Bristol Girls Dance Project: a cluster randomised controlled trial of an after-school dance programme to increase physical activity among 11- to 12-year-old girls

Article history

The research reported in this issue of the journal was funded by the PHR programme as project number 11/3050/01. The contractual start date was in April 2013. The final report began editorial review in July 2015 and was accepted for publication in December 2015. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PHR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Professor Jago has been a member of the Research Funding Board for the National Institute for Health Research (NIHR) Public Health Research (PHR) board since October 2014. Professor Powell was a member of the NIHR PHR Funding Board from June 2011 to September 2015.

Copyright statement

© Queen’s Printer and Controller of HMSO 2016. This work was produced by Jago et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Benefits of physical activity

Regular physical activity (PA) can improve a number of risk factors for chronic disease, including body composition, lipoprotein profiles, insulin sensitivity, glucose levels and blood pressure in children and adults. 1,2 PA is also associated with positive emotional well-being and self-esteem among young people. 3 Despite these benefits, a large number of young people in the UK do not meet the PA recommendations of 60 minutes on most days of the week. 2 The number of young people meeting this guideline fell between 2008 and 2012. 4 There is an age-related decline in PA levels during childhood and adolescence, with the start of secondary school being a key period of downwards transition. 5,6

Girls’ physical activity

The percentage of girls who meet government recommendations of 60 minutes of PA per day is low. 4,7,8 Throughout childhood and adolescence, girls spend less time in moderate- to vigorous-intensity physical activity (MVPA) than boys, with reports of > 60% of girls’ time being spent on sedentary behaviours. 7 Therefore, there is a need for effective strategies to attenuate this decline in PA among girls and to reduce the amount of time spent being sedentary.

A number of reasons have been proposed as possible explanations for the age-related decline in girls’ PA. Research conducted by the Women’s Sport and Fitness Foundation9 found that girls’ attitudes towards PA are positive and that 76% of 15-year-old girls want to do more PA. Girls perceived more barriers to being active than boys, ranging from disinterest and boredom to a lack of opportunity for the desired sport in school. 10 Self-presentation issues also affect participation, with 76% of girls agreeing that they are conscious of their body image. 9

Sport competence has emerged as a common theme relating to the PA levels of girls. Sports competence declines with age in boys and girls, but is consistently lower for girls and falls faster between 17 and 18 years of age. 11 Physical education (PE) in school can expose anxieties and make girls feel embarrassed about their perceived lack of skill, further deterring them from participation. Focus groups based on girls’ attitudes towards PA have reported that girls are often disinclined towards PA because they do not like to appear tired or ‘sweaty’ in front of peers, especially males. 10,12 Girls also worry about not looking ‘cool’ and being teased by boys and girls about not being feminine,9,10 leading to an increased likelihood of giving up sport. During adolescence, girls begin to form their own self-perception of sporting ability, which contributes to the decision-making process about whether or not to participate. 11 Therefore, the factors that contribute to the decline in PA must be handled delicately and be considered when designing interventions to increase girls’ PA.

Interventions to increase girls’ physical activity

A recent meta-analysis concluded that interventions designed to increase girls’ PA levels have a minimal, yet significant, effect (g = 0.314; p < 0.001). 13 On average, girls in intervention groups accrued 12.17% more PA than girls in control groups. The authors suggested that the results led to the belief that behaviour change in this population is challenging but possible. A subgroup analysis found that interventions developed specifically for girls and interventions with multiple components were more likely to produce a significant treatment effect. For example, an intervention that worked with Girl Scout troops to foster healthy behaviour resulted in girls in intervention troops accumulating significantly more MVPA than girls in control troops, with 7.4% of girls meeting MVPA daily targets versus 1.6%, respectively (χ² = 18.4; p < 0.001). 14

A number of approaches have attempted to help girls overcome their anxieties around PA. Intervention approaches have included allowing schools to develop bespoke action plans to prevent the decline in girls’ MVPA. 7 The Nutrition and Enjoyable Activity for Teen Girls intervention15 provided enhanced school sports sessions for 12- to 14-year-old girls, held seminars and workshops, and used various tools to help self-monitor activity. The choice of music and appealing activities during the sports sessions were incorporated to engage girls. The results suggest that these elements may have contributed to the favourability of sessions (41.7% of participants reporting it as the most enjoyable component of the intervention). However, it was found that MVPA did not significantly differ between the intervention and control groups [adjusted difference in change –4.28 minutes/day, 95% confidence interval (CI) –13.82 to 5.25 minutes/day; p > 0.05]. 15 It is important to note that the authors did not reflect on elements of the intervention that were successful or unsuccessful.

Taymoori and Lubans16 used theoretical constructs to aid the development of educational and counselling sessions for girls in order to identify potential mediators of PA behaviour change. Both interventions showed an intervention effect on self-reported PA. However, PA was not measured objectively. The authors concluded that certain behavioural strategies, such as goal setting and activity monitoring, were mediators for behaviour change. This suggests that interventions that focus on these components could be effective in increasing PA. Few studies have focused on increasing girls’ PA as well as on attempting to overcome barriers such as low levels of competence and providing choice. Therefore, there is need for an intervention to be tailored to address these issues.

In summary, research suggests that adolescent girls are less active and experience greater declines in PA than boys because they have disengaged from and/or lack confidence in the types of PA that are traditionally offered. Increasing girls’ perceived competence, strength of identity and interest in activities may help to sustain girls’ participation in PA throughout adolescence.

Dance as a method to improve girls’ physical activity levels

Dance is popular among UK and US adolescent girls. 12,17,18 Dance also provides an opportunity to socialise. 19 The proportion of girls who dance remains high throughout adolescence, with 31.9% engaging at 12–13 years of age and 33.1% at 18 to 19 years of age. 20 This interest in dance indicates that it could be a useful medium through which to help improve girls’ participation in PA.

In order to promote dance as a medium for increasing girls’ PA, it is important to understand what attracts girls to dance. It appears that the style of dance is important. In particular, modern, high-energy dances are seen as more fun. 19 Girls particularly enjoy the opportunity to input into the dance moves that they learn and perform. 19 The type of music (contemporary and upbeat) is a key factor that appeals to girls. 19 Opportunities to socialise and support from peers contribute to whether or not girls participate. 19,21 Dance provides an opportunity for all of the above and can be used as a tool to help overcome barriers to girls’ participation in PA. Dance is usually group-based (and is thus less likely to lead to individuals being on public display), non-competitive and takes place indoors where weather is not particularly important. 10,12,22

Theories of behaviour change: self-determination theory in physical activity research

Interventions based on psychological theory have proved more successful than those not based on psychological theories. 43,44 SDT45 is particularly illuminating in examining adolescents’ motivation for PA46 (e.g. how self-determined their reasons for PA are). The underlying principles of SDT are that self-determined (autonomous) motivation is behaviourally and psychologically adaptive and that it develops to the extent that individuals perceive a sense of ownership or autonomy over their PA, feel able and competent and are supported in their PA by meaningful social relationships. 45,47 This hypothesis is supported by empirical research in the PA/exercise domain among children,48 adolescents49 and dancers. 50

Six motivation types are proposed in SDT, varying in their degree of self-determination or autonomy. Each is hypothesised to be differently associated with individuals’ engagement in a given behaviour (e.g. being active), cognitive processes and affective outcomes. 51 The more self-determined motivational types (intrinsic, integrated and identified behaviours) are grouped as autonomous. Intrinsic motivation sees an individual engage in an activity for inherent satisfaction or enjoyment. Other forms of motivation involve doing the behaviour for a reason(s) distinctive from inherent satisfaction. Integrated regulation is when engaging in a behaviour is in line with an individual’s broader self (e.g. being active as part of a person’s identity). Identified motivation is driven by an outcome, such as health benefits or friendship. The less self-determined motivational forms (introjected and external) are broadly grouped as controlled motivation. Introjected regulation refers to motivation based around internalised pressures, like avoiding feelings of guilt. External regulation is characterised by prods and pushes that are external to the individual, such as avoiding punishment.

Previous research suggests that autonomous motivation for PA is positively associated with an increase in young people’s PA behaviour. 48,52 In addition, autonomous motivation is seen to positively affect psychological outcomes such as quality of life and physical self-concept. 53 Girls with autonomous motivation for exercise have been found to exhibit a greater sense of competence and achievement, and tend to remain involved in PA, whereas those without this form of motivation lose interest and look elsewhere for investment of their efforts. 54 Evidence suggests that social aspects motivate girls to engage in PA, with > 50% of girls being active because of peers. 9 Peer support is an important factor in the participation and retention of after-school dance programmes. 19 It has been shown that children cluster in groups with similar activity levels,55 suggesting that friendship is important for PA participation. As such, an intervention that addresses these factors may be successful in engaging girls in PA and encouraging lifelong attitudes towards being active.

Self-determination theory is particularly amenable to inclusion in behavioural interventions because it specifies the psychological and socioenvironmental conditions that underpin autonomous forms of motivation. 56 In SDT, it is important to note that for interventions delivered by practitioners (e.g. dance instructors), people’s psychological needs can be supported or undermined by the motivational climate that an authority figure creates through their interpersonal or teaching style. 57,58 Needs-supportive motivating styles are underpinned by the provision of autonomy support (i.e. providing a meaningful rationale, offering choice and avoiding the use of controlling language), structure (i.e. setting out clear expectations and providing guidance) and involvement (i.e. being empathic and showing genuine interest for others). 58,59 A controlling motivating style is characterised by strategies such as tangible rewards, feedback aimed to manipulate rather than to inform and the use of controlling language. 60 Among children, PE teachers’ use of autonomy-supportive versus controlling motivation styles has been associated with pupils’ psychological need satisfaction and autonomous motivation for PE. 49

Formative work

Extensive formative work was conducted to inform the design of the current study. This included: (1) qualitative work with Year 7 girls and their parents on how to design an after-school dance programme and recruit children;19 (2) interviews with teachers and dance specialists to identify key elements to incorporate into the intervention;61 and (3) a feasibility trial38 and economic evaluation. 62

Focus groups (including 65 girls) and telephone interviews (with 16 parents) were conducted in four secondary schools across Bristol. 19 Some issues were suggested to affect recruitment. The intervention needed to be marketed as fun and enjoyable (as well as something that provides an opportunity to socialise). To encourage participation, taster sessions were considered useful. Parents suggested that attracting groups of friends and stressing the health benefits of dance may also increase interest. Views on sustaining participation differed for girls and parents. Girls highlighted the importance of dance styles and music (and being allowed to input on both) for keeping them motivated, and parents suggested that setting achievable goals throughout the programme would help retention. Parents also cited enjoyment and teaching style as important.

Interviews were conducted with 11 PE teachers from secondary schools in the local area and 11 dance instructors. 61 PE teacher interviews covered logistical issues of a dance intervention within a school setting and structure, content and recruitment concerns for an after-school dance programme. Dance instructor interviews explored barriers to participation, strategies to aid progression, content of dance sessions and how to retain participants. PE teachers pointed to a lack of dance in school and suggested that they were not confident to deliver dance. All dance instructors concurred that dance sessions should cover a number of dance genres and be structured in a way that helps foster friendships, enjoyment, ownership and a rapport between instructor and pupils.

The BGDP feasibility study63 was a three-arm, parallel-group, cluster randomised controlled study and economic evaluation, with schools as the unit of allocation. The intervention content was informed by the focus groups and interviews discussed above. Seven secondary schools were recruited and all Year 7 girls who were physically able to participate in PE were invited to participate. For practical reasons the sample was limited to 30 girls per school. Three intervention schools received two 90-minute after-school dance classes per week, for 9 weeks. The feasibility trial demonstrated that it is possible to recruit Year 7 girls and record the cost of the programme. The study also showed that girls would attend dance sessions and that it was feasible to collect PA data from the girls at three time points. The feasibility work suggested that it is possible to achieve a mean increase of 10 minutes of MVPA per weekday if the session intensity was increased and inactive creative time reduced. An embryonic resource use checklist was developed for use in the main trial economic evaluation. 62

Summary and rationale for the Bristol Girls Dance Project trial

Physical activity is important for the prevention of a number of diseases and also enhances mental well-being. PA declines during youth, with the start of secondary school being a crucial period of change. Girls are less active than boys at all ages and there is an absence of effective interventions to encourage PA in girls. Dance provides an opportunity for high levels of PA; it is also social and enjoyed by girls. As such, dance has the potential to function as a source of PA across the life-course. After-school dance programmes may be an effective means by which to increase the PA levels of girls. However, dance provision in schools is limited, and in the current economic climate schools have reduced funds to support after-school programmes. The goal of this study was to test whether or not after-school dance programmes can be effectively delivered in UK secondary schools.

Aims and objectives

The specific research aims of the BGDP (locally known as Active7) trial were as follows.

Research design

The trial was a two-armed, cluster randomised controlled trial conducted in 18 secondary schools across the greater Bristol area. The trial included process, economic, quantitative and qualitative evaluations.

