Search:
Cahill, S., Thornton, E., Humphrey, N. (2026). Measuring predictors and prevalence of benevolent childhood experiences in English adolescents: Evidence from the #BeeWell study.
This paper is currently under review which means it is being considered for publication. Adolescence is a time when young people are exposed to a range of difficult life experiences (adversity) but also have the chance to build on positive ones, such as supportive relationships and a sense of belonging. We were interested in how these positive experiences, known as Benevolent Childhood Experiences (BCEs), shape adolescent wellbeing over time, particularly in the face of adversity. We used three waves of #BeeWell data covering 18,616 young people in 136 secondary schools across Greater Manchester, following them from Year 8 (age 12–13) through to Year 10 (age 14–15). Cumulative adversity was measured in Year 8 using 13 indicators of difficult experiences (e.g., bullying, free school meal eligibility, discrimination, feeling unsafe). Emotion regulation (a young person's ability to manage their feelings) was measured in Year 9, and wellbeing in Year 10, while taking baseline wellbeing in Year 8 into account. We found four main things. First, adversity in Year 8 was linked to lower wellbeing in Year 10 mostly through its effect on emotion regulation in Year 9, rather than through a separate direct route. Supporting young people's ability to manage their emotions therefore looks like a high-leverage way to soften the impact of adversity. Second, BCEs (supportive relationships at home, school, and in the community, alongside a sense of self-worth and hope for the future) directly raised both emotion regulation and wellbeing at every level of adversity. They are valuable assets in their own right. Third, contrary to the long-standing assumption that positive experiences protect young people from the harms of adversity, we found that adolescents with more BCEs were, if anything, more affected by adversity, not less, particularly when those experiences came from supportive parents and carers. Strong school relationships worked differently, lifting wellbeing for everyone without changing how adversity translated into outcomes. Fourth, although the same pathways operated for adolescents from sexual and gender minority backgrounds and from minoritised ethnic groups, minoritised young people were exposed to substantially more adversity in the first place. These findings suggest that investing in positive childhood experiences improves wellbeing across the whole adolescent population, but they cannot replace the upstream task of reducing the adversity young people face, particularly the identity-related adversity that disproportionately affects minoritised groups.
Cheng, Q., Thornton, E., Knowles, C., Humphrey, N. (2026). Adolescent internalizing symptoms, school experiences, and attendance: sex-specific within-person pathways. Journal of the American Academy of Child and Adolescent Psychiatry.
In this study, recently published in the Journal of the American Academy of Child and Adolescent Psychiatry, we followed over 25,000 secondary school pupils in Greater Manchester over three years to explore how emotional wellbeing and school life influence one another over time. Focusing on internalising symptoms such as anxiety and low mood, we found that when adolescents experienced higher-than-usual emotional difficulties, they were more likely to feel less connected to their school community later on. We also found that these emotional challenges predicted poorer future relationships with school staff for boys, and lower overall attendance (driven specifically by authorised absences) for girls. Importantly, our research showed that positive school experiences can also help protect mental health, although in different ways for boys and girls: strong staff relationships appeared especially beneficial for boys, while a broader sense of school belonging was particularly protective for girls. Notably, changes in a student’s physical attendance did not appear to drive later changes in their emotional difficulties. Overall, our findings suggest that schools can best support adolescent mental health by fostering environments where students feel connected, supported, and understood, rather than focusing only on whether they are physically present in the building.
Khanna, D., Panayiotou, M., Black, L., Demkowicz, O. & Humphrey. N. (2026). The effects of learning-related and peer-related school experiences on adolescent wellbeing: a longitudinal structural equation model. Children and Youth Services Review.
Schools are complex environments, with multiple experiences occurring within them at the same time. To improve students’ wellbeing, it is important to understand which of these experiences matter, and whether certain areas of school life – academic or social – are more or less strongly associated with wellbeing. This paper considers the effects of three learning-related school experiences (happiness with attainment, careers education, staff-student relationships) and four peer-related school experiences (school belonging, friendships and social support, relational bullying victimisation, physical bullying victimisation) on later wellbeing. Using a sample of 25,249 young people in Greater Manchester over two years, we fit a longitudinal structural model, an approach which allows us to account for the correlation between different school experiences with one another as well as account for prior wellbeing levels. We find that school belonging is most strongly related to wellbeing outcomes while staff-student relationships and physical bullying victimisation do not have significant effects. This has implications for which school experiences future structural interventions are targeted towards to improve students’ wellbeing.
