends				.93
(4) I do similar types of exercise to my friends				.80
(2) When my friends exercise, I exercise too				.72
(17) Exercising with my friends is fun				.71
	Factor
Item	5	6	7	8	9
Eigenvalue	1.91	1.70	1.33	1.22	1.12
Variance explained (%)	4.07	3.63	2.84	2.61	2.39
Cronbach’s alpha	.72	.81	.81	.84	.84
Factor 5: Parent exercise expectations (6 items)
(8) I exercise because my parent(s) say that I should, even though I would rather not exercise	.63
(3) I feel pressure from my parent(s) to exercise	.57
(9) I feel that I am not as good at exercising as my parent(s)	.55
(7) My parent(s) encourage me to exercise to change my body size and shape	.48
(20) If my parent(s) tease me about my exercise ability, it makes me want to exercise less	.45
(14) I avoid exercises that I am not as good at as my parent(s)	.43
Factor 6: Parent co-participation (4 items)
(16) I exercise with my parent(s)		.84
(17) Exercising with my parent(s) is fun		.70
(2) When my parent(s) exercise, I exercise too		.70
(4) I do similar types of exercise to my parent(s)		.62
Factor 7: Proving exercise ability (5 items)
(23) If my parent(s) tease me about my exercise ability, it makes me want to exercise more			.79
(23) If my sibling(s) tease me about my exercise ability, it makes me want to exercise more			.76
(23) If my friends tease me about my exercise ability, it makes me want to exercise more			.73
(21) If my parent(s) are competitive when they exercise, it makes me competitive			.49
(21) If my friends are competitive when they exercise, it makes me competitive			.45
Factor 8: Sibling body-related pressure (4 items)
(18) When I exercise with my sibling(s), I compare my body to theirs				.81
(13) I exercise so my sibling(s) think my body looks good				.78
(12) I exercise because I want to have a similar body to my sibling(s)				.66
(19) I worry what my body looks like when I exercise with my sibling(s)				.51
Factor 9: Social media modelling (3 items)
(2) Exercise content that I see on social media makes me do more exercise					.82
(3) I use exercise-related content on social media to guide what type of exercise I should do					.79
(1) I regularly view exercise-related content on social media					.72

Note. Numbers in brackets are the original question numbers.

Reviewing and naming the subscales

Each of the nine factors on the SIEBAQ assessed at least one sociocultural influence on exercise behaviours and attitudes (see Table 2). The first factor (9 items; ‘peer exercise pressures’) included items that related to peer pressures and expectations around exercise, including pressures around exercise ability and exercising for body image. The second factor (6 items; ‘traditional media influence’) included items that related to the influence of traditional media, such as feeling pressure to exercise when seeing exercise-related content on traditional media platforms and exercising to model the body types seen in traditional media. The third factor (6 items; ‘sibling co-participation’) included items relevant to sibling influence, such as co-participating in exercise with siblings and awareness of sibling perceptions of exercise. The fourth factor (4 items; ‘peer co-participation’) comprised items related to exercising with peers. The fifth factor (6 items; ‘parent exercise expectations’) contained items that related to feeling pressure from parents about level of ability in exercise. The sixth factor (4 items; ‘parent co-participation’) consisted of items related to exercising with parents. Factor seven (5 items; ‘proving exercise ability’) contained items related to proving level of exercise ability to parents, peers and siblings, and reflected a competitive nature towards exercise. The eighth factor (4 items; ‘sibling body-related pressure’) reflected body-related pressure from siblings, such as body-image comparisons when exercising. The ninth factor (3 items; ‘social media modelling’) represented modelling of exercise behaviours seen on social media platforms.

Internal consistency and validity of the SIEBAQ

Internal consistency for each of the subscales and the global score was calculated. All nine subscales on the SIEBAQ were found to have acceptable (≥.72; range 0.72 to 0.92) internal consistency (e.g. Nunnally & Bernstein, 1994; see Tables 2 and 3). Weak to moderate positive significant correlations were found between all nine factors on the SIEBAQ (r range = .093 − .572; p < .01).