The study was granted ethical approval from the School for Policy Studies, University of Bristol, and was sponsored by the University of Bristol. The trial was registered with International Standard Randomised Controlled Trial Register under the reference number ISRCTN 52882523. The original study protocol was sent to the National Institute for Health Research (NIHR) on 26 April 2013 and published as a journal article. 64 A small number of revisions were made to the protocol during the study. A summary of changes that were made to the original protocol paper are given in Table 1.

Study population and recruitment

The study sought to recruit 18 mainstream secondary schools (excluding special educational needs providers, dance academies and privately/independently funded schools) operating within three local authorities (Bristol City Council, North Somerset Council and Bath and North East Somerset Council). To participate, schools needed to have at least 30 Year 7 girls and be able and willing to allocate space for two after-school sessions per week for 20 weeks between January and June/July 2014. All schools fulfilling the inclusion criteria were invited to participate and the first 18 schools that agreed to participate were enrolled. Additional schools were placed in a reserve pool. We aimed to recruit between 25 and 33 Year 7 girls from each school (450–594 girls). To participate schools had to recruit at least 25 girls to the study. In the case of five schools, it was not possible to recruit the minimum number of girls, so reserve schools acted as replacements. This extended the recruitment period into November 2013.

Following school recruitment, a participant recruitment campaign was initiated in all 18 schools. A taster session was provided for all Year 7 girls who were able to engage in PE classes. Taster sessions were delivered by external dance instructors within school time (usually in a PE session). At the end of the taster session pupils were told about the study (including details of the randomisation and data collection commitments). The only exclusion criterion applied at pupil recruitment was that girls were able to participate in standard PE lessons.

Where more than 33 consent forms were returned, pupils were randomly ranked, with the first 33 pupils being selected at random using a computer-generated algorithm. Any girl dropping out of the study prior to baseline data collection was replaced by the subsequent pupil in rank order. This process was repeated as necessary. No replacements were made after baseline data collection.

School contacts signed a ‘school study agreement’, which stated that the school contact was happy for their school to take part in the study and for data to be anonymously stored in data sets and be used for publication. The form also collected contact details and basic demographic information of Year 7 girls. Written informed parental consent was obtained for all girls, and school contacts and dance instructors (taking part in the end of study qualitative work) also provided written informed consent. Following each taster session, all girls were given information packs for themselves and their parents/guardians. Girls wishing to participate had to return a completed parental informed consent form to school.

Our pilot work63 indicated that small reimbursements enhance data provision. As such, all girls received a £10 gift voucher on completion of each data collection phase (£30 in total). Control schools were each given a £500 donation to offset any costs incurred in facilitating data provision, which was received after the third stage of data collection.

Public and patient involvement

A Local Advisory Group (LAG) was formed to improve the relevance and delivery of the intervention for the schools and girls taking part. The LAG consisted of a variety of individuals in order to address the views of as many stakeholders as possible. The group met on three occasions during the study and played an important part in informing the intervention materials (dance diaries), behaviour management and attendance. The LAG also provided extensive input into the dissemination materials, which were developed in consultation with them. The group consisted of local council staff, school teaching staff, dance instructors, creative directors, school sport development managers and parents.

Baseline data

Both objective and self-reported data were collected from various stakeholders. A summary of the data collected can be found in Box 1.

Randomisation

Schools were randomised to control (n = 9) or intervention (n = 9) in a 1 : 1 ratio after baseline data had been collected using Stata version 12.1 (StataCorp LP, College Station, TX, USA) code specifically designed to balance the two arms by minimisation, which took account of four school-level factors, namely local authority, school size, baseline mean MVPA (%) and deprivation (based on the percentage of children with free school meals). Deprivation was assessed as the percentage of pupils in the school eligible for the Department of Education’s Pupil Premium (additional funding given to schools to support disadvantaged pupils and bridge the attainment gap between them and their peers). Randomisation was conducted by Keeley Tomkinson. Mark J Edwards made schools aware of their allocation soon after randomisation was conducted.

Intervention group

The nine schools that were randomised to the intervention arm received a 20-week dance intervention, consisting of two 75-minute after-school sessions per week (maximum 40 sessions), which were provided between January and June/July 2014. Dance sessions were led by 10 professional self-employed dance instructors who delivered a standardised programme which was developed in the feasibility trial. Instructors attended an induction session before the intervention began (December 2013) and a ‘booster session’ after the first term of intervention delivery (April 2014).

The dance programme focused on building girls’ autonomous motivation to be active and perceived dance autonomy, competence and relatedness through an autonomy-supportive environment. The programme provided exposure to a wide range of dance styles (instructors were given flexibility in what they delivered) (see Appendix 1). Girls in intervention schools were each given a ‘dance diary’ (see Appendix 2) which they were encouraged to complete between sessions. The diaries were intended to help children reflect on their learning and to encourage them to set their own goals.

Development of dance instructor training and manual

The BGDP aimed to increase girls’ autonomous (mainly identified and intrinsic) motivation for both dance and PA more broadly by providing them with fun and need-satisfying dance sessions led by an autonomy-supportive instructor. This was targeted through the ‘Guide for dance instructors’ (‘session-plan manual’) (see Appendix 1 for excerpts), the session design and content, and dance instructor intervention training. The ways in which the dance instructor manual and session content mapped on to the theoretical targets of the intervention are presented in Table 2. Dance instructors received a 1-day training session prior to the start of the intervention. The training was developed collaboratively between a dance instructor who had been involved in the feasibility study and study team. From the theoretical perspective it included a 2-hour session on SDT that highlighted the key features of the training manual (i.e. definitions and descriptions of motivation types, need-satisfaction and autonomy-supportive motivating styles) and how it can be applied in after-school dance sessions. Practical activities were used throughout the training to provide dance instructors with the opportunity to practice autonomy-supportive styles, ask questions and receive feedback. At the mid-point of the intervention, dance instructors attended a half-day booster session where the central tenets of the SDT-based teaching style were revisited and instructors were able to share their experiences of delivering the intervention and collaborate with study staff to resolve any problems they were facing.

Template for Intervention Description and Replication checklist

In an attempt to standardise the reporting of interventions, the Template for Intervention Description and Replication (TIDieR) checklist and guide has been proposed. 65 The 12-item checklist aims to improve the reporting and replicability of interventions. Table 3 summarises the BGDP trial in accordance with the TIDieR checklist.

Control group

Schools in the control arm did not receive the dance intervention and continued with their normal practice (no data were collected regarding what was offered during the intervention period in control schools). Control schools received a £500 donation to the school fund after all data had been collected from girls.

Measurements

Data were collected from all girls (intervention and control) at three time points:

1. Time 0 [T0 (baseline)]: September to November 2013.

2. Time 1 [T1 (weeks 17–20 of the intervention)]: June 2014.

3. Time 2 [T2 (baseline + 52 weeks)]: September to November 2014.

To assess any contamination of the control group from dance classes locally, we collected data on the extracurricular provision (including dance) offered by all 18 schools. This was done at each measurement point. In addition, girls were asked if they attended dance classes outside school at each measurement point. At baseline, parents completed a demographic questionnaire which included the free-text item ‘Does your Active7 child take part in any other after-school clubs?’. To assess differences between the intervention and control girls at T1 and T2, and to minimise any burden on parents beyond baseline, girls were asked to complete one questionnaire item at T1 and T2 measuring the types of activities (active and sedentary) that they had performed in the after-school period, in the evenings and at the weekend over the past 7 days. This item was intended to help understand the PA behaviour of the girls in both groups at all time points. All data were collected during school time on school premises.

Psychosocial questionnaire

At each time point girls were asked to complete a psychosocial questionnaire. Questionnaires were completed on a Google Nexus 7 (ASUSTek Computer Inc., Taipei, Taiwan) tablet. We hypothesised that the variables below would mediate the effect of the intervention on girls (Figure 1). The psychosocial questionnaire was therefore designed to provide all the information that would be needed to address this question, specifically the following constructs were assessed:

• motivation for dance and PA

• perceived autonomy, competence and relatedness in PA.

FIGURE 1.

Hypothesised mediation model.

All girls were asked to complete a 67-item questionnaire at T0. The questionnaire assessed psychosocial variables that could be influenced by the intervention and/or mediate the effect of the intervention on MVPA. Questionnaires assessed autonomous and controlled motivation for dance and PA,68 perceptions of autonomy, competence and relatedness69,70 within PA and self-esteem. 71

A modified questionnaire was used at T1 and T2. The girls were asked about the after-school activities in which they took part and whether or not they took part in organised dance outside school. At T1 only (during the intervention), girls rated the degree to which they perceived their dance instructor to be autonomy-supportive on a seven-point scale (strongly disagree to strongly agree). Items were based on the Health Care Climate Questionnaire72 and have been adapted for use in an activity setting. 73 At T2 these six questions were removed from questionnaires, as they were relevant only at T1 when girls were taking part in the intervention. A summary of the variables that were assessed at each time point can be found in Table 5.

TABLE 5 - Psychological questionnaire items

Items assessed	T0 (baseline, September to November 2013)	T1 (weeks 17–20 of the intervention, July 2014)	T2 (follow-up, September to November 2014)
SDT-derived questions (67 items)	✓	✓	✓
Weekday and weekend day activity questions (8 items)		✓
Sport climate questionnaire (6 items)		✓

Sample size

Sample size calculations were performed to detect a mean difference of 10 minutes of weekday MVPA between the intervention and control groups. The uninflated sample size required for analysis to detect a difference of 10 minutes/day MVPA, assuming a SD of 18 minutes63 with 90% power and a 5% two-sided alpha of 68 per arm. We estimated the 95% CI for the school-associated intraclass correlation (ICC) in the pilot study to be < 0.001 to 0.087. On the assumption that 20% of girls would not provide primary outcome data, the mean cluster size for analysis would be 24, resulting in a design effect of 3.0 using the upper 95% confidence limit for ICC. Thus, we aimed at (and succeeded in) recruiting a total of 18 schools and at least 450 girls.

Blinding

All project staff, excluding the Trial Manager and Process Evaluation Research Associate, were blinded to the allocation of intervention arms. Primary and secondary analyses were conducted blind by the statistical analysis team. Blinding was broken at the presentation of the main trial results to the Trial Management Group on 2 February 2015. Statisticians were unblinded at this point in order to undertake the per-protocol analysis which examined the intervention effect for girls who had adhered to the intervention and the statistical elements of the process evaluation which was conducted only in intervention schools.

Process evaluation methods

Qualitative analysis

The framework method, a form of thematic analysis defined by the systematic production of a matrix which reduces data into a series of codes, was used. 79 This method is useful for condensing and summarising large qualitative data sets. It is an approach that does not exclusively align with a particular theoretical or epistemological perspective and, therefore, can be applied using an inductive or deductive approach. 79 Analysis consisted of the following six steps.

1. Transcripts were read and reread to form preliminary impressions of the data.

2. Initial codes were created to summarise and interpret data. Both inductive and deductive coding was used. Deductively, the analysis probed data to understand whether or not the intervention was delivered in line with SDT. Pre-defined codes were created from the central tenets of SDT. 45 In addition, a pre-defined code ‘school context’ examined the potential interaction between school context (e.g. participant- and area-level demographic factors and school culture) and intervention delivery and experience. The pre-defined codes were broad, with the primary purpose of categorising relevant information, which was further interrogated. Initial codes were inductively produced independently by four researchers (JK, ME, SS and TM) who each coded three transcripts (one from each informant group).

3. Initial codes were discussed, refined and combined to produce three coding frameworks, one for each informant group.

4. The coding frameworks were applied to all remaining transcripts by three researchers (JK, ME and TM). New and redundant codes were discussed in frequent meetings of the analysis group (JK, ME, SS and TM) and iterative amendments were made to each framework.

5. The coded data were then entered into a framework matrix in NVivo version 10 (QSR International, Warrington, UK). The framework condensed the volume of data and summarised codes with illustrative quotations. This process facilitated reflections on salient codes between participants within each informant group.

6. The three frameworks were triangulated using the convergence coding matrix approach,80 in which the codes for informant groups were compared to assess the degree of convergence as either: ‘Agreement’, ‘Partial agreement’ and ‘Silence’ or ‘Dissonance’ (see Appendix 6). In total there was agreement between informant groups in 22 themes, partial agreement in 26, silence in 39 and dissonance in 6 out of 77 themes.

To ensure trustworthiness, we applied four criteria: credibility, transferability, dependability and confirmability (see Appendix 7). 81 The main themes elicited from the interviews are presented in Chapters 6–8 and are supported by illustrative, relevant and representative quotations.

Statistical analysis

The analysis and presentation of the trial was carried out in accordance with Consolidated Standards of Reporting Trials (CONSORT) guidelines, with the primary comparative analyses being conducted on an intention-to-treat basis and due emphasis placed on CIs for the between-arm comparisons. To take appropriate account of the hierarchical nature of the data, we used multivariable, mixed-effects, linear regression to estimate the difference in the primary outcome for intervention group versus control, adjusting for baseline MVPA and randomisation variables.