Parker, G., Jacob, J., Deighton, J. (2026). How do UK secondary schools support young people’s wellbeing? Using public involvement to develop a school provision survey. Journal of Participatory Research Methods.
Schools are crucial in the prevention and early intervention of mental health disorders, yet evidence on school provision to support pupil wellbeing is hard to capture accurately. This includes the wider aspects of school provision that, whilst often not categorised as mental health and wellbeing provision, often has the potential to improve pupils’ mental health and wellbeing. This research aimed to design and implement a school provision survey in collaboration with stakeholder, school staff, and young person groups using patient and public involvement (PPI) methods, to capture how schools within the #BeeWell programme support children and young people’s wellbeing. Targeted and embedded consultation methods were conducted with these groups from January to October 2024, who provided input on survey conceptualisation, design, and implementation. The final survey was rolled out to schools in October 2024. The PPI impact on the research, researcher, and individuals involved was also measured. This work resulted in PPI having a large impact on the creation of a #BeeWell school provision survey and demonstrates how PPI can be meaningfully captured within a practical timeframe to collect data on secondary school provision.
Thornton, E., Mills-Webb, K., Irizar, P., Knowles, C., Marquez, J., Humphrey, N. (2025.) Substance-general and substance-specific influences on adolescent vaping, smoking, alcohol consumption, and illicit drug use: context, inequalities, and determinants.
Substance use before the age of 15 is a key risk for requiring substance use treatment later in life. Adolescent’s lives have changed considerably in recent years, potentially creating new patterns of risk and protective factors. Using data from #BeeWell, we looked at over 30, 000 young people aged 12-15 years across Hampshire, Isle of Wight, Portsmouth, and Southampton, to understand what puts young people at risk of vaping, smoking, drinking alcohol or trying illegal drugs. We compared the influence of both schools and neighbourhoods, and then explored which individual and social factors mattered most. Using a statistical technique called cross-classified multi-level modelling, we found that schools played a bigger role than neighbourhoods in explaining differences in substance use. We also found inequalities between different groups of young people. Some factors (for example, relationships with friends, family and teachers, and using mental health support at school) were linked to several types of substance use. Other factors were more substance‑specific, for example, feeling low or being bullied did not predict trying illegal drugs. Overall, our findings highlight that prevention efforts should recognise both shared and substance-specific risks, account for inequalities between groups, and make good use of school environments as a key setting for support.
Farzinnia, R., Black, L., & Humphrey, N. (2025). A cross-lagged panel network model to document the relationships between psychosocial risk factor exposure and internalising symptoms in early-to-mid adolescence.
This paper is currently under review, meaning it is being considered for publication. Adolescence is a time of rapid physical, emotional, and social change, and many young people begin to experience mental health difficulties during this period. However, research often examines individual risk factors and internalising symptoms separately, rather than considering multiple factors and how they influence each other over time. Using data from nearly 27,000 young people aged 12–15 in Greater Manchester who took part in the first three waves of the #BeeWell project, we applied a statistical approach called a cross-lagged panel network model. This method allowed us to examine how everyday difficulties (such as being bullied and feeling stressed) and internalising symptoms (such as worry and crying) were connected both at the same time point and across time. We found that these difficulties were closely linked and tended to reinforce one another over time; for example, increases in one problem were often followed by increases in others. Some patterns differed when looking at changes within individuals compared to differences between individuals, showing that population-level trends do not always reflect individual experiences. Overall, our findings suggest that adolescent mental health is shaped by many interconnected and mutually reinforcing factors. This highlights the importance of prevention and support programmes that address multiple risks at the same time, rather than focusing on a single issue in isolation.
Knowles, C., Thornton, E., Humphrey, N. (2025). Mapping the missed days: Latent patterns of attendance and their predictors in English secondary schools.
This paper is currently under review, which means it is being considered for publication. Why do some young people start missing school while others keep attending every day? Using #BeeWell survey data linked to school attendance records for more than 28,000 pupils in Hampshire and Southampton, this study tracked how attendance changed across the 2023–24 school year. Most pupils (about 9 in 10) attended school regularly throughout the year. However, four distinct patterns emerged. A second smaller group had persistently low attendance across all terms, a third showed a decline as the year went on, and a fourth group started with poor attendance but improved over time. These patterns matter because missing school is strongly linked to poorer mental health, lower attainment and fewer opportunities later in life. Young people were more likely to struggle with attendance if they lived in more deprived areas, had special educational needs, poorer mental or physical health, or felt they did not belong at school. Feeling under pressure from schoolwork, being bullied, or experiencing emotional difficulties also increased the risk of falling into lower-attendance groups. The findings show that attendance is shaped by much more than rules or motivation alone. Building a strong sense of belonging at school and offering early, targeted support to pupils who are struggling with wellbeing or pressure could help prevent long-term absence and support young people to stay engaged in their education.