Summary: part 1

The first part of this study aimed to develop and psychometrically test a new, comprehensive measure of sociocultural influences on exercise behaviours in adolescents (SIEBAQ). Following exploratory factor analysis, nine factors were identified: (i) peer exercise pressures; (ii) traditional media influence; (iii) sibling co-participation; (iv) peer co-participation; (v) parent exercise expectations; (vi) parent co-participation; (vii) proving exercise ability; (viii) sibling body-related pressure; and (ix) social media modelling. All factors demonstrated acceptable-to-good levels of internal consistency (Nunnally & Bernstein, 1994; Ursachi et al., 2015). The next step, via Part 2 of this study, sought to establish the convergent validity of the SIEBAQ with related constructs and to explore its predictive validity in relation to compulsive exercise outcomes, eating practices and broader adolescent mental health.

Part 2: convergent validity of the SIEBAQ and exploring sociocultural predictors of compulsive exercise

Method

Participants and procedure

The same participant sample as in Part 1 was used for Part 2 of this study (n = 905). However, participants were required to have completed both the SIEBAQ and the Compulsive Exercise Test (CET) to be included in this part of the study, as these were the key variables of interest in analyses. Fifty nine (6.5%) participants were removed through listwise deletion due to non-completion of the CET and SIEBAQ. The final sample for this study comprised 846 adolescents, aged between 12-15 years (mean age = 13.63; SD = .93; girls: n = 414). In comparison to the original, larger sample, 92% (n = 777) reported having at least one sibling and 93% (n = 787) reported using social media. Preliminary analysis exploring potential differences in demographic characteristics and scores on the included measures between the larger, original participant sample in Part 1 and the sub sample in Part 2 revealed no significant differences. This suggests the sub sample of participants in Part 2 were representative of the original, larger sample.

Measures

The following section provides an overview of the measures used in addition to the SIEBAQ (Supplementary Material), in the order in which they were distributed to participants. Mean scores were calculated for each of the SIEBAQ subscales. Reliability of the measures in the present study is also discussed.

Compulsive exercise test (CET; Taranis et al., 2011)

The CET measures an individual’s compulsivity towards exercise and comprises five subscales: (1) avoidance and rule-driven behaviour; (2) weight control exercise; (3) mood improvement; (4) lack of exercise enjoyment; and (5) exercise rigidity. A CET total score is calculated from the sum of the mean scores of each subscale. The CET contains 24 items, each with a 6-point Likert scale ranging from 0 (never true) to 5 (always true). Greater scores represent greater levels of compulsive exercise. The CET has been validated for use with adolescent samples, with Cronbach’s alpha scores of .88 for boys and .89 for girls (Goodwin et al., 2014). The internal consistency for this sample was .90.

Dutch eating behavior questionnaire – restrained eating subscale (DEBQ; van Strien et al., 1986)

The DEBQ measures three eating styles: (1) emotional eating; (2) external eating; and (3) restrained eating. Each item is measured using a 5-point Likert scale ranging from 1 (never) to 5 (very often). As restrained eating is typically the component of disordered eating most strongly associated with compulsive exercise (e.g. Dalle Grave et al., 2008), only the 10-item restrained eating subscale of the DEBQ was used for the present study. An overall restrained eating score is calculated with the mean score of the ten items. The DEBQ has shown good levels of reliability with adolescent samples, with Cronbach’s alpha scores of .88 for the restrained eating subscale (e.g. Hunot-Alexander et al., 2019). Internal consistency for the current sample was .96.

Sociocultural attitudes towards appearance questionnaire – 4 (SATAQ-4; Schaefer et al., 2015)

The SATAQ-4 is a 22-item measure which assesses sociocultural influences and pressures related to appearance and body image. The five subscales of the SATAQ-4 measure both thin and athletic body-ideal internalisation and perceived body-related pressure from family, peers and traditional media sources. The SATAQ-4 has been validated for use with adolescents (Amiri & Navab, 2018; Yamamiya et al., 2016), and internal consistency has been found to be excellent for each of the three ‘sociocultural influence’ subscales: family pressures (α = .84); peer pressures (α = .88); and media pressures (α = .95; Palmeroni et al., 2021). Similar internal consistency was found in the current sample (α ≥ .81).

Short Warwick-Edinburgh mental wellbeing scale (SWEMWBS; Stewart-Brown et al., 2009)

The SWEMWBS is a short, 7-item version of the WEMWBS (14 items). The WEMWBS measures mental wellbeing with positively worded statements (e.g. “I’ve been thinking clearly”), and participants are required to rate their experiences over the last two weeks on a 5-point Likert scale ranging from 1 (none of the time) to 5 (all of the time). Total scores range from 7 to 35 and raw total scores are converted into metric scores (Stewart-Brown et al., 2009). Higher scores indicate higher levels of mental wellbeing. The SWEMWBS has shown good internal consistency with adolescent populations (α = 0.78; McKay & Andretta, 2017). The internal consistency for the current sample was good (α = 0.89).