For the baseline (T0) summary statistics, the mean (SD) was reported for normally distributed variables and the median [interquartile range (IQR)] for skewed variables. Normality was determined by the Shapiro–Wilks test, histograms and normal probability plots. Proportions were reported as percentages. Summary statistics were reported for each group. A kernel density estimation was used to calculate the probability density function of non-parametric continuous variables, and, thus, kernel smoother plots were used to describe such distributions.

Linear mixed models were used to assess the primary and secondary outcomes, and a logistic model was used for proportional data. The variables to be included in the regression models were predetermined in the analysis plan. A typical model included the outcome at T2 as the dependent continuous variable (e.g. weekday MVPA at T2), whereas the independent variables consisted of the main variable of interest (the two arms of the trial) adjusted for T0 (e.g. weekday MVPA at T0), the number of valid weekdays at T0, the number of valid weekdays at T2 and the variables used in the randomisation process. Model assumptions were checked by testing the residuals for normality and plotting the linear predictor against the residuals to ensure that no trend was present.

After the statistician was unblinded, a per-protocol analysis was performed. The trial allocation and attendance data were merged with the accelerometer data and a new variable ‘adherence’ was generated. For the T2 analysis, all girls who had not adhered to the intervention in the intervention arm were dropped from the data set and the regression model for outcome ‘weekday MVPA at T2’ was rerun on this data set. This type of analysis was repeated for ‘weekend MVPA at T2’, ‘weekday MVPA at T1’ and ‘weekend MVPA at T1’. Owing to the bias introduced by dropping girls who did not adhere in the intervention group in the per-protocol method, a complier average causal effect (CACE) analysis was conducted, which included all girls, regardless of whether or not they adhered. Each girl in both groups was given a new variable of adherence (0 = no, 1 = yes) whereby all girls who adhered in the intervention group were coded 1 and all girls who did not adhere were coded 0, and all girls in the control group who adhered to the control (i.e. did not move to an intervention school and start a dance class) were also coded as 0. Instrumental variable regression was then conducted to identify whether or not there was any difference between the two arms with respect to adherence.

We intended to explore whether or not any effect of the intervention on the primary outcome was mediated by autonomous and controlled motivation for PA and/or perceptions of autonomy, competence and relatedness need satisfaction in PA, using the methods described by Emsley et al. 82 However, as there was no effect on the primary or secondary outcomes, we did not conduct the mediation analysis.

Quantitative process evaluation data (e.g. attendance rates) were analysed using appropriate descriptive statistics for normally distributed variables (using the mean and SD) and variables without such a distribution (using the median and IQRs). Ratings of instructors’ teaching styles were made from visual observations and audio recordings of 21 items. Each item was rated in 5-minute intervals during the session, these values were summed and divided by the number of intervals in the session across five teaching elements: relatedness support; structure before the activity; structure during the activity; autonomy support; and controlling teaching behaviour. A rating of the overall impression of the session was also given for autonomy-supportive teaching style, structure and instructor relatedness support. All analyses were performed using Stata 13.1 (StataCorp LP, College Station, TX, USA).

Economic evaluation

Full details of the economic evaluation are shown in Chapter 4.

Recruitment

Issues encountered during recruitment

A less conventional aspect of the recruitment campaign was the provision of taster sessions in schools. Organising taster sessions was logistically difficult. A total of 65 taster sessions took place within a short time period. These were delivered by six instructors and study staff made recruitment presentations at each taster session. Schools were largely confined to scheduling taster sessions in place of PE lessons, meaning that there was little flexibility on their part.

When schools withdrew from the study (owing to insufficient pupil sign-up) there were time pressures on recruiting replacements. Data collection was planned to take part between September and October 2014; thus, when replacement schools were sought they had to be recruited quickly (as taster sessions had to be arranged and necessary information handed out and returned).

No adverse events were reported during the intervention delivery.

Results

Primary analysis

In order to conduct the primary analyses, the T0 and T2 data sets were merged, including information on the variables used for school randomisation (local authority, school size, baseline mean MVPA and deprivation) for each girl. The main analyses (Table 9; this table also includes results of the per-protocol CACE analysis, which are discussed further below, see Complier average causal effect analysis), which adjusted for baseline values, found no evidence for a difference in weekday MVPA at T2 in girls allocated to the intervention group (mean 56.55 minutes) compared with those in the control group (mean 53.15 minutes) (adjusted difference in means –1.52, 95% CI –4.76 to 1.73). There was also no difference for the secondary outcome of weekday MVPA at T1 (adjusted difference in means –1.52, 95% CI –5.03 to 1.98). No evidence was found for a difference in any accelerometer-derived variables at T1 or T2.

TABLE 9 - Means and SDs by trial arm and linear mixed model adjusted for imbalance at baseline for weekday MVPA at T1 and T2

For between-group differences the control group is the reference group, with models adjusted for baseline mean weekday MVPA, local education authority, school size, school-level deprivation, school-level baseline MVPA, the number of total valid weekdays at T0, the number of total valid weekdays at T2 (or T1) and school-level clustering.

Primary comparison.

Key secondary outcome.

Dashes indicate that there are no data available.

Variable	Control			Intervention			Intervention vs. control adjusted difference in means (95% CI)a	p-value
	n	Mean	SD	n	Mean	SD
T2 mean weekday MVPAb	262	53.15	19.61	246	56.55	21.92	–1.52 (–4.76 to 1.73)	0.359
T1 mean weekday MVPAc	265	57.69	19.39	256	60.46	22.98	–1.52 (–5.03 to 1.98)	0.395
T2 MVPA weekday CACE analysis	508	–	–	–	–	–	–4.79 (–14.53 to 4.96)	0.336
T1 MVPA weekday CACE analysis	521	–	–	–	–	–	–4.86 (–18.41 to 6.91)	0.365

Model checking

To ensure that the model was a good fit for the data, residuals in Figure 3 should have a mean of zero and constant variance. The mean of the residuals is –4.30 × 10^–8 and the mean of the standardised residuals is –1.34 × 10^–9, so the fit is relatively normal.

FIGURE 3.

Residual plots for the main model. (a) Plot of residuals; and (b) plot of standardised residuals. Twenty-two observations lie outside 1.96, expected to be 5%, 22/508 = 4.3%.

Secondary analysis

Crude results

Means and SDs for accelerometer-assessed outcomes at T1 and T2 are presented in Table 10. No effect was detected at T1 or T2 for any outcome.

TABLE 10 - Means and SDs by trial arm and linear mixed model adjusted for imbalance at baseline for accelerometer assessed secondary outcomes at T1 and T2

For between-group differences the control group is the reference group, with models adjusted for baseline value, local education authority, school size, school-level deprivation and school-level clustering.

Odds ratios presented in coefficient column.

Variable	Control			Intervention			Intervention vs. control adjusted difference in means (95% CI)a	p-value
	n	Mean	SD	n	Mean	SD
T2
Mean weekend day minutes of MVPA	145	36.56	26.16	124	39.65	23.21	–1.75 (–7.51 to 4.01)	0.552
Mean weekday CPM	262	446.83	137.81	246	478.75	144.94	–2.44 (–25.25 to 20.38)	0.834
Mean weekend CPM	145	405.04	228.96	124	450.98	263.89	–4.11 (–61.07 to 52.86)	0.888
Proportion of girls meeting 60 minutes of MVPA per weekdayb	262	0.32	0.47	246	0.39	0.49	–1.18 (–1.82 to 0.76)	0.458
Proportion of girls meeting 60 minutes of MVPA per weekend dayb	145	0.12	0.33	124	0.15	0.36	–1.11 (–2.39 to –0.52)	0.787
Mean weekday sedentary minutes	262	533.01	80.54	246	515.12	80.22	–6.79 (–23.60 to 10.03)	0.429
Mean weekend sedentary minutes	145	475.14	95.22	124	463.66	105.92	0.62 (–22.42 to 23.66)	0.958
T1
Mean weekend day minutes of MVPA	159	42.57	27.71	130	48.92	32.19	1.26 (–5.70 to 8.22)	0.723
Mean weekday CPM	265	500.35	177.32	256	529.42	157.72	–7.48 (–35.06 to 20.11)	0.595
Mean weekend CPM	159	492.21	371.47	130	543.67	284.39	6.27 (–72.10 to 84.65)	0.875
Proportion of girls meeting 60 minutes of MVPA per weekdayb	265	0.42	0.49	256	0.47	0.50	–1.11 (–1.68 to –0.73)	0.637
Proportion of girls meeting 60 minutes of MVPA per weekend dayb	159	0.17	0.38	130	0.27	0.45	–0.82 (–1.56 to –0.43)	0.543
Mean weekday sedentary minutes	265	522.96	85.46	256	502.91	87.01	–7.72 (–27.32 to 11.87)	0.449
Mean weekend sedentary minutes	159	464.13	92.28	130	452.46	98.21	–8.94 (–31.91 to 14.04)	0.446

Model checking

Figure 4 shows the residual plots for the key secondary model. The mean of the residuals is –1.60 × 10^–8 and the mean of the standardised residuals is 6.48 × 10^–10, so approximately 0.

FIGURE 4.

Residual plots for the key secondary outcome model. (a) Plot of residuals; and (b) plot of standardised residuals. Twenty-seven observations lie outside 1.96, expected to be 5%, 27/521 = 5.18%.

Summary of quantitative results

The main analysis provided no evidence that the BGDP had any effect on accelerometer-derived PA levels of Year 7 girls, during the intervention and post-intervention periods. One-third of girls in the intervention arm met the adherence criteria of two-thirds of sessions delivered in schools. When the data were reanalysed using a CACE per-protocol analysis, the models provided no evidence of difference in weekday MVPA during and post intervention. Further exploratory analyses indicated that intervention group girls attending dance classes during the measurement period achieved more minutes of MVPA, more minutes of light-intensity PA and more accelerometer CPM between 15.00 and 17.00 on dance days versus non-dance days. Therefore, data show that any impact was relatively small and is weakened after considering data from non-dance days. An analysis of the psychosocial variables suggested that autonomous motivation towards both dance and PA decreased in the intervention group compared with the control group at T1 and T2. Although there was strong evidence against the null for these differences, the magnitude of the differences was relatively small when interpreted using the scale of measurement. Similarly, although change in autonomy need satisfaction did not differ between intervention and control groups from T0 to T2, perceptions of competence and relatedness need satisfaction showed small decreases in intervention arm girls compared with control arm girls at T1 and T2. The results were similar when the data were reanalysed using a CACE model.

Aims and perspective

This chapter describes the methodology and results of an economic evaluation of the BGDP. The economic evaluation was conducted from a public-sector perspective and included costs to local authorities and to the NHS.

The aims of the economic component of the project were:

• to determine and categorise the resources used and monetary value of cost inputs in training, preparation and delivery of the BGDP programme

• to determine the cost-effectiveness of the BGDP programme compared with no active intervention

• to determine the cost–utility of the BGDP programme compared with no active intervention.

This short-term analysis estimates the cost-effectiveness and cost–utility of the intervention compared with no active intervention over the 1-year period of the trial. For these analyses it was assumed that there were no differences in costs or intervention effects between the intervention and control group beyond the 1-year follow-up period.

Methods

Results

Summary

The BGDP was inexpensive to deliver, costing an additional £73 per girl compared with no active intervention over the 1-year period of the trial. It is important to note that the £73 cost per child may underestimate the true cost per girl, as it is based on all children who signed up in a given school actually attending. However, as demonstrated in the attendance data, many children in study schools did not attend sessions, so the cost per attendee is higher. The additional cost per school was relatively consistent, ranging from £2257 to £2551. There was no evidence to suggest that the BGDP was effective at increasing MVPA. Self-reported health-related quality-of-life measures were also similar between the intervention and control groups at T2. Therefore, owing to a lack of effectiveness, the intervention, despite its low cost, was not cost-effective.

This study provides information on resources used and monetary value of cost inputs in training, preparation and delivery of the BGDP programme. These findings will be relevant to commissioners who have limited resources to improve the health of children in schools and who need to know whether or not investments in after-school PA programmes are affordable within a set budget. 88 Findings reveal that the BGDP is of comparable cost to other school-based PA interventions. 89 This study demonstrates that collating full cost data via a resource use checklist is feasible and practical, and can produce an accurate estimate of mainstream costs.

It is important to highlight that the data presented in this study indicate that the EQ-5D-Y was unresponsive to changes in a ‘healthy’ school population, with no differences identified between the intervention and control groups. In addition, health utility values were generated from adult UK population norms for EQ-5D,87 as child values were unavailable at the time. Future research should consider alternative health-related quality-of-life measures, which are appropriate to the population under investigation, which is young and healthy.