Cahill, S, Vagnetti, S, Knowles, C, Thornton, E, Cheng, Q, Humphrey, N. (2025). Breaking point: identifying the factors that predict suspension from school.
This paper is currently under review, which means it is being considered for publication. This study explores the factors linked with school suspensions in order to understand who is most at risk and why. Suspensions interrupt learning and can harm young people’s wellbeing and future opportunities. Although behavioural issues are often seen as the main cause, this research shows that suspensions are shaped by a wider mix of personal, social and school-level influences. Using linked #BeeWell survey data and official school records for nearly 20,000 pupils aged 12–15 in Hampshire and Southampton, we looked at how family disadvantage, wellbeing, relationships, behaviour and school practices relate to suspension. Pupils eligible for free school meals, those with special educational needs and those living in more deprived areas were all more likely to be suspended. Pupils with strong feelings of belonging at school, positive relationships with staff, satisfaction with schoolwork and good emotional skills were less likely to be suspended. Unexpectedly, pupils reporting emotional difficulties or high school pressure were less likely to be suspended, perhaps because distress can show up as disengagement rather than disruptive behaviour. Pupils who had used substances or experienced bullying were more likely to be suspended. Importantly, suspension was more common in schools where suspensions overall were already frequent. The findings suggest that reducing suspensions requires more than discipline alone. Strengthening school belonging, improving relationships between pupils and staff, supporting emotional skills and addressing disadvantage may help prevent pupils from reaching the “breaking point” that leads to exclusion.
Cahill, S., Thornton, E., Humphrey, N. (2025). Measuring predictors and prevalence of benevolent childhood experiences in English adolescents: Evidence from the #BeeWell Study.
This paper is currently under review, which means it is being considered for publication. This paper uses data from the #BeeWell survey to better understand the everyday positive experiences that help young people feel safe, supported and able to thrive. While research often focuses on adversity, we also need to know what goes well in the lives of young people: things like feeling cared for at home, having trusted adults to talk to, enjoying free time and feeling accepted at school. Drawing on responses from over 120,000 adolescents aged 11–15 across England, the study measured how common different positive experiences are and how they vary for different groups of pupils. Most young people reported several positive experiences, especially enjoying their free time and feeling happy at home. However, support from adults at school or very strong self-esteem were less commonly reported. Importantly, positive experiences were not equally shared. Pupils from more disadvantaged backgrounds, those with special educational needs and those identifying as LGBTQ+ reported fewer positive experiences overall. Older pupils also reported fewer positives than younger ones. These patterns show that social circumstances influence access to everyday supports that matter for wellbeing. By highlighting where inequalities exist, this research points to the importance of strengthening supportive relationships and opportunities across home, school and community settings so that all young people have the experiences that help them flourish.
Thornton, E., Cheng, C., Demkowicz, O. & Humphrey, N. (2025). Lost learning: Prevalence, inequalities and outcomes of internal exclusion in mainstream secondary schools. British Educational Research Journal.
Internal exclusion (isolation) is a behaviour management approach involving the temporary relocation of a pupil from their classroom to a designated isolation space as a consequence of disruptive behaviour. In this study, we looked at what proportion of young people are isolated, who is more likely to be isolated, and whether isolation impacts school belonging and other outcomes. We analysed data from over 34,000 young people in 121 secondary schools in Greater Manchester. 8.3% of pupils reported being internally excluded at least once a week, for an average of 8.44 hours per week. Older pupils, boys, LGBTQ+ young people, those with special educational needs, eligible for free school meals, living in more deprived neighbourhoods and from minoritised ethnic groups were all disproportionately internally excluded. Our analyses revealed that isolation contributes to significantly worse school belonging, relationships with school staff, and (for girls) mental wellbeing. We conclude that internal exclusion is a very widely used but likely discriminatory and harmful practice; accordingly, alternative approaches to managing behaviour in secondary schools are warranted.
Cheng, Q. and Humphrey, N. (2025), Preliminary Independent Evaluation of Free2B: A Targeted Intervention to Promote the Mental Wellbeing of LGBTQIA+ Youth. Journal of Adolescence.