International physical activity questionnaire – short form (IPAQ-SF; Craig et al., 2003)

The IPAQ-SF was used to determine participant levels of physical activity engagement. The IPAQ-SF consists of seven questions that assess daily time spent sitting, walking and engaging in moderate and vigorous physical activity over the last seven days. For the purposes of this study, only the questions related to walking, moderate and vigorous physical activity behaviours were used. Total minutes reported by participants are converted into Metabolic Equivalent Task minutes per week (MET-min/week). The MET assignments for each level of physical activity intensity are walking (3.3 METs), moderate (4 METs), and vigorous (8 METs). The official scoring system for the IPAQ-SF (Craig et al., 2003) then classifies participants into one of three categories according to the cut-off of total MET scores: low, moderate and high. Internal consistency for the current sample was .90.

Data analysis

All data were analysed using IBM SPSS Statistics 27. Data were initially screened for normality. Shapiro Wilk tests identified a non-normal distribution for all study variables. Non-parametric tests were therefore used where possible. A Kruskal-Wallis test was also conducted to explore gender differences in the study variables prior to analyses. The test identified significant gender differences between scores on some of the study variables (see Table 3). Pairwise comparisons revealed that the gender differences were significant between boys and girls.

Table 3. Means, standard deviations and tests of difference for study variables by gender.

	Mean (SD)			Test of difference H
	Boys (n = 402)	Girls (n = 414)	Prefer not to say (n = 30)
Compulsive Exercise Test	7.53 (4.45)	8.34 (4.92)	6.66 (3.89)	6.48
Avoidance & rule-driven behaviour	.98 (.96)	1.06 (1.11)	.90 (.99)	.54
Weight control exercise	1.91 (.97)	2.23 (1.30)	1.81 (1.21)	11.40**
Mood improvement	1.85 (1.25)	1.94 (1.28)	1.32 (1.02)	7.14
Lack of exercise enjoyment	1.36 (.96)	1.69 (1.14)	2.41 (1.53)	25.43**
Exercise rigidity	1.61 (1.30)	1.74 (1.34)	1.20 (1.11)	5.40
DEBQ restrained eating	1.75 (.86)	2.24 (1.16)	2.36 (1.38)	34.19**
SWEMWBS	20.04 (5.59)	18.75 (5.09)	16.03 (7.56)	25.59**
IPAQ-SF (MET-min/week)	2250.46 (2903.80)	1824.16 (2312.10)	1707.30 (3204.12)	3.30
Low: n (%)	125 (31.1%)	160 (38.6%)	13 (41.9%)
Moderate: n (%)	111 (27.6%)	103 (24.9%)	6 (19.4%)
High: n (%)	101 (25.1%)	91 (22.0%)	4 (12.9%)
SATAQ-4 total	1.88 (.79)	2.15 (.99)	1.87 (.94)	11.76*
Thin ideal internalisation	1.81 (.87)	2.00 (1.02)	1.88 (1.03)	5.25
Athletic ideal internalisation	1.60 (.79)	1.96 (1.11)	1.82 (1.04)	16.63**
Family pressure	1.92 (.93)	2.12 (1.11)	1.81 (.99)	5.48
Peer pressure	1.99 (.88)	2.45 (1.17)	1.95 (1.07)	28.55**
Media pressure	2.09 (1.07)	2.20 (1.09)	1.86 (1.08)	4.98
SIEBAQ
Peer exercise pressures	1.58 (.69)	1.70 (.88)	1.68 (.86)	.28
Traditional media	1.50 (.73)	1.85 (1.01)	1.70 (.94)	33.43**
Sibling co-participation	2.15 (.92)	2.36 (.95)	2.27 (.98)	12.75*
Peer co-participation	2.76 (1.11)	2.66 (1.05)	2.20 (1.03)	9.40
Parent exercise expectations	1.65 (.61)	1.68 (.73)	1.83 (.94)	1.04
Parent co-participation	2.26 (.97)	2.44 (1.01)	1.91 (.96)	12.16
Proving exercise ability	1.95 (1.11)	1.90 (.95)	1.55 (.85)	4.54
Sibling body-related pressure	1.34 (.64)	1.35 (.72)	1.54 (.85)	2.14
Social media modelling	2.14 (1.02)	2.42 (1.07)	1.73 (1.03)	22.54**

Note. DEBQ = Dutch Eating Behavior Questionnaire; IPAQ-SF – International Physical Activity Questionnaire Short Form; SATAQ-4 = Sociocultural Attitudes Towards Appearance Questionnaire – 4; SIEBAQ = Sociocultural Influences on Exercise Behaviours in Adolescents Questionnaire; SWEMWBS = Short Warwick-Edinburgh Mental Well-Being Scale.