Qualitative sample characteristics

A total of 59 girls took part in the focus groups (Table 38) and group sizes ranged from three to eight girls. Girls in the highest attendance third were more likely to agree to participate in a focus group. Participation in one school was low (n = 3) which, in part, may be due to the interview being organised (by the school contact) at an unappealing time of day (after school). A total of 10 dance instructors were interviewed, including one reserve instructor, two instructors who each delivered half of the intervention in one school and one dance instructor who taught in two schools. One school contact was interviewed in each of the intervention schools.

Process evaluation findings

Chapters 6–8 reflect on key findings from the process evaluation. Chapter 6 addresses the implementation and impact of the intervention on the stakeholders. Chapter 7 focuses on the theoretical fidelity within the intervention. In particular, it looks at dance instructors’ experiences of delivering a theory-based intervention and girls’ perceptions on delivery styles and impact on their psychological needs of autonomy, competence and relatedness. Chapter 8 provides a summary of the challenges of and lessons learnt through conducting an extracurricular PA intervention. The main aim of Chapter 8 is to identify key areas of improvement in order to facilitate successful delivery of such future interventions. Collectively, these chapters provide a comprehensive overview of the process evaluation data collected.

Implementation

This section reports on the intervention dose, compliance adherence, dance instructor training and the degree to which instructors adhered to the session-plan manual. The section provides an overview of the extent to which the intervention was implemented in line with the initial design and guiding materials.

Dose

Four schools delivered the maximum of 40 dance sessions (range of sessions 37–40). The average number of girls signed up to the study in each school was 31.56 (range 26–33). The average proportion of girls attending two-thirds of all possible dance sessions was 9.10 per school (range 1–20).

Compliance adherence

Figure 5 displays the attendance by school over the 40 sessions. The mean number of girls per session across all intervention schools was 12.84 girls (SD 6.96 girls) (maximum 32 girls). Mean attendance at the first session was 24.33 girls (SD 5.50 girls), this decreased to 10.25 girls (SD 7.59 girls) by the final session. Sessions (n = 12) in which fewer than two girls attended are explained in general by a school event or communication breakdown between instructor and school; in some cases, reasons were not given. There is considerable variation within and between schools, with attendance fluctuating throughout the intervention period. Low attendance in School 53 [mean 6.41 girls (SD 3.39 girls)] can partly be explained by the high number of girls who never attended a dance session (n = 13).

FIGURE 5.

Attendance per dance session across all schools.

Table 39 shows additional attendance data. Percentage attendance across all sessions ranged from 20.85% in School 21 to 64.94% in School 23. Percentage attendance is slightly higher when calculated as a product of girls who attended one of the first four sessions, particularly in School 53, where average attendance increased from 22.89% to 37.71%; however, overall average attendance across all schools increases to only 44.50%.

TABLE 39 - Additional attendance data

Dashes indicate that no data are available.

School ID	Number of sessions completed	Number of girls at sign-up	Average attendance over all sessions, mean (SD)	Average attendance over all sessions (%)	Attendance relative to those attending one of the first four sessions (%)	Number of girls who never attended	Number withdrawn from dance sessions
21	40	33	6.88 (4.94)	20.85	24.55	6	3
23	40	33	21.43 (5.76)	64.94	64.92	0	0
32	37	33	13.30 (5.22)	40.30	45.85	4	1
42	37	33	14.16 (7.07)	42.91	44.26	1	1
51	40	32	14.68 (5.71)	45.88	45.86	0	0
53	39	28	6.41 (3.39)	22.89	37.71	13	1
61	37	33	14.51 (5.29)	43.97	50.05	3	8
62	40	26	9.75 (5.86)	37.50	39.00	6	3
72	38	33	14.66 (5.04)	44.42	48.86	2	6
Average	39	32	12.84	39.90	44.50	4	3
Total	348	284	–	–	–	35	23

Although the attendance levels may seem concerning, the qualitative data suggest that a decline in attendance is to be expected in after-school activities. School contacts and dance instructors generally viewed the attrition as similar to other after-school clubs.

We tend to have a lot of very high numbers to start with. So for example, when we start netball club in Year 7 we might have 50, 60 girls come into netball club and then the drop off, it goes down to about 30 by the end.

School contact 23

My numbers dropped which is to be expected in summer term.

Dance instructor 42

When you start a new project you quite often might get loads of people and then you [. . .] whittle it down to your core people I guess, which I guess can be quite usual.

Dance instructor 23

However, two school contacts suggested that the attendance decline was notably high.

You’ve always got all the enthusiasm at the start and then ‘Oh, I’m bored now’. Naturally, but not by the amount that it did [in Active7]. That’s quite unusual.

School contact 21

It did drop off . . . you know the number of girls not coming . . . we’d have to sort of talk to them about [. . .] I was quite surprised in a way that, it was  . . . that numbers did drop down.

School contact 51

The high attendance in School 23 (see Figure 5) was cited as being a result of the ‘novelty’ of the study.

We haven’t really had something like this, like Active7. That’s why loads of people started attending.

Focus group 23

One dance instructor thought that these girls valued their place on the project.

I felt like they wanted to stay in the project but they also understood that this was exclusive to them [. . .] so I think they really valued their place in the class.

Dance instructor 23

The school contact in School 23 attributed the high attendance (along with attitudes and behaviour) to the relatively high socioeconomic position of girls’ families.

The type of students we’ve got in this school . . . they don’t want to let people down so I think they’ve got that in the back of their minds. They are aware that it’s a good opportunity for them, and they’ve got parental support so I think that’s a major impact.

School contact 23

I think it’s just because the school’s in a good area that the students are more behaved . . . well behaved, got better attendance compared to an area like, I don’t know, maybe [area] or something like that.

Dance Instructor 23

The inclusive rather than exclusive attendance policy of the BGDP allowed girls to return if they missed sessions. This may have encouraged girls to return to the club after an absence.

Active7 is very different to a normal group and when you’re working towards a performance and you haven’t been there for eight weeks it’s usually you’d just be like ‘no’. So I think that openness is really nice.

Dance instructor 32

Absences followed by reattendance was sometimes difficult because the instructor was perceived to be annoyed with girls for missing sessions. However, some thought that this was justifiable and found it easy to catch up.

Going back to like pressure, she always used to get a bit cross if we weren’t there for a few weeks.

Focus group 32

She was strict because she wanted you to be dedicated and turn up.

Focus group 53

But you did find it okay to come back?

Yeah.

I learnt the dance really quickly anyway when I came back.

Focus group 53

Factors that influenced attendance

Figure 6 shows the reasons why girls did not attend sessions. The most common reason that girls cited (in the questionnaire completed at follow-up) was ‘I did something else on the days Active7 was run’. Another often-cited reason for not attending was ‘When I signed up I thought it would be different’.

FIGURE 6.

Self-reported reasons for not attending Active7 sessions. TV, television.

The qualitative data suggest that attendance was influenced by girls’ commitment to the project, events outside of Active7, friendship groups and the design of the intervention. Dance instructors felt that attendance reflected the girls’ level of commitment.

It dropped down to those that really were committed.

Dance instructor 21

They got the impression that once you committed to the project that you have to stick with it.

Dance instructor 23

There’s lots of other groups going on at the school and when the weather was nice, or if it was a friend’s birthday or anything like that, it meant that the attendance was all over the place.

Dance instructor 32

All participant groups discussed how competing priorities outside Active7 affected attendance.

We sometimes had sports matches or something, so we couldn’t go.

Focus group 23

A lot of the students might be doing other things in the week and things like that it was a big commitment.

School contact 51

They’re not used to kind of having to do so much stuff after school and there’s obviously lots of other things that are offered to them in secondary school.

School contact 53

Friendship groups within Active7 affected attendance levels in both positive and negative ways. Some girls went because their friends attended, whereas if a friend dropped out, this could negatively impact attendance.

Not sure whether they were really [. . .] interested in dance. They’d come every week because they’re part of the gang.

Dance instructor 42

There were times I know when they would walk home together and so if one decided that they didn’t want to do the session then I might lose like four.

Dance instructor 62

So like when you’re doing stuff out of school and with like you’re friends and that and then you can’t go because you have to go to Active7.

Focus group 61

Various elements of the intervention may also have contributed to the declining attendance. This is discussed at length in Chapter 8, Project design, however, a brief summary is provided here. Supporting the qualitative evidence, the days on which Active7 was scheduled, specifically Fridays, was a factor that negatively affected attendance.

We finish early on Fridays as well so it’s just like there’s no point in finishing earlier [. . .] That was one of the main reasons why I kind of like left.

Focus group 51

Some school contacts and dance instructors suggested that the length of the project and the quantity of sessions per week had a negative effect on commitment.

They don’t normally last 20 weeks or whatever, whether that was just too much for them.

School contact 51

The two sessions a week, I think it perhaps was too much because it was [. . .] a big commitment.

School contact 42

I’ve said about changing to one session a week [yeah]. I think that . . . yeah, I think the commitment . . . they did brilliantly to commit but I think as they get older that commitment is – if it was to stay with two sessions a week – it would definitely, um [yeah], definitely dwindle away and float off I think.

Dance instructor 23

However, feelings were mixed for girls. Some felt that two sessions helped them to remember dance material and keep them fit, whereas others thought that it got boring.

[Two sessions per week] gives us chance to remember things [. . .] it gives us more chance to practice.

Focus group 23

It almost because like a chore to some people.

Focus group 72

Withdrawal from sessions

Table 40 presents the number of girls in intervention schools withdrawing from the study, by school. Pre-intervention withdrawal relates to those girls who did not attend any sessions. Only three schools had no pre-intervention withdrawals. Three schools had pre-intervention withdrawal in excess of 10% of pupils signed up, with School 53 seeing almost 30% of girls not attend any sessions. During-intervention withdrawal refers to girls who stopped attending Active7 sessions, and complete withdrawal refers to the number of girls who withdrew from the study completely (including data collection). Similar to pre-intervention withdrawal, during-intervention withdrawal was highest in schools 21 and 53, with > 36% and 48% of girls in these schools, respectively, withdrawing at some point.

TABLE 40 - Withdrawal rates in intervention schools

School ID (n pupils at T0)	Pre-intervention withdrawal: attended no sessions, n (% of total sign-up in school)	During-intervention withdrawal, n (% after attending at least one session)	Complete withdrawal, n (%)
21 (33)	6 (18.18)	5 (18.52)	0 (0.00)
23 (33)	0 (0.00)	0 (0.00)	2 (6.06)
32 (33)	4 (12.12)	0 (0.00)	1 (3.03)
42 (33)	0 (0.00)	1 (3.03)	0 (0.00)
51 (32)	0 (0.00)	4 (12.50)	0 (0.00)
53 (28)	8 (28.57)	4 (20.00)	1 (3.57)
61 (33)	3 (9.09)	1 (3.33)	0 (0.00)
62 (26)	1 (3.85)	0 (0.00)	1 (3.85)
72 (33)	3 (9.09)	2 (6.67)	0 (0.00)
Overall (284)	25 (8.80)	17 (6.56)	5 (0.70)

Impact of attendance levels on intervention delivery

Some dance instructors were frustrated and disappointed by the decrease in attendance.

It was quite depressing to write down and see it, how many people didn’t want to come.

Dance instructor 23

You can take that [poor attendance] really personally, which you shouldn’t do. But yeah it’s definitely interesting to hear that it wasn’t just me.

Dance instructor 32

I haven’t really experienced that [low attendance] before . . . in the other places that I’ve taught it all remains, you know, on a higher level, so I was like ‘mm, am I doing something wrong?’.

Dance instructor 62

Poor attendance resulted in the repetition of previous session content, a method used to allow absent girls to keep up with the progressive building of dance pieces. However, dance instructors found this challenging, particularly as it prevented them from moving forwards.

You’re getting different people every time and that’s a real juggling act because if they want to make their own material they have to make something new every session.

Dance instructor 23

From my point of view it disrupts the class a lot because it means I have to always go back to the beginning of the choreography to make them feel included [. . .] it creates a bit of boredom for the others.

Dance instructor 23

You can’t really progress to another section of a dance really if everyone doesn’t know it.

Dance instructor 62

As attendance declined, however, smaller groups of ‘committed’ girls were preferred by the instructor and those attending. This also made teaching easier and facilitated the formation of closer instructor–participant connections.

As the numbers decreased, although it was heart-breaking sometimes, it was so much easier to teach.

Dance instructor 21

The girls said on a number of occasions ‘I love the group like this, like much smaller’.

Dance instructor 32

I don’t really know if like you can do this . . . but like less people because now there’s not that many people there it’s so much more relaxed and like everyone can just be themselves and everything. When there’s like loads of new people you don’t want to like stand out or anything. If you could do Active7 with less people.

Focus group 61

Good attendance in one school gave the instructor a confidence boost and was experienced as reassuring because she had taken over from another instructor and felt that there was a risk that girls might stop attending during this period of transition.