Young people who identify as LGBTQIA+ are more likely to experience mental health challenges than their cisgender, heterosexual peers, but there is little evidence on effective interventions. Our study provides preliminary independent evidence of the effectiveness of Free2B, a one-to-one mentoring service. Using data collected in England between 2016 and 2023, we compared 55 young people aged 10–18 who took part in Free2B with a matched control group from the #BeeWell study (N = 14,715, aged 12–15) using rigorous statistical methods. Participation in Free2B led to a statistically significant improvement in mental wellbeing – an average increase of 3.81 points on the Short Warwick-Edinburgh Mental Wellbeing Scale (SWEMWBS), a large effect size (d = 0.80) – and this finding was robust across multiple sensitivity analyses. These results are highly encouraging and suggest Free2B could be an effective way to improve the mental wellbeing of LGBTQIA+ young people, warranting further research through a larger trial.
Hugh-Jones, S., Ray, S., Wilding, A., Sutton, M., Humphrey, N. & Munford, L. (2025). Does regular engagement with arts and creative activities improve adolescent mental health and wellbeing? A systematic review and assessment of causality. Social Science and Medicine: Population Health.
There is growing evidence of and interest in the role that arts and creative activities can play in young people's mental health and wellbeing. This study reviewed all of the existing quantitative evidence for this relationship, focusing specifically on regular, everyday engagement in arts and creative activities. We applied a framework (the Bradford-Hill viewpoints) to assess to what extent this existing evidence suggests a 'causal' association (meaning that a change in mental health and wellbeing can be attributed to a change in how much arts someone does). This is particularly important as we think that this is a very complex relationship; arts engagement may affect mental health, but mental health may also affect arts engagement, and other factors (such as socioeconomic status) may affect both. The review found a wide variety of evidence for a range of arts activities, settings, populations, and methodologies. Results were generally positive, and there was at least some support for all of the viewpoints on causality, giving some support to the existence of a causal relationship between regular engagement in arts and creative activities, and better mental health and wellbeing for young people. However, more experimental evidence (and methods that can replicate some aspects of experimental conditions) would be beneficial in bolstering our confidence in this.
Knowles, C., Thornton, E., Mills-Webb, K., Petersen, K., Marquez, J., Stojiljkovic, S., Humphrey, N., (2025). Local landscapes, evolving minds: mechanisms of neighbourhood influence on dual-state mental health trajectories in adolescence. International Journal of Environmental Research and Public Health
This study explores how the areas where young people live can affect their mental health over time. Drawing on data from over 27,000 adolescents in Greater Manchester, collected between 2021 and 2023 as part of the #BeeWell programme, researchers tracked changes in life satisfaction and emotional difficulties during secondary school. The findings suggest that while most young people report consistently high life satisfaction and low emotional difficulties, a substantial minority follow less favourable patterns. For example, around 14% reported consistently low life satisfaction while a further 6% deteriorated over time. Similarly, 8% experienced high and worsening emotional difficulties. The study found that young people living in more socioeconomically deprived neighbourhoods were slightly more likely to follow sub-optimal wellbeing trajectories. To understand why, the research team looked at features of the local community that might help explain this link. They found that aspects such as access to green space, housing quality, and neighbourhood equality partly explained the effect of deprivation, suggesting that improving local environments may help protect young people’s mental health, even in more disadvantaged areas. Overall, the research shows how different aspects of community life may play a small but meaningful role in shaping mental health in adolescence and identifies housing quality alongside other environmental factors as potential targets for improving outcomes among young people in deprived communities.
Lereya, S. T., Constantinou, M., Parker, G., Thornton, Jordyn, T., Knowles, C., Humphrey, N. (2025). Bullying victimisation, self-esteem and perceived appearance among minoritised sexual and gender identities in adolescence.
This paper is currently under review, which means it is being considered for publication. Bullying remains a widespread problem in schools and is known to affect how young people feel about themselves and their appearance. This is particularly relevant for adolescents with minoritised gender and sexual identities, who often report higher rates of victimisation and poorer wellbeing. To better understand these patterns, this study explored the relationships between bullying, self-esteem, and perceived appearance across gender and sexual identity groups. Using a large cross-sectional dataset of just over 20,000 adolescents, we tested whether gender and sexual minority youth experienced more bullying and lower self-concept, and whether the negative effects of bullying were stronger for these groups. While we found that minoritised adolescents reported higher bullying and poorer outcomes, the impact of bullying on self-esteem and perceived appearance did not differ significantly by identity group. This suggests that bullying has a broadly negative effect, regardless of gender or sexuality. These findings contribute to ongoing discussions around minority stress and identity-based victimisation and suggest that while LGBTQ+ youth face greater levels of bullying, its psychological impact may not be uniquely intensified. This has implications for universal and targeted approaches to anti-bullying efforts in schools, highlighting the need for inclusive, identity-aware support strategies for all young people.