*p < .01; **p < .001; Kruskal Wallis H test based on gender differences between boys, girls and ‘prefer not to say’.

Thirty participants in the Part 2 participant sample (n = 846) reported ‘prefer not to say’ for their gender; this sample size was deemed too small for separate inclusion in inferential testing. This group could also not be accurately described in relation to their gender identity. As a result, adolescents who reported their gender as ‘prefer not to say’ were excluded from inferential analyses where gender differences were explored.

Spearman’s rho two-tailed correlations were initially conducted separately for boys and girls to explore the convergent validity of the SIEBAQ with related constructs (e.g. sociocultural attitudes towards appearance, compulsive exercise). To test the study’s hypothesis, Spearman’s rho two-tailed correlations were then used to explore whether significant relationships exist between compulsive exercise outcomes and related constructs, such as disordered eating, mental wellbeing, physical activity engagement and sociocultural influences on appearance. Where significant associations were identified, these were controlled for in the first step of the regression analyses. After accounting for any significant covariates, hierarchical stepwise regressions were conducted separately by gender to identify which sociocultural influences predicted compulsive exercise in boys and girls. Sociocultural influence subscales in the SIEBAQ that were identified as being significantly correlated with each compulsive exercise outcome were then added into the next step of the regression. Pairwise deletion of cases was implemented for all analyses. To reduce the risk of a Type 1 error, a p value of p < .01 was employed.

Results

Descriptive statistics

The descriptive statistics and tests of difference between gender for participants in Part 2 (n = 846) are presented in Table 3. CET and DEBQ scores for the current sample were generally slightly lower than existing research conducted with non-clinical adolescent samples (e.g. Goodwin et al., 2014; Hunot-Alexander et al., 2019), and SWEMWBS scores were slightly lower in comparison to existing research with UK adolescents (e.g. McKay & Andretta, 2017). Physical activity levels in the current sample were broadly in line with existing research with adolescents (e.g. Aktürk et al., 2019; Fernández-Bustos et al., 2019) and identified that approximately 49% of the total sample met criteria for moderate to high levels of physical activity.

Tests of difference (Table 3) also showed that girls scored significantly higher than boys on CET weight control exercise, DEBQ restrained eating, the SATAQ total score, SATAQ athletic ideal internalisation and SATAQ peer pressure, SIEBAQ traditional media, SIEBAQ sibling co-participation and SIEBAQ social media modelling.

Convergent validity

Two-tailed Spearman’s correlations were conducted separately for boys and girls to explore the convergent validity between the SIEBAQ and other related constructs (i.e. IPAQ-SF, SATAQ-4; see Supplementary Tables 1 and 2). For boys, all SIEBAQ subscales significantly, positively correlated with all subscales of the SATAQ-4, with weak to moderate significant correlations, highlighting convergent validity of the SIEBAQ with a measure of a related construct (r ≥ .178, p < .01). Correlations between the SIEBAQ and physical activity using the IPAQ-SF only identified weak to moderate significant, positive correlations with peer co-participation and proving exercise ability (r ≥ .160, p < .01; see Supplementary Table 1). For girls, most SIEBAQ subscales significantly, positively correlated with the subscales of the SATAQ-4 (r ≥ .144, p < .01), aside from SIEBAQ parent co-participation (which did not correlate), supporting convergent validity of the SIEBAQ with a measure of a related construct (see Supplementary Table 2). Correlations between the SIEBAQ and physical activity using the IPAQ-SF identified weak to moderate significant, positive correlations with traditional media influence, social media modelling, parent and peer co-participation, and proving exercise ability (r ≥ .136, p < .01).

Correlation analyses

Two-tailed Spearman’s correlations were conducted separately for boys and girls to explore relationships between compulsive exercise outcomes and all related constructs (e.g. disordered eating, mental wellbeing, sociocultural influences on exercise and attitudes towards appearance) to inform regression analyses.