When the sessions started from my point onwards, the attendance, in my opinion, stayed really, really good and quite consistent.

Dance instructor 23

Fidelity to the session-plan manual

In total, 93 (26.72%) sessions were reported as being ‘very’ much like the manual, 164 (47.13%) were ‘somewhat’ like the manual and 90 (25.86%) were ‘not at all’ similar to the manual (Table 41). The extent to which the manual was followed throughout the intervention period fluctuated. However, in the first five sessions, instructors said that the session was followed ‘somewhat’ and ‘very’ like the manual. This varied more between sessions 6 and 40 (Figure 7).

TABLE 41 - Fidelity to the session-plan manual per school

Missing data attributable to no girls attending one session.

School ID	Very		Somewhat		Not at all		Missing
	n	%	n	%	n	%	n	%
21	17	42.50	20	50.00	3	7.50	0	0.00
23	4	10.00	16	40.00	20	50.00	0	0.00
32	10	27.03	24	64.86	3	8.11	0	0.00
42	18	48.65	7	18.92	12	32.43	0	0.00
51	17	42.50	20	50.00	3	7.50	0	0.00
53	1	2.56	7	17.95	30	76.92	1a	2.56
61	0	0.00	37	100.00	0	0.00	0	0.00
62	8	20.00	16	40.00	16	40.00	0	0.00
72	18	47.37	17	44.74	3	7.89	0	0.00
Total	93	26.72	164	47.13	90	25.86	1	0.29

FIGURE 7.

Dance instructor self-reported adherence to manual.

Views on the session-plan manual were generally mixed. For some, it was viewed as a useful resource.

To actually be handed all of this information that someone had put together for you and just a really, a really good resource.

Dance instructor 61

However, instructors felt that the level of detail was unnecessary given their existing level of training and experience.

Too much. Way too much.

Dance instructor 53

In particular, the suggested timings of sessions were thought to be unrealistic and cause worry and stress, leading to a less well-delivered session.

The first class for me, I was so worried about keeping it to the book that actually that kind of made it worse because the book was great but it was to the minute and therefore when you’ve got a class that takes five minutes to actually get in line, that’s five minutes gone [. . .] I think I was so worried about that.

Dance instructor 21 and 51

The timings were just silly, because [. . .] in reality you’re not going to be looking at your watch.

Dance instructor 42

I found it hard to stick to and I think the very specific timings in there are often unrealistic.

Dance instructor 62

However, one instructor, although she did not follow the timings, thought that they were a useful guide and should be kept in the manual.

She’s got like three minutes for this, two minutes for that, and that was totally irrelevant for me. But that does cater for everyone’s style because you can ignore it or you can use it so it’s good to have it there.

Dance instructor 23

Furthermore, session-plan content in initial sessions was at odds with how instructors would have normally led the sessions.

Where it went wrong for me was [when] trying to stick to the manual I maybe did things that near the beginning that I wouldn’t have done and that maybe set things up slightly against me in terms of managing behaviour and stuff.

Dance instructor 42

As such, dance instructors altered session plans accordingly; the manual was used as a ‘guide’ and allowed participant input to structure sessions.

My girls were like my manual if that makes sense. They were like guiding [. . .] the programme for me.

Dance instructor 61

As mentioned above, adhering to the manual was often difficult when faced with intermittent attendance.

I found it [the manual] hard to stick to. But that’s because I wouldn’t get the same kids from one session to another so you can’t have continuity.

Dance instructor 62

During the training and within the session-plan manual the instructors were encouraged to develop behavioural guidelines with their groups. However, only four dance instructors discussed creating these guidelines with their groups.

On maybe [the] second session, third session, we together wrote some guidelines and rules if you like, which were about not talking when other people were talking and all the usual sort of things and we always came back to those.

Dance instructor 32

A total of 14 sessions were covered by different instructors, because the main instructor was not available to deliver that session. In these cases, the session may have been a ‘one off’ and the delivery did not refer to the manual.

I just go in and treat it [covered sessions] like a taster session [. . .] so it was just going in and having a bit of fun really. Kind of letting them pick the music . . . or they show me a dance that they’ve done.

Dance instructor 53

Receipt of the intervention

This section focuses on receipt of the intervention, including the perceived levels of enjoyment and exertion of girls in the sessions, and the qualitative perceptions of the impact of the intervention.

Enjoyment

The average perceived enjoyment ratings are presented by school and visit number (1–4) in Figure 8. The self-reported findings suggest that girls very much enjoyed the dance sessions [mean 4.3/5 (SD 0.3)]; these high levels were maintained throughout the intervention.

FIGURE 8.

Mean perceived enjoyment per school on four occasions during the 20-week intervention.

Qualitative work with girls further suggests that the girls enjoyed the sessions. Group work and dancing to popular music were identified as particularly enjoyable.

I think because the sessions were actually really good and I enjoyed them quite a lot.

Focus group 72

[The second dance instructor] had like different music every lesson, like recent music and stuff. So that made it like more fun because we like knew the songs and stuff.

Focus group 23

Dance instructors also felt that the girls enjoyed the sessions.

We did musical theatre which they absolutely loved.

Dance instructor 21 and 51

Girls did not enjoy some dance styles; for example, ‘dance style A’ was particularly disliked because the choreography was considered embarrassing. This was seen as contributing to the decline in attendance among girls. Repetition of dance routines and learning sequences/moves when they had missed a session were also disliked by girls.

It dropped off a little bit when she changed styles to [dance style A], and after there was a huge drop off. I actually think a lot of them would have carried on and stayed doing street.

School contact 61

A bit embarrassing really [dance style A]. But I didn’t even do it but it just looks embarrassing because like you do weird moves.

Focus group 61

And like when someone like didn’t come and then they come the next day [. . .] we would have to like show it all again.

Focus group 42

They’d get bored of re-teaching it [the material that was missed] when there would be another new person at the next session and then another new person.

Dance instructor 53

Exertion

Mean exertion levels were low [3.7 (SD 0.9)] and remained consistently so at each measure (Figure 9). In contrast, the qualitative data suggest that many girls did find sessions physically challenging. Several dance instructors took note of this, suggesting that some of the girls’ fitness levels meant that they struggled to keep up.

We did this exercise [. . .] After 30 seconds the first time we were tired and couldn’t do it. Then after a few sessions [. . .] we could do it for like 10 minutes.

Focus group 53

Some of . . . their fitness levels were so low that they would be dipping in and out purely because they were out of breath or they did need a sit down.

Dance instructor 42

FIGURE 9.

Mean perceived exertion by school on four occasions during the 20-week intervention.

Summary

The data presented in this chapter highlight issues regarding attendance at dance sessions. The decrease in attendance over time was considerable; however, the number of girls who never attended a session or withdrew from sessions was relatively low. Dance instructors and girls felt that the dance sessions brought about positive health and psychological changes, particularly as regards their confidence to dance. Enjoyment levels remained high; however, some of the girls disliked the repetition of content, certain dance styles and having to perform to a crowd. Exertion scores were low throughout, although both the instructors and girls described the sessions as physically exerting. Training was felt to adequately prepare the instructors to deliver the intervention, although some areas were described as inappropriate given the level of experience of some instructors. Fidelity to the manual was initially relatively high, but as instructors got to know the girls, they moved away from the manual to allow for greater input from girls.

The decline in attendance was noted as being common to after-school programmes and not indicative of dance interventions per se. This pattern in attendance is reflected in other after-school studies. 38,90,91 Although some dance instructors were disappointed with the decline, it was not always perceived to have a negative impact. A variety of informant groups described the decline as facilitating stronger relationships between the girls and dance instructors who continued to attend. Dance instructors also found it easier to teach smaller groups. Various authors promote and have used small groups of children in the delivery of interventions. 92,93 Despite this, a result of the fluctuating attendance was that dance instructors found it difficult when having to repeat content. As such, greater emphasis needs to be placed on consistent attendance.

The data demonstrated that the key reasons for not attending sessions related to factors outside the study, such as having other commitments or events to attend. These findings are comparable to previous interventions with young people. 91,94 Friendship groups were seen both to challenge and to encourage attendance. Some girls would attend sessions because their friends did; however, this meant that some stopped attending if friends missed sessions. Girls have previously reported that they would be more likely to attend a session if their friends encouraged them to go. 19 In the current study, girls described enjoying the sessions, further suggesting that the decline in attendance is not related to content; indeed, 12.3% of the girls did not attend any sessions. This is in line with other studies that have reported high enjoyment yet have still experienced declines in attendance. 39,94,95

Aspects that were enjoyed by girls included learning new dance material, group work and dancing to current music. The perceived levels of exertion for dance sessions were low, similar to findings from the BGDP feasibility trial in which average exertion was 3.5 (out of 10). 38 The qualitative data in the current study suggest that most girls found the sessions physically exerting; some enjoyed this challenge, whereas others found it tiring. This raises the question of how appropriate this measure is for adolescents, as it appears that it may be misunderstood or ineffective in gathering accurate assessments of exertion. For example, the measure was taken at the end of the session and, therefore, required girls to reflect back over the entire session when reporting their level of exertion. Both girls and school contacts thought that the intervention had a positive impact on participants’ fitness, confidence to dance and mental well-being. This may have contributed to girls expressing an interest to continue dancing. Some suggested that they would like to take part if the BGDP was offered in the following year.

Dance instructors felt prepared to deliver the intervention as a result of the training. However, many instructors felt that elements were inappropriate given their significant experience. Having formed a social network as a result of the training, instructors found the mid-intervention booster session useful and valued the opportunity to reflect on their experiences. Similar findings were reported in the HEALTHY study. 96 The session-plan manual was considered a useful resource, although many found that it contained excessive detail (again, particularly for those instructors with a great deal of experience). This is likely to contribute to the varying levels of fidelity to the manual. The majority of instructors adhered to the manual in the earlier sessions, but as the intervention progressed they began to develop sessions based on what the girls wanted to cover. Fidelity was slightly lower than has been reported in other studies;90,95 however, this was expected to an extent, as instructors were encouraged to be autonomy-supportive and responsive to the needs of the girls.

A final point of note is that communication with, and support from, school contacts was highly valued by the dance instructors. Poor communication and/or an unwillingness by school contacts to help dance instructors made elements of managing sessions more difficult than if assistance was received. Such issues can be avoided if school contacts are aware of what is required of them.

Conclusion

This chapter reveals the major issues and influencing factors attached to the delivery of the BGDP. The decline in attendance, which in most part was affected by friendship groups and influenced by competing commitments, was a key factor. However, this resulted in smaller, more manageable, groups that the girls enjoyed. Different elements that girls enjoyed and did not enjoy were outlined, providing valuable information for future research. Fidelity to the session-plan manual was lower than in previous studies. However, the low level of session-plan fidelity may, in fact, signify high levels of theory fidelity and demonstrate that instructors were enabling the girls’ autonomy.

Dance instructor training and acceptance of the intervention theory (self-determination theory)

Dance instructors held positive views about the training and believed that the principles of SDT were appropriate to underpin the dance sessions. Most instructors believed that their existing teaching style conformed with the central tenets of SDT. They reported benefiting from the close alignment of their existing approach with that of the intervention.

A couple of those things [motivational techniques] come naturally anyway because I feel I encourage them to do events and give them the roles within their group . . . it’s something that I always do anyway.

Dance instructor 23

I think it’s what I would do anyway [teaching that aligns with SDT], but it’s always useful. And it also helps you to highlight and to actually see what you’re already doing.

Dance instructor 61

Some instructors found that aspects of their existing teaching style was challenged to some extent after completing the training. They described thinking about how to use vocabulary that was supportive of SDT during lessons.

You can get into a groove with how you teach and [the introduction of SDT] really made you challenge those sort of key phrases that you say throughout the class.

Dance instructor 23

So in terms of like how to make improvements, how to develop things. That made me think more to, you know, be careful of how I’m wording things.

Dance instructor 61

Observed dance instructor teaching practices

During four randomly observed sessions in each school (details of method and selection of sessions are discussed in Chapter 2, Process evaluation methods), an observer rated dance instructors’ teaching style on four teaching practices, namely, relational support, structure before and during the activity, and autonomy support. The results are presented in Table 42.

The most frequently observed teaching practice among all instructors was relatedness support. During the dance activities overall, instructors provided moderate structure before the sessions, and structure during the sessions was, on average, ‘sometimes observed’ in the sessions led by instructor 4. The least observed teaching practice was autonomy support.

The agreement between observers exceeded the statistical expectation (Table 43). The weighted Kappa matrix is presented in Table 44 and ICCs in Table 45.

Table 44 shows the weighting of scores between coders that was used in the calculation of the κ-statistic. Coders scoring the same number was weighted at 1.0, coders scoring one number apart was weighted at 0.75, coders scoring two numbers apart was weighted at 0.3 and coders scoring three numbers apart was weighted at 0.0. For example, when coder 1 gave a score of 0 and coder 2 also gave a score of 0, this was weighted at 1.0; when coder 1 gave a score of 0 and coder 2 gave a score of 1, this was weighted at 0.75.