Ray, S., Cheng, Q., Demkowicz, O., McAuliffe, J. & Humphrey, N. (2025). A preliminary independent evaluation of the Factory International Schools Programme.
This paper, currently under review for publication, evaluates the impact of the Factory International Schools Programme - an arts and creativity intervention - on young people’s wellbeing. The programme ran from October 2023 to July 2024 as a pilot, offering weekly two-hour sessions led by professional artists for Year 8 and 9 students. Each half term focused on a different art form, ranging from poetry to digital music creation, and concluded with a public showcase of participants' work. A total of 107 participants completed an adapted version of the #BeeWell survey. We used a method called Propensity Score Matching to compare these participants with similar young people from the wider #BeeWell dataset who did not take part in the programme. This allowed us to control for background differences, helping us isolate the effects of the intervention. We then used a technique called Difference-in-Differences, which compares changes in wellbeing over time between the two groups - those who took part in the programme and those who didn’t. This helps to account for other factors that might influence wellbeing more generally, making it easier to identify changes that are more likely to be due to the programme itself. The results showed that participation in the Factory International Schools Programme improved young people’s perceived ability to cope with stress and their life satisfaction. However, we did not observe changes in other areas of wellbeing, including mental wellbeing, autonomy, emotion regulation, self-esteem, stress, or sense of school belonging.
Cameron, L., Farzinnia, R., Cheng, Q., Jenkins, R., & Humphrey, N. (2025). The association between school staff support and wellbeing for students with poor quality social relationships.
This paper, currently under review for publication, explored whether school staff support is particularly beneficial for students with poor quality social relationships (e.g., those lacking parent/carer support, peer support, and/or who are bullied by their peers). Using data from over 25,000 adolescents from the #BeeWell study, we fitted a series of linear regression models to estimate the association between school staff support and psychological wellbeing; optimism; self-esteem; and, life satisfaction. We then conducted moderation analysis to explore whether the association between school staff support and wellbeing varies based on students’ level of parent/carer support, peer support, and/or exposure to bullying. Analyses revealed that higher levels of school staff support are significantly associated with higher levels of all four student wellbeing outcomes. Moderation analyses revealed a complex picture, with the nature and magnitude of effects varying by relationship and wellbeing outcome. For example, for self-esteem and life satisfaction, the association with school staff support is stronger for students reporting low (vs high) parent/carer support or who are exposed (vs unexposed) to bullying. In contrast, for psychological wellbeing and optimism, it is smaller for students reporting low (vs high) social support from peers. Collectively, these results highlight the importance of having supportive school staff for promoting student wellbeing. They also suggest that increasing school staff support is a potentially useful mechanism for bolstering the wellbeing of vulnerable students (e.g., those who are bullied).
Cheng, Q., Panayiotou, M., Finserås, T. R., Andersen, A. I. O., & Humphrey, N. (2025). How do social media use, gaming frequency, and internalising symptoms influence each other over time in early-to-middle adolescence?
This paper, currently under review for publication, explored the complex relationships between social media use, gaming, and internalising symptoms like anxiety and depression among young adolescents. Using the #BeeWell survey in Greater Manchester, we tracked over 25,000 adolescents, starting from ages 12-13, across three years, asking: Does increased social media use or gaming frequency lead to increased anxiety and depression, or do adolescents experiencing these mental health challenges tend to spend more time on these platforms? We used a Random-Intercept Cross-Lagged Panel Model (RI-CLPM) to distinguish between comparing different teenagers and tracking changes within individuals over time. Our analysis revealed unexpected findings, challenging popular beliefs: We found no evidence that increased time spent on either social media or more frequent gaming predicted higher levels of emotional difficulties, such as anxiety and depression, in the following year. However, we uncovered gender-specific patterns; for girls, higher gaming frequency predicted decreased social media use a year later, while boys experiencing more symptoms like anxiety or depression tended to reduce their gaming time a year later, possibly due to loss of interest or increased parental restrictions. Whether social media use was "active" (posting and chatting) or "passive" (scrolling) had a very small impact on our overall findings, although a few minor differences were observed. These results suggest that the relationship between technology use and adolescent mental health is more nuanced than previously thought; rather than technology use driving mental health challenges, emotional well-being might influence how teenagers engage with technology, particularly among boys. These findings are based on self-reported data, which has limitations.