Boys

Most associations identified between the SIEBAQ and CET outcomes were small to moderate (r ≥ .143, p < .01). Almost all SIEBAQ subscales positively, significantly correlated with CET outcomes, aside from SIEBAQ parent exercise expectations where the positive correlations with CET mood improvement was non-significant. Correlations were also non-significant between CET lack of exercise enjoyment and SIEBAQ peer exercise pressure, traditional media influence, parent co-participation and sibling body-related pressure. For CET lack of exercise enjoyment, significant negative correlations were identified with SIEBAQ sibling and peer co-participation, proving exercise ability and social media modelling. In addition, DEBQ restrained eating and all SATAQ variables, significantly positively correlated with all CET outcomes (r ≥ .329, p < .001), aside from CET lack of exercise enjoyment. Mental wellbeing and physical activity levels all significantly correlated with all CET outcomes (r ≥ .-0.242, p < .01), aside from CET avoidance and rule-driven behaviour and weight control exercise (see Supplementary Table 1).

Girls

Significant positive and negative correlations were found between most of the SIEBAQ sociocultural influences and CET outcomes, and correlations were generally small to moderate (r ≥ .143, p < .01). For CET lack of exercise enjoyment, significant negative correlations were found with SIEBAQ sibling, peer and parent co-participation, and social media modelling (see Supplementary Table 2). In addition, DEBQ restrained eating and all SATAQ variables, aside from SATAQ peer and media pressure, significantly positively correlated with all CET outcomes (r ≥ .154, p < .01). Physical activity levels significantly correlated with all CET outcomes aside from weight control exercise, where positive correlations were non-significant. Mental wellbeing significantly positively correlated with CET mood improvement and exercise rigidity, and significantly negatively correlated with CET lack of exercise enjoyment.

Regression analysis

Hierarchical regression analyses were then conducted separately by gender to determine which sociocultural influences predicted compulsive exercise outcomes in boys and girls. Covariates (i.e. restrained eating, mental wellbeing, sociocultural attitudes towards appearance variables, and physical activity levels) which were significantly correlated with the compulsive exercise outcomes were controlled for and entered into the first step of the regression. Sociocultural influences which were significantly correlated with each compulsive exercise outcome were entered into the second step, using a stepwise regression (see Tables 4 and 5).

Table 4. Hierarchical stepwise regression analysis predicting CET scores from covariates and sociocultural influences for boys (n = 402).

Predictor	CET avoidance			CET weight control			CET mood improvement			CET lack of enjoyment			CET exercise rigidity
	β	Adj R^2	F change	β	Adj R^2	F change	β	Adj R^2	F change	β	Adj R^2	F change	β	Adj R^2	F change
Step 1		.40	30.79**		.35	31.30**		.35	22.19**		.06	11.89**		.37	24.70**
IPAQ	.21**
DEBQ restrained eating	.20**			.28**
SWEMWBS							.18**			−0.22**				.16**
IPAQ-SF							.17**							.21**
SATAQ thin ideal				.27*			.24*
SATAQ athletic ideal							−0.27*
SATAQ family pressure	.23*													.26*
SATAQ media pressure							.30**							.35*
Step 2		.50	11.53**		.36	8.06*		.51	8.79*		.20	7.13*		.46	11.04**
IPAQ	.16**
DEBQ restrained eating	.20**			.27**
SWEMWBS										−0.19**				.14*
IPAQ-SF														.16**
SATAQ thin ideal	.27*
SATAQ media pressure														.22*
Social media modelling	.30**			.16*			.22**			−0.18*				.27**
Proving exercise ability	.17**						.15*			−0.16*				.17**
Peer co-participation							.24**
Parent co-participation							.13*
Parent exercise expectations										.31**

Note. DEBQ = Dutch Eating Behaviour Questionnaire; IPAQ-SF = International Physical Activity Questionnaire Short Form; SATAQ = Sociocultural Attitudes Towards Appearance; SWEMWBS = Short Warwick-Edinburgh Mental Well-Being Scale.

SIEBAQ Sociocultural influences on exercise behaviours italicised. Only significant predictors at p < 0.01 reported.

*p < .01; **p < .001.