Table 45 indicates that there was poor similarity between observers’ ratings for relatedness and fair-to-moderate similarity between ratings for autonomy, structure before and structure during sessions. 98,99

Need support and satisfaction

Compared with reports of fidelity to the dance session manual, the results suggests that dance instructor fidelity to the theoretical underpinnings of the project (i.e. adopting an autonomy-supportive teaching style and supporting girls’ needs for autonomy, competence and relatedness) may have been stronger. Instructors described attempting to deliver dance sessions in a manner that supported the psychological needs of the girls. Confirming this, girls’ experiences generally suggest a level of need satisfaction. However, there was greater diversity in girls’ experiences of this than the instructors’ reflections on providing need support. For a summary of the qualitative assessment of fidelity to SDT and potential areas for improvement, see Appendix 8.

Challenges of delivering a self-determination theory-based physical activity intervention for girls

Managing disruptive behaviour and the use of an end-of-project performance emerged as issues that challenged the instructors’ theoretical intervention fidelity. When faced with challenging behaviour or girls who ignored instructions, some instructors found it difficult to use SDT-based teaching practices.

When you’ve got 25 plus of them all going a bit mental [. . .] you do have to sort of change tactics unfortunately.

Dance instructor 42

Because we were quite happy to say ‘oh yeah’, like talk in a calm manner. But then after a while if the children aren’t responding to you in that way, to actually find solutions to that problem – but still using the ABC [autonomy, belonging, competence] technique – I think that’s what we kind of discovered in the [booster session].

Dance instructor 21 and 51

It would take them ages to get changed and things like that and because I was trying not to be in that ‘come on!’ role, I couldn’t really push that.

Dance instructor 23

In addition, one instructor saw the autonomy-supportive strategies as a firm set of guidelines that she had to abide by in order to deliver the session appropriately.

They were running wild and I was trying to be, you know, use the ABC and it was very hard to try and keep to that [. . .] I think that’s because I was so worried about sticking to ‘this is what we had to do’ to then kind of trying to actually respond to the children themselves.

Dance instructor 21 and 51

Some instructors suggested that role-play-based learning during the training would help them deal with such problems in ways that were consistent with the key themes behind SDT.

I would say it would be good to . . . set up some situations where that skill [teaching in an autonomy-supportive manner] could be practised, because I felt frustrated with myself sometimes that I didn’t know [. . .] what to say and I didn’t know how to do it.

Dance instructor 23

Another challenge to intervention fidelity concerned the use of dance performances (in front of other, non-group audiences) as a motivational tool and whether or not girls wanted to participate in these performances. Girls received a choice over whether or not they wanted to work towards and contribute to performances. The majority of the time, girls chose to perform. Dance instructors and school contacts considered an end of project performance to be important for motivating participation.

Things to work towards, so if you know that there’s a show [. . .] something like that will definitely keep them [attending].

Dance instructor 62

I think the performance element of it is a really vital part actually.

Dance instructor 61

So I think it (performing) does a lot for young people in terms of confidence and experience so I think it’s really important.

Dance instructor 32

I think as the girls highlighted maybe if they’re working towards something. So I know a lot of them when they do dance outside school they’re doing their exams and you know, their grades and so on whereas if they knew it was to do a big dance show at the end of the year or you know, a performance somewhere . . .

School contact 23

Instructors felt that most girls wanted a performance and that this was a positive experience for them; however, they also recognised that girls may be anxious about performing.

Maybe they don’t [want a performance] at the start because their confidence isn’t there yet, but by the end pretty much every young person I’ve worked with wants to perform.

Dance instructor 32

Even if they’re slightly petrified of it or whatever . . . it’s kind of a good fear [. . .] They did all love doing this as well.

Dance instructor 62

There were mixed views among the girls regarding a final performance. Some liked the motivation that it provided and felt that there could have been more opportunities for performing different routines that they had learnt.

If you know you’re not doing it in front of lots of people you kind of lack a bit. But if you know you’re doing it in front of people then you know that you’ve got to try and do your best and try and get the steps right.

Focus group 23

Yeah, I love going on stages and sort of showing off what we’ve been doing.

Focus group 32

However, for others, although a performance was considered a good experience, it was a source of anxiety and the support of the instructor was needed.

It’s kind of embarrassing but then it’s fun at the same time. Like everybody’s watching you in the audience.

Focus group 32

I felt a bit worried because I thought I was going to mess up.

Focus group 62

Some people like crying and really nervous and [saying] they weren’t going to do it.

Focus group 61

I think the thought [of a performance] for some of them might be too much.

School contact 51

They were really embarrassed to show their parents.

Dance instructor 21 and 51

Putting on a performance was a source of pressure for some, and one instructor suggested that it negatively affected her teaching. Some instructors found it hard to keep an environment supportive of SDT when working towards the performance.

When it came to rehearsal and things like that maybe not so much [participant ownership over the class].

Dance instructor 32

They’d got all these questions and all these different ideas but, you know, they also want to perform [. . .] you know you’ve got to choose something appropriate and it was difficult juggling all their different ideas.

Dance instructor 23

Motivation

The qualitative data regarding the girls’ motivation for (1) dance and (2) being physically active were similar to the quantitative data presented above (see Chapter 3, Psychosocial data), in that they revealed little indication of any change in motivation for dancing and being active. In this section we quantify the responses given by girls for some motivational types and give an almost exhaustive account of the data, in order to demonstrate its sparsity.

Statements made by two girls relate to an identified form of motivation. In both cases the girls expressed enjoyment at engaging in dance, as it is beneficial to them or keeps them from being ‘lazy’.

Sometimes, it’s like you’re put on the spot but I suppose that’s good if you’re going to do a lot more dancing things.

Focus group 53

If we’re not keeping active then you’re just gonna be . . . lazy.

Focus group 62

The physical activities and dancing were enjoyed by many girls because of how they made them feel.

After doing physical activities it makes you feel [. . .] happier, kind of. I kind of felt good about myself . . .

Focus group 53

In year six I did sports in the summer holidays. I kind of stopped and then when Active7 started I did sports again and [. . .] it’s just made me feel better because.

Focus group 53

There was one example of a girl viewing dance as ‘her thing’ (an integrated form of motivation) at the end of Active7.

One of my girls, she was really talented. Gifted and talented I’d say – really great rhythm. And she said to me at the end ‘I didn’t really used to like dance’. And that really surprised me because I think ‘but you’re really naturally quite, you know, gifted in that sense’. ‘I didn’t really used to like dancing but now I absolutely love it, it’s totally my thing’.

Dance instructor 72

Some instructors and girls believed that attendance was affected by girls feeling fortunate for being able to participate in the study (as places were restricted within each school).

They also understood that this was exclusive to them and there was nobody else in the school [. . .] so I think they really valued their place.

Dance instructor 23

We got chosen to do it when some people didn’t so it would be unfair if we didn’t go, for them, because they could have had our slot.

Focus group 32

External motivation to attend Active7 sessions was experienced in relation to parental encouragement/involvement, intervention components (attendance records and performance opportunities) and the adoption of forfeits by one instructor. One instructor felt that parents told girls that they must attend. Girls were also motivated to attend and indeed sign-up in order to socialise or through peer pressure.

With the star charts, how long we’ve been there . . . I felt because I’ve been off for so many weeks, I feel really bad that there are loads of crosses.

Focus group 32

I think their parents kind of told them to be there.

Dance instructor 21 and 51

I was telling my friend ‘Shall we do it?’ and she was like ‘Oh I don’t want to do it.’ I was like ‘It could be fun’ and she was just, she decided to go.

Focus group 21

Several girls, from two schools, attended to avoid feelings of disapproval from the instructor, a blend of external and introjected motivation. The following comments are interesting as they suggest that attendance was prevented by feelings of guilt.

I felt [dance instructor] would be really upset with me if I didn’t go and then I kind of stopped going because I was scared that she’d be angry with me every day I didn’t go.

Focus group 72

If we felt sick and we had to tell her she’d say ‘oh, try and join in but if not go sit down, but I would really, really like it if you would join in’. And like then she’d make us feel guilty.

Focus group 62

The types of amotivation mentioned included a lack of motivation or interest in completing dance diaries and attendance records, learning dances if girls had not attended the session when the dance was introduced and lack of interest in dance sessions generally.

I found it [completing dance diary] a bit annoying because I think we could have been doing more dances and maybe cooling down [. . .] we kind of got a bit stuck. Sometimes I didn’t know really what to say.

Focus group 32

Theory-based process evaluation summary

The data suggest that the intervention was delivered in a manner supportive of the concepts of SDT to a modest degree. Most of the instructors felt that the central tenets of SDT already aligned with their teaching style and were able to provide autonomy, competence and relatedness support through various aspects of the dance sessions. There were certain areas in which this was challenged, from both the instructors’ and the girls’ points of view. However, as the qualitative data attest, training was well received by the majority of instructors, who felt that the key principles of SDT ‘matched’ their style of teaching, and that the training served as a reminder to attend to the ‘how’ of their teaching (i.e. communication practices).

Autonomy support, as described by dance instructors, came mainly in the form of music choice, dance styles, warm-up activities and choreography. Dance instructors reported providing opportunities for participant choice, which was enjoyed by girls. Having choice over these elements made sessions more engaging and relevant to the girls. The level of autonomy granted was appreciated by girls and noticed by school contacts. However, provision of choice was predominantly option choice (e.g. choosing a type of music or a dance move) and some girls reported lacking action choice (e.g. not having control over the speed of task progression). Previous experimental research suggests that providing action choice promotes self-determination and intrinsic motivation more than option choice. 100 It is likely that option choice is easier to provide than action choice and this is perhaps why this was reported as the form of autonomy support more frequently used. However, instructors who relied only on providing option choices may have believed that this was sufficient autonomy support and may have neglected other elements such as action choice. Providing both action and option choices is likely to lead to greater autonomy than using one strategy in isolation.

In order to achieve autonomy need satisfaction among the girls, it was necessary for the instructors to provide guidance and instruction. Despite successful attempts to support autonomy, dance instructors sometimes found it difficult to do so owing to behaviour issues within the class and during the lead-up to the final dance performance. This was corroborated in the teaching style ratings which showed the structure during the activities to be modest. Behavioural issues resulted in the use of controlling teaching styles. Indeed, the opinions of some girls that their instructor was controlling suggest that not all instructors used an autonomy-supportive style and, thus, there were pockets of low theoretical fidelity. This result could reflect the instructors’ usual style which was unchanged by the intervention training. Alternatively, and in line with the dance instructors’ views that the SDT training aligned with their usual teaching style, adopting a more controlling style may be a response to the different challenges of teaching a large group of beginners in a school setting.

Experimental research suggests that action choice, that is, being involved in decision-making processes, rather than option choice, whereby an individual’s choice is restricted to selecting from a series of options, significantly affects internal loci of control, volition, perceived choice and intrinsic motivation. 100 On a practical level, this has utility in the design of programmes underpinned by SDT; providing autonomy in sessions by allowing individuals to have control over the timing, progression and differentiation of tasks, drills or performances may be one way in which to promote sufficient ‘action choice’. As such, certain choices are conducive to feelings of autonomy and competence, and to subsequent levels of intrinsic motivation.

Future work is needed to identify ways in which PA intervention deliverers, who may have used controlling strategies previously, can be equipped with techniques to use in response to challenging behaviour without resorting to controlling techniques. One instructor alluded to role play techniques that mimic SDT strategies for dealing with behavioural difficulties. This is one example of a potential strategy in the future implementation of theory-based interventions. Furthermore, a challenge for developers of theory-based interventions is to identify how best to teach intervention deliverers about the underpinning theoretical nuances (e.g. action vs. option choice) in ways that are easy to understand and implement in practice.

Dance instructors appeared to be confident in providing competence support and used a multitude of techniques that could be adopted in other interventions, such as using peer-to-peer teaching, facilitating self-reflection and providing genuine encouragement, and using differentiation to account for varied ability levels. This finding probably reflects the instructors’ experience in teaching dance techniques and highlights the importance of identifying well-trained intervention deliverers who can bring beneficial innovation. Indeed, instructors felt that they provided competency support via a number of methods, including differentiation of dance sequences, offering leadership opportunities and feedback, asking the girls to reflect on their own level of competence and the use of peer role models. Instructors were able to identify the girls’ different abilities, allowing them to carry out and deliver the sessions using SDT more effectively. Qualitatively, both instructors and girls felt that levels of competence and confidence in dance increased over the course of the project; girls reported positive experiences of having their competence supported and some reflected directly on their increased perceived competence. However, some girls felt that their instructor was unaware of different abilities and did not give them the attention they needed when learning new skills. Being able to contribute to the speed and pace of the sessions may have afforded the girls a greater sense of competence support and satisfaction. This finding is in contrast to the quantitative need satisfaction findings in which perceived competence in dance and PA decreased pre–post intervention in the intervention group. It is possible that both results are correct and that, although some girls did feel more competent after the intervention, on average, girls did not. Using qualitative and quantitative methods has allowed us to identify this.