Stepanous, J., Irizar, P., Mills-Webb, K., Kapadia, D., Cheng, Q., Marquez, J. & Humphrey, N. (2025). Ethnic inequalities in adolescent mental wellbeing: an intersectional analysis of social identity markers, risk and protective factors. SSM - Mental Health
This study looked at ethnic differences in mental wellbeing among #BeeWell adolescents aged 12-15 and analysed how various social identity factors (like gender) and life experiences (such as discrimination or support from parents) predict mental wellbeing across different ethnic groups. Key findings were that several minoritised ethnic groups, such as Black African and Indian young people, reported better mental wellbeing compared to White British young people, while no minoritised groups reported worse wellbeing. Some factors, like the traditional male/female gender gap in wellbeing and the positive impact of peer support, were consistent across all ethnic groups. Other factors interacted with ethnicity, such as racial or religious discrimination having a particularly negative impact on Indian and Pakistani youth, and the protective effect of living in areas with a higher density of people from the same ethnic background being beneficial only for certain groups. Overall, the study provides new insights into the complex nature and factors contributing to ethnic disparities in mental wellbeing.
Khanna, D., Panayiotou, M., Black, L., Demkowicz, O. & Humphrey. N. (2024). Reciprocal associations between school experiences and adolescent wellbeing: a developmental cascades approach.
This paper is under review, which means it is being considered for publication. This paper looks at how school belonging, staff-student relationships, relational bullying, and wellbeing are associated over time, that is, how do these factors ‘cascade’ in relation to one another over adolescents’ development. While previous studies has considered developmental cascades for certain school experiences and wellbeing, none have looked at these specific factors together while separating between-person differences (stable trait-like differences across individuals) from within-person changes (state-like changes in an individual over time). To better understand developmental cascades at an individual level, this separation is necessary. This paper uses a statistical technique known as a random intercepts cross lagged panel modelling to achieve this, studying these factors over three years in a sample of 27,623 young people. The study finds positive spirals between school belonging and wellbeing, negative spirals between bullying victimisation and wellbeing, and also that wellbeing strongly predicts all school experiences. This has implications for both the target area and timing of structural interventions that take place in schools to improve wellbeing.
Marquez, J., Farzinnia, R., Cheng, Q. & Humphrey, N. (2025). Longitudinal relationships across sleep, physical activity, and mental wellbeing in early- to-mid-adolescence: a developmental cascades investigation. Quality of Life Research
This study looks at how sleep (SL), physical activity (PA), and wellbeing (WB) are connected during adolescence. While some research suggests these factors influence each other, few studies have examined their reciprocal associations over time, distinguishing between between-person and within-person effects, and none have explored all three together. Because of this, we still don’t fully understand how SL, PA, and WB interact over time at an individual level. This study contributes to filling this gap by investigating these links during the important transition from early to mid-adolescence. Using #BeeWell dat, the study analysed responses from 27,949 adolescents (aged 12/13 at the start) in Greater Manchester. It tracked changes over three years (T1, T2, T3) to explore how SL, PA, and WB influence each other over time, both between different people and within individuals. The results showed differences between boys and girls. For girls, sleep and wellbeing were reciprocally associated between T1 and T2. For boys, sleep predicted better wellbeing at both T1→T2 and T2→T3, but wellbeing did not predict sleep. Additionally, for boys, wellbeing predicted increased physical activity at T2→T3, but this effect was smaller and varied depending on how the model specification. These findings highlight the crucial role of sleep in shaping adolescent wellbeing and suggest that gender plays a role in how these factors influence each other over time.