Boys (Table 4)

The regression models to assess the predictive roles of covariates and sociocultural influences on compulsive exercise outcomes were significant for all five CET subscales (Table 4). Significantly correlated covariates differed for all five CET subscales (e.g. restrained eating, mental wellbeing, physical activity levels and sociocultural attitudes towards appearance; see Supplementary Table 1) and initially explained 40% (Avoidance and Rule-Driven Behaviour), 35% (Weight Control Exercise), 35% (Mood Improvement), 6% (Lack of Exercise Enjoyment) and 37% (Exercise Rigidity) of the total variance in the first step. Sociocultural influence variables explained an additional 1-14% of the variance in Step 2 of the regression models. Social Media Modelling made the largest contribution for most CET variables (between 16 and 30% of the variance explained) aside from CET Mood Improvement and CET Lack of Exercise Enjoyment. For CET Mood Improvement, Peer Co-participation made the largest contribution followed by Social Media Modelling. For CET Lack of Exercise Enjoyment, Parent Exercise Expectations made the largest contribution followed my Social Media Modelling. For CET Avoidance and Rule-Driven Behaviour and CET Exercise Rigidity, Proving Exercise Ability made the second largest contribution (17% of the variance explained).

Girls (Table 5)

The regression models were significant for all CET subscales (Table 5). Significantly correlated covariates differed for all five CET subscales (e.g. restrained eating, mental wellbeing, physical activity levels and sociocultural attitudes towards appearance; see Supplementary Table 2) and initially explained 39% (Avoidance and Rule-Driven Behaviour), 59% (Weight Control Exercise), 37% (Mood Improvement), 22% (Lack of Exercise Enjoyment) and 27% (Exercise Rigidity) of the total variance in the first step. Sociocultural influence variables explained an additional 4-12% of the variance in Step 2 of the regression models. Social Media Modelling made the largest contribution for CET Avoidance and Rule Driven Behaviour (23% of the variance explained) and CET Exercise Rigidity (38%), followed by Proving Exercise Ability (15 and 19%). Social Media Modelling also made the largest contribution for CET Mood Improvement (30%), followed by Parent Co-Participation (15%). For CET Weight Control Exercise, Traditional Media made the largest contribution (19%), followed by Social Media Modelling (14%). For CET Lack of Exercise Enjoyment, Parent Exercise Expectations made the largest contribution (31% of the variance explained), followed by Peer Exercise Pressures (22%).

Table 5. Hierarchical stepwise regression analysis predicting CET scores from covariates and sociocultural influences for girls (n = 414).

Predictor	CET avoidance			CET weight control			CET mood improvement			CET lack of enjoyment			CET exercise rigidity
	β	Adj R^2	F change	β	Adj R^2	F change	β	Adj R^2	F change	β	Adj R^2	F change	β	Adj R^2	F change
Step 1		.39	31.82**		.59	82.48**		.37	25.59**		.22	16.51**		.27	16.17**
DEBQ restrained eating	.21*			.32**			.18*
SWEMWBS							.32**			−0.21**			.25**
IPAQ-SF	.17**						.30**			−0.27**			.24**
SATAQ family pressure							.32**
SATAQ peer pressure				.32**
Step 2		.48	7.72*		.63	11.08**		.49	8.86*		.34	10.17*		.37	13.74**
DEBQ restrained eating	.21**			.26**
SWEMWBS							.19**			−0.15*			.15*
IPAQ-SF	.11*						.23**			−0.18**			.19**
SATAQ peer pressure				.32**
Social media modelling	.23**			.14**			.30**						.26**
Parent co-participation							.15**			−0.19**
Proving exercise ability	.19**						.14*						.15*
Traditional media				.19**
Parent exercise expectations	.13*									.31**
Peer exercise pressures										.22*

Note. DEBQ = Dutch Eating Behaviour Questionnaire; IPAQ-SF = International Physical Activity Questionnaire Short Form; SATAQ = Sociocultural Attitudes Towards Appearance; SWEMWBS = Short Warwick-Edinburgh Mental Well-Being Scale.

SIEBAQ Sociocultural influences on exercise behaviours italicised. Only significant predictors at p < 0.01 reported.

*p < .01; **p < .001.