It is important to recognise that girls’ baseline and follow-up reports of dance competence were based on different cognitive representations of dance for girls in the intervention group given their new experience. Before the intervention, most girls did not have experience of formal dance classes. However, they rated their dance competence as relatively high, which could be based on more informal experience (e.g. dancing with friends). Post intervention, having been exposed to some of the physical challenges and having been able to more clearly reflect on one’s competence compared with that of others, the girls are likely to have had a different perspective on dance. Thus, girls may interpret things differently before and after the intervention and may rate their competence post intervention as lower than at baseline. This could have masked some of the more positive findings identified in the qualitative results.

Every dance instructor attempted to form trusting relationships with and between girls. They asked about girls’ lives outside of class, responded to feedback in dance diaries and took time in class to reflect on sessions. Girls within some focus groups did not feel a connection with their instructor, describing them as strict or not genuine. Only a few girls reported lacking a connection with others; this was thought to be down to differences in ability and confidence levels. Girls would often focus on upwards comparison, which in some circumstances can be beneficial when it comes to enhancing competence levels; however, as the evidence here demonstrates, the use of role models must be approached carefully.

Previous research suggests that relatedness towards PE teachers is associated with engagement in PE among high-school-aged girls (15- to 18-year-old girls). 101 The qualitative data showed that dance instructors were, in most cases, able to build trusting and meaningful relationships with and between the girls through a range of techniques, representing excellent intervention innovation102 which could be adopted in other interventions. However, for some girls, relatedness towards their instructor was undermined by perceptions of the relationship being superficial or not being treated with respect. Given the variation in instructors’ provision of relatedness support, peer-based learning, where effective techniques are shared with instructors who find creating a meaningful bond more difficult, could hold promise. Although this was included to an extent in the BGDP mid-intervention booster session, more could be done to share effective practice (either in person or via social media interactions) and challenge low fidelity during the intervention period.

Autonomous motivation was represented by some of those girls recognising the benefits of engaging in dance and inherently enjoying the sessions. Forms of controlled external motivation were evident in forms of parental encouragement, attending to maintain one’s attendance record and one instructor’s use of PA-based forfeits for talking. Some girls described feeling guilty if they missed sessions or would attend to avoid disapproval from their instructor.

Based on these findings, it is evident that dance instructors may need more time in training to become familiar with the theoretical components of the intervention. Incorporating role play opportunities to put the theory into practice, particularly in situations that prove challenging (e.g. misbehaviour), may be effective. Allowing and organising more time for dance instructors to meet and share experiences and ideas would provide them with more opportunities for peer learning. Future work should ensure that dance instructors know how to manage performances outside of the class with SDT and how it translates to programmes encouraging PA.

Conclusion

The data presented in this chapter pertain to the theoretical fidelity of the BGDP intervention. Recommendations for the future delivery of interventions based on theory are also presented. Together, the quantitative and qualitative data suggest that there were some areas of very strong theoretical fidelity and that the intervention ‘tone’ underpinned by SDT was accepted by both girls and instructors. However, the findings also reveal a number of gaps and room for improvement in theoretical fidelity within the intervention. As such, a number of recommendations are put forth which may have utility in the future design and implementation of theory-led interventions. Overall, in line with the views of Rothman,44 basing interventions such as BGDP in well-articulated theory, translating the theory to practical intervention activities and subjecting it to thorough evaluation has resulted in the advancement of SDT-based PA interventions in trials.

Project design

As a theme, ‘project design’ encompassed subthemes relating to the logistical arrangements of BGDP, namely, recruitment, timetabling, session quantity and project duration.

Session content

This section relates to the combination of a number of subthemes that were concerned with the delivery of sessions, variety in session content, repetition, group work, dance diaries and behaviour.

Project organisation

This section relates to ideas for improving attendance, session design and project organisation. The section is split into six subthemes: open enrolment, parental involvement, working towards end goals, cost, facilities and communication issues.

Summary

This chapter has reflected on the various issues associated with the implementation of BGDP and has highlighted some implications for the delivery of similar interventions in the future. It is hoped that the various recommendations will have utility in the design, implementation and success of future trials and after-school PA interventions in general. A summary of these recommendations can be found in Table 47.

Various methods of recruitment for each participant group are recommended (see Recruitment), reflecting the diversity of girls and their requirements. Research altruism represents a potential lever with which to encourage school contact involvement; in light of this, it may be worth dedicating more space to recruitment materials that highlight the academic need for and merit of the research. Future projects should ensure school contact familiarity with the girls, as this could ease issues surrounding increased workloads and teaching burdens. Recruiting dance instructors based on endorsements from other dance instructors and dance agencies was also recommended as a fruitful method of recruitment. It was useful to recruit multiple reserve instructors from different areas, as they can then provide necessary support over a wider geographical range. Targeting peer groups among the girls (to aid sign-up and maintain attendance), espousing the credentials of the project to instil a type of research ‘privilege’, and using vouchers sparingly were cited as sensible and realistic methods for recruiting girls.

Timetabling of sessions (see Timetabling) needs to consider other after-school activities and the requirements of girls. Such an approach would counter issues associated with projects overlapping with existing extracurricular activities. Session quantity and project duration (see Session quantity and duration and Project duration) were viewed as long, and a more targeted approach whereby only one session per week is delivered in 5-week ‘blocks’ was considered a sensible modification.

Given the significance of dance styles in influencing participation, it was suggested that future interventions should encourage participant ‘choice’ in dance styles (see Session content). This could be facilitated through a participatory approach, whereby girls are empowered to promote their ideas and suggestions in a way that supports their engagement with the project. Training dance instructors in dance styles with which they are unfamiliar is also encouraged, as qualitative findings suggested that instructors were not familiar with some styles and the methods for teaching them prior to the intervention. Furthermore, the repetition of content discussed above (see Repetition) was found to limit the enjoyment of sessions, and it is recommended that measures are installed to ensure that girls who miss classes are familiarised with missed content. This could include implementing a more rigorously applied ‘buddy’ system.

As seen above (see Group work), group work is to be encouraged, as it enabled girls to feel like they had ownership of the project, helped the instructor to cope with various levels of competence within the group, developed girls’ leadership skills and was helpful with inconsistent attendance. Owing to the lack of consensus regarding the use of diaries (see Dance diaries), future projects may wish to clarify why they are needed. A collaborative method whereby dance instructors define boundaries regarding acceptable behaviour and language in classes was seen to be useful for managing behaviour (see Behavioural difficulties and management techniques). Maintaining communication and contact between dance instructors and school contacts is encouraged, as is a positive support network among instructors (see Communication and management issues).

An open enrolment method (see Open enrolment), whereby girls can ‘drop in’ to sessions, was viewed as a potential modification to project organisation that would positively affect attendance. Although restricting the number of girls was a necessary trial design measure, it does not reflect usual educational practice. Increased contact between the trial team and parents – who were significant in directing attendance – was viewed as a sensible method for retention (see Parental involvement). This could be achieved via information workshops or the dissemination of information to parents. Implementing payment for sessions should be approached with caution.

Although it is often not possible for intervention designers to ‘choose’ facilities within schools, it was suggested that being able to respond to participant concerns regarding facilities and act upon them is useful (see Facilities). It is important to identify what the optimal and realistically available facilities will be and how these can be adapted for PA promotion. This is a salient finding, given the quantity of responses from girls who suggested that facility issues hindered their engagement with the intervention.

Conclusion

The data presented in this chapter have shown how future school-based PA interventions could learn from the BGDP experience. We have provided recommendations that may be of use in future after-school PA interventions. Although the findings provide evidence that after-school interventions have the potential to increase PA levels among adolescent girls, the qualitative data suggest that there is a need to tailor the design and implementation of sessions to individual schools.

Summary of findings

In this report we have outlined the results of the BGDP. As stated in Chapter 1, there was a strong rationale for developing dance-based after-school interventions for adolescent girls. Dance is a desirable activity for adolescent girls and, when delivered on the school site immediately after school, it can be an accessible activity. Offering dance sessions for free also avoids any financial barriers to participation. We therefore hypothesised that providing dance sessions would yield increases in girls’ MVPA.

The main objective of the BGDP was to investigate how effective after-school dance sessions were at increasing the PA levels of Year 7 girls. Both intention-to-treat and per-protocol analyses found no evidence of a difference in weekday MVPA at T2 or T1 between those girls who were assigned to the intervention group and those assigned to the control group. However, an additional exploratory analysis discovered that the dance sessions did provide girls who attended with 4.6 minutes more of MVPA and 14.3 minutes more of light PA than on a day on which there were no dance sessions. At T2 evidence was found to support a difference in motivation scores (excluding autonomy need satisfaction and self-esteem) between groups, with higher scores among girls in the control group. The evidence followed a similar pattern at T1.

The good recruitment rate of schools (compared with the sample population) indicates that schools are keen to be involved in dance interventions, thus suggesting that dance is an appropriate activity for adolescent girls. Pupils and dance instructors were also interested in the study, as demonstrated in the strong pupil recruitment rates in schools and the willingness of local dance instructors to be involved. However, it is important to note that the girls who were recruited to the study were, relative to the wider population, an already active sample. Dance instructors found the intervention training and session-plan booklet to be useful resources and, on the whole, felt that they developed professionally as instructors as a result of this exposure. There is scope for the refinement of the intervention materials, in particular in terms of ensuring that the materials cater for wide-ranging skills and experience.

This study shows that it is possible to deliver a relatively long (6–7 months’ duration) PA intervention within secondary schools in the after-school period. All intervention schools received at least 37 of the planned 40 intervention sessions (undelivered sessions were largely attributable to illness or competing school events). Many stakeholders felt, however, that the intervention period was too long to maintain interest and enthusiasm. A major consequence of this was seen in the declining attendance rates throughout intervention schools. Attendance was low at the start of the intervention and declined over time. Although this is common to many after-school activities, it does raise concerns about the potential of the after-school period for engaging children in long-term interventions. This may be an appropriate time at which to consider an increase in PA, as schools provide a setting in which all children can be reached. In addition, more attention may be required to encourage better attendance in future studies. This might be achieved through, for instance, more formalised agreements with schools or pupils.

Regarding the length of intervention sessions, views were more ambiguous. Many interview respondents felt that sessions were too long and that any future interventions should be reduced in length. This is similar to the feedback from our earlier pilot work,38 which found 90-minute sessions to be too long, hence the reduction in session time in the current study. A similar issue emerged vis-à-vis the timing of sessions, with respondents expressing largely positive views on sessions beginning immediately after school (as our pilot work suggested), but some maintaining that they needed more time to transition between school classes and the start of Active7 sessions. However, the end time of sessions was often considered too late, especially in winter months. Although the results were disappointing, with attendance and fidelity to the session-plan manual being lower than hoped, we feel that a modified after-school PA intervention does have the potential to increase interest in physical activities and improve child PA levels.

Active7 dance sessions were enjoyed by the girls who attended. The elements of the intervention that were disliked included having to readdress content because of pupil absences, disliking certain dance styles and having mixed views about performing to a group. The enjoyment data collected from girls suggest that when girls did attend, they had a good experience. This, combined with qualitative data, suggests that the reasons for girls missing sessions were not attributable to disliking the content. In fact, missing sessions was largely attributed to competing activities or responsibilities outside school (family and social commitments were often cited) that would impact upon any other commitments girls had. In addition, as the intervention period spanned a 6-month period, the commitments that girls had changed. The self-reported exertion data collected towards the end of the dance sessions indicate that girls did not find the sessions physically demanding. Although this is not entirely corroborated by the qualitative data, it is an issue that would need to be addressed in future interventions.

The dance intervention is likely to appeal to schools, as it is financially viable and can be delivered relatively cheaply. The average cost was £2401 per school, with an average of £73 per pupil over the 40 sessions (£1.83 per child per session). In a context in which schools purchase external provision for many after-school activities, the findings are likely to be appealing. Approximately 30 children can participate in an enjoyable activity all year round (indoor dance sessions are not affected by weather variations).

A summary of findings from the BGDP will be disseminated to a number of stakeholders after the main trial findings have been published. Head teachers and school contacts in all schools (intervention and control) will receive a letter summarising the trial findings (see Appendices 9 and 10). All girls who participated in the study will be given a ‘thank you’ postcard (see Appendix 11) and parents/guardians will receive a letter (see Appendix 12) summarising the study results. All schools will be sent a poster presenting the findings (see Appendix 13). We have included a list of local dance activities on our website, which have also been signposted on the dissemination materials (see www.bristol.ac.uk/sps/research/researchprojectpages/active7/dance/).