Knowles, C., Thornton, E., Petersen, K., Lereya, S. T. & Humphrey, N. (2024). Latent classes of adolescent health behaviour, social covariates and mental wellbeing: a longitudinal birth cohort study. BMC Public Health
This study used data from the first two annual waves of #BeeWell data. The aims were to evaluate the extent to which adolescent health behaviours (i.e., physical activity, sleep, fruit and vegetable consumption, and confectionary consumption) cluster together into distinct patterns; 2) whether a broad range of social and demographic covariates influence the likelihood of engaging in healthy or unhealthy patterns of behaviour; and 3) whether any of these patterns could be linked to mental wellbeing one year later. Using a statistical approach known as Latent Class Analysis we were able to sort adolescents into one of three distinct groups with a high degree of accuracy. The Wellness Weary (15% of the sample) were characterised as the least likely to be active or get sufficient sleep, and ate fruit and vegetables relatively infrequently (i.e., the unhealthiest group). The Balanced Bunch (40% of the sample) exhibited a moderate likelihood of being active, getting sufficient sleep and eating fruit and vegetables. Finally, the Green and Dream Team (45% of the sample) were most likely to be active and get sufficient sleep, and ate substantially more fruit and vegetables than the other two groups. Confectionary consumption was largely equivocal across all groups although the Wellness Weary did consume sweets, chocolate, crisps and fizzy drinks slightly more often than the rest. Several social and demographic factors influenced the health behaviour patterns adolescents displayed. Most notably, socio-economic disadvantage, Black or Asian ethnicity, and social media use were all linked with less healthy patterns. Conversely, cisgender heterosexual girls were more likely to have favourable health behaviour patterns. After controlling for the effects of these covariates among others, there remained a significant difference between the Green and Dream Team and the Balanced Bunch one year later whereby the Green and Dream Team reported greater wellbeing. Findings illustrate the tendency for adolescents to engage in multiple health behaviours simultaneously, and that a well-rounded healthy lifestyle may have a positive impact on long-term wellbeing. Efforts to reduce both social disparity and social media use could encourage adolescents to engage in health behaviour patterns classically considered beneficial for both physical and mental wellbeing.
Thornton, E., & Humphrey, N. (2024). Prevalence, inequalities, and sequelae of bullying in adolescence: insights from the #BeeWell study. International Journal of Bullying Prevention
Being bullied can lead to mental health difficulties, as well as other negative outcomes for young people. Although it is thought to be common in childhood and adolescence, estimates vary across studies due to differences in how bullying is measured. It is also important to identify any inequalities in exposure among different groups, as some groups (for example, those with special educational needs) may be more at risk of being bullied and the associated consequences than others. We therefore investigated whether changes in how we qualify someone as being bullied, or the specific type (physical, relational, or cyber) of bullying experienced, influences what our analyses tell us about how many young people are bullied, who is more likely to be a victim of bullying, and the extent to which exposure to bullying associated with mental health difficulties, using a range of statistical analyses of the #BeeWell dataset. The extent of bullying exposure ranged between 5% and 16%, depending on how this was defined, with young people who reported being bullied frequently on at least one of physical, relational, or cyber bullying being the most common (16%). Inequalities in exposure to bullying were consistently found among LGBTQ+ young people (compared to cisgender heterosexual males); those with special educational needs (compared to those without); younger students (compared with older students); and those from more disadvantaged neighbourhoods (compared to more advantaged neighbourhoods). However, inequalities among cisgender heterosexual females (compared to cisgender heterosexual males) and ethnic minority groups (compared to those of a White ethnicity) varied based on how bullying exposure was defined and the type of bullying being considered. Our findings also indicated that those who are bullied (regardless of how this was defined) are more likely to experience significant mental health difficulties. One of our analyses indicated that if bullying could be effectively prevented, we could reduce the rate of such difficulties by nearly 20%. Finally, the impact of being bullied on mental health difficulties was stable across sociodemographic groups, with the exceptions of being more strongly associated among LGBTQ+ young people and cisgender heterosexual females (compared to cisgender heterosexual males), and less strongly associated among Black students (compared to White students). We discuss the implications of our findings, for both research and practice, including the potential of targeting bullying to reduce rates of mental health difficulties among young people, and the importance of explicitly defining how bullying has been measured, so that any findings can be interpreted within that specific context.
Cheng, Q. & Humphrey, N. (2024). Preliminary evaluation of a targeted, school-based social and emotional learning intervention for at risk youth: Football Beyond Borders. Evaluation Review
Football Beyond Borders (FBB) is a targeted, school-based social and emotional learning (SEL) intervention. Young people who participate in the programme are considered at-risk by their school, passive learners, or role models. The programme is run throughout the school year and includes: (i) weekly project-based classroom SEL sessions, implemented alongside weekly pitch-based football activities by a youth practitioner; (ii) weekly 1:1 sessions for those requiring therapeutic support led by a counsellor; (iii) engagement activities with parents/carers and, (iv) reward trips, including visits, residentials, and work experience opportunities. Our aim was to provide a preliminary robust independent evaluation of the efficacy of FBB in improving the mental wellbeing of at-risk young people, drawing on the #BeeWell dataset so that we had a comparison group who hadn’t participated in the intervention. To do so, we used a framework for analysis that can be used to test the effects of an intervention when it isn’t possible to conduct a randomised trial (difference-in-differences estimation) alongside a statistical technique that can be used to ensure that an intervention and comparison group in an evaluation are well matched (propensity score matching). We found a positive, statistically significant impact of FBB on the mental wellbeing of at-risk young people, equivalent to just under 2 and a half points on the SWEMWBS scale (an effect size of d=0.44, or a 17-percentile point increase). To check the robustness of this effect, we ran a series of additional checks (e.g. using different PSM approaches to the one we had chosen for our main analysis). In most cases, this did not substantively change the intervention effect we had identified, leading us to conclude that FBB improves the mental wellbeing of at-risk young people. This evaluation showcases the utility of the #BeeWell dataset in helping to understand how best to support young people’s wellbeing. We hope to undertake similar evaluations of other interventions in the future.