Discussion

The first aim of this study was to develop and explore the psychometric properties of a measure designed to assess a variety of sociocultural influences that are theoretically associated with exercise psychopathology in adolescents (the SIEBAQ). Results from the initial factor analysis identified nine subscales that capture five key sociocultural influences on exercise behaviours in adolescents: parents, siblings, peers, social media and traditional media. The second aim of this study was to explore the convergent validity of the SIEBAQ and to use the SIEBAQ to explore sociocultural predictors of compulsive exercise in adolescents. There was evidence for convergent validity of the SIEBAQ, which was positively related, as expected, to measures of exercise psychopathology and sociocultural attitudes towards appearance. Cross-sectional analyses with the SIEBAQ identified significant differences in sociocultural influences on compulsive exercise outcomes between adolescent boys and girls. As previous research has only identified the significance of parents, peers and traditional media in relation to compulsive exercise in adolescents (e.g. Goodwin et al., 2014; Reynolds et al., 2022), this study has uniquely identified the added importance of sibling and social media influence in further understanding sociocultural influences on exercise psychopathology in adolescent boys and girls.

Modelling of exercise behaviours seen on social media was evidenced as the strongest predictor of compulsive exercise outcomes in both boys and girls. As the relationship between social media influence and compulsive exercise has only been researched in adult populations to date (e.g. Raggatt et al., 2018; Reynolds et al., 2022), our novel findings contribute to the need to understand social media influence on maladaptive health behaviours prior to adulthood. As body ideal internalisation also significantly correlated with compulsive exercise outcomes in our adolescent sample, and adolescents in qualitative research have reported exercising to model the body types they see on social media (Reynolds et al., 2022), it is important for future research to explore the potential contribution of body ideal internalisation in the relationship between modelling of exercise behaviours on social media and compulsive exercise in adolescents. Our findings also add further evidence to support the inclusion of social media in addition to traditional media within the tripartite influence model (e.g. Roberts et al., 2022).

Co-participating in exercise with parents and peers was associated with mood improvement in adolescents. However, proving exercise ability to significant others (e.g. parents/peers) was a strong statistical predictor of compulsive exercise outcomes in both boys and girls. As co-participating in exercise activities with peers elicits greater peer approval in adolescent boys (e.g. Lawler et al., 2021), and both adolescent boys and girls in qualitative research have reported a perceived pressure to meet a socially constructed ‘ability threshold’ to co-participate in exercise activities with others (Reynolds et al., 2022), the novel contribution of the present study could suggest that the perceived need to meet this ‘threshold’ can predict maladaptive exercise outcomes at this age. It is therefore important for exercise-related interventions that promote co-participation of exercise activities with parents and peers to note the potential unintended implications that such sociocultural influences could have on the promotion of maladaptive exercise behaviours and attitudes in adolescents.

Adolescents also co-participate in exercise behaviours with siblings (e.g. Lisinskiene & Juskeliene, 2019; Reynolds et al., 2022), hence sibling influence was also included in the development of the SIEBAQ. The factor analysis pointed towards some novel influences and potential mechanisms of sibling influence on exercise attitudes and behaviours in adolescents, such as body-related pressure from siblings. Correlation analyses conducted in Part 2 of this study with the SIEBAQ identified significant, positive relationships between compulsive exercise outcomes and SIEBAQ subscales inclusive of sibling influence (i.e. sibling co-participation, body-related pressure and proving exercise ability), for both boys and girls. Further research is necessary to understand the key mechanisms and circumstances through which siblings can influence compulsive exercise outcomes in adolescents. For example, as older and same-gender siblings are more likely to serve as role models (e.g. Whiteman et al., 2011), and girls are more likely to engage in social comparisons with their older siblings (e.g. Campione-Barr & Killoren, 2019), it would be of interest for future research to explore the extent to which factors such as sibling age, gender, exercise attitudes/behaviours and body image might influence maladaptive exercise attitudes and behaviours. The SIEBAQ offers a suitable tool by which to achieve this, by assessing sibling influences on exercise behaviours and supporting research to further understand the extent of sibling influence on exercise psychopathology throughout adolescent development, and understand how siblings can support the prevention of compulsive exercise through intervention strategies.

Significant, positive relationships were identified between mental wellbeing and most compulsive exercise outcomes, which is contrary to existing literature that has identified mental wellbeing to be negatively impacted by compulsive exercise (e.g. Lichtenstein et al., 2017). However, as research with adolescents has found a significant, positive relationship between both functional and dysfunctional emotion regulation behaviours and compulsive exercise (Goodwin et al., 2012), it could be suggested that compulsive exercise is maintained through both positive and negative reinforcement mechanisms, such as post-exercise mood improvement or exercising to avoid negative repercussions even if exercise is not enjoyable (e.g. weight-related comments from others). Further work is needed to explore the seemingly complex relationship between compulsive exercise and mental wellbeing in adolescents.