Interpretation

Physical activity interventions have been found to achieve small to negligible effects in children’s PA, with some improvements to levels of MVPA. 36 Interventions targeted at increasing girls’ PA in particular have had similarly small effects (p< 0.001). 13 Biddle et al. 13 have suggested that, although behaviour change is possible for female pre-adolescent populations, it is difficult. This conclusion was recently supported by Lubans et al. ,15 who also reported that an intervention did not increase PA levels of their adolescent sample. The present study recruited a relatively active sample of girls. Median baseline weekday MVPA was 49.15 (control) and 53.25 (intervention). This issue has been noted in several other studies that have found that active participants tend to sign up to PA interventions, thus indicating that the individuals most in need of the intervention – those with low PA levels – are not being recruited. This not only has the effect of ‘missing’ those individuals with low activity levels who are of primary public health concern but may also have a secondary effect of limiting any improvements in the MVPA levels of participants.

The effect of the BGDP intervention on MVPA can be seen as similar to effects reported in the wider literature. The only notable effect detected was for those who attended dance sessions while wearing an accelerometer at T1 (during the dance intervention). There was evidence of an increase in levels of light PA and MVPA on dance days compared with non-dance days for these girls. O’Neill et al. 27 also found increases (p < 0.001) in levels of light PA and MVPA on days on which children attended their dance sessions. Results are comparable to the GoGirlGo! study, in which 5- to 13-year-old children accumulated 11 minutes of MVPA on PA session days compared with 8 minutes of MVPA on non-activity days. 103

As we found girls to be more active on the days on which they attended dance sessions, we suggest that the after-school sessions do not offset other PA but rather provide a new way in which girls can be active. An important point is that we were unable to detect the effect of the intervention on school travel mode throughout the intervention period. The T1 measures (which provided the accelerometer-assessed measure of PA during the intervention period) were taken during a summer month (with longer hours of daylight); research has found that independent mobility of children during winter months is somewhat lower than during summer months, suggesting that the girls had more options for travelling home from school. 104 In addition, although the findings suggest that sessions need to incorporate more opportunities for MVPA, it is important to note that the T1 measures were taken at a period in which most intervention schools were preparing routines to showcase. Thus, the content of sessions was likely to be focusing on the minutiae of dance sequences. This would result in increased amounts of light and sedentary activity than would be expected from a typical session. An additional factor that may have curtailed the effect of the intervention was the high baseline levels of MVPA in both trial arms (mean intervention group 53.25 minutes; control group 49.15 minutes). Other studies have found that increasing PA in an already active population is difficult. 39,105,106 In future, it is clear that the sessions need to incorporate higher levels of high-intensity PA in order to increase the MVPA levels of participants.

Contrary to much evidence,43 Biddle et al. ’s13 meta-analysis found stronger effect sizes in atheoretical studies. Noting that this was not anticipated, the authors of the review contend that behaviour change interventions may be successful without a theoretical underpinning. The qualitative results presented above suggest that the participants received positive experiences of competence and greatly valued the autonomy granted to them during sessions. The type of choice that participants received in Active7 sessions can be seen to be largely that of the ‘option’ as opposed to ‘action’ variant. 100 Option choice relates more to issues such as music and dance move choice. Action choice, however, focuses more on progression and choice over the speed and complexity of tasks. As the latter variant may be more directly associated with PA levels, a greater focus on action choice would probably promote more self-determined and intrinsic motivation among participants,100 which may have a greater effect on PA levels. A more detailed understanding of the type of choice granted could be explored in the lengthened training sessions that those delivering the intervention sought. As the intervention was not successful in increasing PA levels, we were unable to conduct mediation analyses that examined the effect of the principles of SDT in greater depth. The intervention did not have the expected effects of increasing intervention arm participants’ perceptions of autonomous motivation or perceptions of psychological need satisfaction.

The findings suggest that those delivering the intervention adhered more to the theoretical underpinning of the study (SDT) than to the session-plan manual guiding the structure of sessions. Compared with other studies (with > 70% of delivered sessions only ‘somewhat’ or ‘not at all’ like the manual), the fidelity to the session-plan manual was low. 90,95 Ironically, this can again be interpreted as demonstrating a greater adherence to the theoretical underpinnings, whereby instructors were encouraged to provide choice and allow the girls to take ownership of the content of sessions. Although it appears contradictory that session-plan fidelity was low and theoretical fidelity high, we contend that this is not particularly surprising, as skilled and experienced dance instructors adapted sessions to suit the needs and wishes of the participants and attempted to provide choice for girls. However, theoretical fidelity was variable and a number of factors (e.g. behavioural issues, misinterpretation of the theory, using performance pieces and intervention drift) undermined delivery, which was consistent with an autonomy-supportive style. This said, instructors reported using many need-supportive teaching styles which were received as intended by the participants, and these ideas can be incorporated into future interventions.

The process evaluation found that several factors encouraged girls to participate in the study. The reasons for participation were largely social (making new friends and building existing relationships) and dance specific (learning new styles, trying something new). Developing social relationships was important for not only recruitment but also attendance and retention. This supports the contention that children are more likely to remain engaged in PA if they have good social ties within a given activity. 107,108 As reported by dance instructors, after the initial settling-in periods, a core group of participants who attended regularly remained, and these groups pivoted around strong social networks. In terms of the decline in attendance, the qualitative research is particularly insightful. Corroborating the views of the BGDP participants, who suggested reducing the overall length of the intervention period, research13,34 has found that studies were more effective when the intervention period was short (< 12 weeks). A long intervention period was considered potentially boring and unsuitable to young people; again, this was a view supported by the focus groups and by interviews.

Similarly to the findings of previous research,94,95 participants enjoyed the after-school sessions. Exposing girls to activities that they find enjoyable is crucial, as girls are more likely to participate in PA over the longer term if they find the activity enjoyable. 109 The high levels of enjoyment, combined with the largely extraneous reasons for not attending sessions, implies that the intervention was designed and delivered in an appropriate manner for the girls. If greater retention and consistent attendance can be achieved, the impact on MVPA could be significantly greater. The process evaluation highlighted some important factors that, if incorporated appropriately in future interventions, could maximise the relevance to the target audience.

The intervention was relatively inexpensive compared with similar PA interventions. 110,111 As such, dance-based after-school activities should be considered by school management as a way to enhance extra-curricular activities. The average cost per pupil was £73 for 40 sessions (range £68–77). Thus, if following the recommendation of reducing the length of the intervention period, the financial cost could be significantly less. As the intervention was not successful in achieving the primary or secondary outcomes, we cannot comment on the cost-effectiveness of the intervention. We found that the EQ-5D-Y was ineffective as a quality-of-life measure for the healthy child population in this study. We suggest that future research explores different quality-of-life measures for such populations.

Implications for future research

Dance appeals to girls and schools and is enjoyed by adolescent girls. It can be considered an effective means by which to initiate secondary school-aged girls’ engagement in PA. Future research could test whether or not dividing the intervention into a series of smaller blocks (i.e. five sessions), each focusing on different dance styles, appeals more to participants and whether or not it can improve retention and attendance rates. Important findings from the process evaluation include: adopting an open-enrolment policy should be considered; behavioural guidelines should be collaboratively instigated by the intervention deliverer and participants; group work should be encouraged; greater theoretical fidelity through peer-to-peer support and a greater training provision should be facilitated; there should be a reduction in the repetition of content if children are absent (this would be eased by introducing ‘blocks’ of sessions), as well as a reduction in the overall intervention length and individual session length. Future studies may wish to better control delivery across schools to ensure that the intervention is delivered in a consistent manner across sites. Table 48 provides details of some of the lessons learnt during the delivery of the BGDP intervention and offers potential solutions to increase the effectiveness of future interventions.

Strengths and limitations

This study has been a systematic and rigorous trial of a PA intervention in secondary schools. The study had sufficient statistical power to detect any effect on the primary and/or secondary outcomes. Rates of both school and girl recruitment were high, and there was excellent retention of girls to the study. All schools were retained to the study. We collected objective assessments of PA from participants at three time points, providing sufficient data to detect any long-term intervention effect. We also conducted an extensive process evaluation, which provided both qualitative and quantitative assessments of the intervention delivery and reception from a range of stakeholders. The intervention has been reported in accordance with the TIDieR framework and results have been reported in keeping with the CONSORT reporting guidelines (with the CONSORT flow diagram also reported; see Figure 2). As the study was free to attend and delivered in state schools, there was no financial barrier to participation (at the school or participant level). In addition, all girls able to participate in PE sessions were able to take part and there were no skill-specific prerequisites for participants.

The study was limited by the focus on one geographical region in the south-west of England. Participants were an already active sample. This issue not only affects the results of the study but is indicative of a wider problem with PA interventions, namely, that recruiting participants with low activity levels (arguably those most in need of an intervention) is difficult. The immediate effect of the intervention upon PA levels is largely limited to self-reported exertion levels and a one-off accelerometer assessment of those attending a session at T1 (the number providing data is low). An objective measure, such as heart rate monitoring, may have provided a more accurate reflection of within-session exertion levels. Anecdotal evidence from staff collecting data suggests that there was some degree of peer influence in the reporting of exertion levels and psychosocial questionnaire responses. We were largely unable to establish any intervention effect on school travel mode as girls were not asked about how they travel to and from school at any time point. As other research has suggested,39 the high baseline levels of MVPA may indicate that the study appealed to an already highly active group of adolescent girls, and, therefore, those most in need of a PA intervention did not sign up.

Our qualitative data and informal conversations with dance instructors delivering the intervention suggest that those involved in delivering sessions received a number of benefits. A peer-support network was established as a result of the instructors’ engagement in the study. The largely unanticipated positive effect on the local dance instructor network is a significant impact of the study, but no objective data were recorded to capture this effect. Future studies may seek to assess the impact on those delivering the intervention more rigorously. A more sustainable approach to improving PA opportunities may be through the upskilling of teaching staff (as opposed to hiring external providers), so that schools can deliver dance activities themselves. For present purposes, this was not feasible owing to the cost and expertise necessary for the dance-specific training that would have been required.

Contributions of authors

Russell Jago (Professor of Paediatric Physical Activity and Public Health) was the principal investigator on the grant funding the work, and Simon J Sebire (Senior Lecturer in Psychology of Physical Activity/Exercise), Ashley R Cooper (Professor of Physical Activity and Public Health) and Jane E Powell (Professor of Public Health Economics) were coapplicants.

Russell Jago, Mark J Edwards (Research Associate, Trial Manager), Kathryn Banfield (Research Assistant, Physical Activity Interventions among Youth) and Thomas May (Research Assistant, Qualitative Research) wrote the first draft of the monograph.

Simon J Sebire and Joanna M Kesten (Research Associate, Qualitative Research) were responsible for the process evaluation and completed first drafts of the process evaluation results chapters. Joanna M Kesten collected all process evaluation data. Russell Jago, Mark J Edwards, Simon J Sebire, Joanna M Kesten and Thomas May analysed the qualitative process evaluation data. Ashley R Cooper provided input into the design and conduct of the study, and critically reviewed and edited the monograph.

Emma L Bird (Senior Lecturer in Public Health) and Jane E Powell were responsible for the economic evaluation design and analysis.

Keeley Tomkinson (Research Assistant in Medical Statistics) and Peter S Blair (Reader in Medical Statistics) led the statistical analysis and reporting of the trial, and served as the main links to the Bristol Randomised Trials Collaboration, a UKCRC-Registered Clinical Trials Unit.

All authors made contributions to sections of the report and critical comments on drafts of the monograph and approved the final submission.

Publications

Edwards MJ, May T, Kesten JM, Banfield K, Bird EL, Powell JE, et al. Lessons learnt from the Bristol Girls Dance Project cluster RCT: implications for designing and implementing after-school physical activity interventions. BMJ Open 2016;6:e010036.

Jago R, Edwards MJ, Sebire SJ, Cooper AR, Powell JE, Bird EL, et al. Bristol girls dance project (BGDP): protocol for a cluster randomised controlled trial of an after-school dance programme to increase physical activity among 11–12 year old girls. BMC Public Health 2013;13:1003.

Jago R, Edwards MJ, Sebire SJ, Tomkinson K, Bird EL, Banfield K, et al. Effect and cost of an after-school dance programme on the physical activity of 11–12 year old girls: The Bristol Girls Dance Project school-based cluster randomised controlled trial. Int J Behav Nutr Phys Act 2015;12:128.

Sebire SJ, Edwards MJ, Kesten JM, May T, Banfield K, Bird EL, et al. Process evaluation of the Bristol Girls Dance Project. BMC Public Health 2016;16:349.

Sebire SJ, Kesten JM, Edwards MJ, May T, Banfield K, Tomkinson K, et al. Using self-determination theory to promote adolescent girls’ physical activity: exploring the theoretical fidelity of the Bristol Girls Dance Project. Psychol Sport Exerc 2016;24:100–10.

Data sharing statement

All available data can be obtained from the corresponding author.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the PHR programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the PHR programme or the Department of Health.