While significant correlations between the SIEBAQ and physical activity engagement were limited, the SIEBAQ subscales which did significantly correlate with physical activity engagement were reflective of exercise co-participation and proving level of exercise ability to others; areas where higher levels of exercise behaviours would be expected. Given that exercise attitudes are more closely linked with maladaptive health behaviours (e.g. disordered eating) than exercise frequency (e.g. Taranis et al., 2011), this can support the overall limited correlations between the SIEBAQ and physical activity levels and could suggest that sociocultural influences can potentially affect exercise attitudes more than exercise frequency in adolescents. Using the SIEBAQ in future research will therefore support researchers to understand the way in which sociocultural influences can support exercise engagement, and to understand perceived sociocultural barriers associated with exercise activity.

This study has successfully developed a new measure that reliably assesses the five key sources of sociocultural influence theoretically associated with exercise behaviours in adolescents and can reliably assess the relationship between sociocultural influences and exercise psychopathology in adolescents. This study is the first to identify social media modelling for exercise as a significant predictor of compulsive exercise in adolescents, and the first to find significant relationships between sibling influence and compulsive exercise outcomes, addressing the key gaps in existing literature. Individual SIEBAQ subscales can be selected by researchers based on the sociocultural influences that are relevant to their research aims. A further strength of this study is the large sample size of adolescent girls and boys recruited from diverse ethnic groups, as sociocultural influences can differ between males and females (Field et al., 2001) and exercise activity levels can vary among ethnic groups (e.g. Eyre & Duncan, 2013).

While there are many strengths, this study is not without its limitations. Further research is required to determine whether item loadings on the SIEBAQ can be replicated in other samples and across other cultures. To do this, a Confirmatory Factor Analysis (CFA) is required. Conventional CFA goodness of fit assessment was not conducted during the development of the SIEBAQ and this is an important avenue for further psychometric testing in order to confirm the nine-factor structure identified through EFA. CFA criteria, however, are often restrictive, particularly where there are multiple factors (5-10), each with a reasonable number of items (e.g. 5-10 per scale minimum). It is acknowledged that achieving acceptable fit indices will be challenging in these instances (Marsh, 2007). While there were significant differences in sociocultural influences on compulsive exercise behaviours in adolescent boys and girls, a small sample of the participants in Part 2 selected their gender as ‘prefer not to say’. This sample of participants, however, was deemed too small for inferential testing and therefore excluded from analyses where gender differences were explored. It is important to acknowledge the potential inequity in understanding sociocultural influences on exercise behaviours in identities beyond the traditional binary gender identities. Future research could therefore explore potential differences in sociocultural influences on exercise behaviours and attitudes in gender identities beyond males and females.

Conclusion

This research has successfully developed and psychometrically tested a measure of sociocultural influences on exercise behaviours in adolescents, which encompasses a multitude of sociocultural constructs that are theoretically associated with exercise behaviours in adolescent populations (i.e. parents, siblings, peers, social media and traditional media). As previous research has assessed the relationship between sociocultural influences and exercise psychopathology using questionnaires designed to measure parent, peer and traditional media influences on appearance and weight-related concerns, this new measure can be used to more reliably assess a wider range of sociocultural influences on unhealthy exercise behaviours and attitudes in adolescents, such as compulsive exercise. In addition, development of the SIEBAQ and preliminary cross-sectional analyses has identified the importance of sibling and social media influence in compulsive exercise behaviours in adolescents; influences under-researched in adolescent populations. Further psychometric testing of the measure is now required to confirm the factor structure identified through EFA, and it would be of interest for future research to explore the predictive validity of the measure against objective measures of physical activity.

CRediT author statement

Kalli A Reynolds: Conceptualisation, Formal Analysis, Investigation, Writing—Original Draft Emma Haycraft: Conceptualisation, Writing—Review & Editing, Supervision Carolyn R Plateau: Conceptualisation, Writing—Review & Editing, Supervision

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data available on request from the authors. The data that support the findings of this study are available from the corresponding author (CRP) upon reasonable request.

Additional information

Funding

Kalli A. Reynolds is funded by a PhD studentship awarded by the School of Sport, Exercise and Health Sciences at Loughborough University, UK.