Thornton, E., Petersen, K., Marquez, J. & Humphrey, N. (2024). Do patterns of adolescent participation in arts, culture and entertainment activities predict later wellbeing? a latent class analysis. Journal of Youth and Adolescence
The ways in which we spend our free time (i.e., leisure) can benefit us in terms of our mental wellbeing. This includes participation in arts, culture, and entertainment (PACE) activities (e.g. creative hobbies, playing videogames, reading for pleasure). We used a statistical method called Latent Class Analysis (LCA) to identify common profiles (groups) of PACE among young people in Greater Manchester, based on the amount of time they reported spending doing each activity. We could then assess whether certain socio-demographic characteristics (e.g., socio-economic disadvantage) predicted membership of a given PACE group, and whether PACE group membership predicted later wellbeing. Patterns of PACE fell into four distinct groups, which we called 1) the Dynamic Doers, who had high probabilities of taking part in a wide range of activities; 2) the Mind and Body Crew, who were likely to spend their time reading for enjoyment, participating in arts and crafts or other creative hobbies, doing exercise or other physical activities, and playing video games; 3) the Game and Gain Squad, who were highly likely to play sports, do exercise or other physical activities, and play video games; and 4) the Activity Free Association, who were unlikely to take part in any PACE activities. Associations between socio-demographic characteristics and PACE classification were established (e.g., socio-economic disadvantage increased the likelihood of Activity Free Association classification, compared to Game and Gain Squad classification). We also found that PACE classification predicted later wellbeing (e.g., Dynamic Doers reported significantly higher wellbeing than the Activity Free Association). These findings emphasise the role of PACE in supporting young people’s wellbeing and underscore the importance of making PACE activities appealing and accessible to all.
Khanna, D, Black, L., Panayiotou, M, Demkowicz O & Humphrey, N. (2024). Conceptualising and measuring adolescents’ hedonic and eudemonic wellbeing: discriminant validity and dimensionality concerns. Child Indicators Research
‘Wellbeing’ can be defined and measured in many different ways. Commonly, there are two ways of thinking about wellbeing as ‘hedonic’, which considers wellbeing associated with immediate pleasure, moods and more cognitive aspects, or ‘eudaimonic’, which considers wellbeing associated with ideas of flourishing, functioning and even self-esteem and positive social relationships. This paper looks at some of the wellbeing domains used in #BeeWell (positive affect, negative affect, life satisfaction, autonomy, self-esteem, optimism, friendships and social support) to see whether they can be grouped together into categories of hedonia and eudaimonia, or whether the questions asked in the survey are better thought of as encompassed by a more general idea of wellbeing. We test three models, one that looks at each domain together at one level (correlated model), one that groups them into separate higher-level hedonic and eudemonic factors (higher-order model), and one that considers each of the individual items also aligned with a general wellbeing factor (bifactor model). We find that the traditional ways of separating wellbeing as hedonic versus eudemonic does not apply; there is a disconnect between the theoretical ideas and the empirical reality. This work is therefore useful in thinking about how established theories and conceptualisations often guide the design of new measures and analyses of wellbeing, without the preliminary conceptualisation first being tested in different samples eg. that of adolescents.
Marquez, J., Francis-Hew, L., & Humphrey, N. (2023). Protective factors for wellbeing resilience in adolescence: a longitudinal analysis using the residuals approach. Child and Adolescent Psychiatry and Mental Health
We used an innovative approach (the residuals approach) to advance our understanding of wellbeing resilience processes in adolescence. We analysed longitudinal #BeeWell data on 12, 130 young people to examine how adversity exposure impacts later wellbeing (life satisfaction, and internalising mental health difficulties) in the transition from year 8 (age 12/13) to year 9 (age 13/14); whether gender and ethnic differences in wellbeing resilience exist; which internal and external factors confer protective effects for wellbeing resilience; and whether the protective effect of these factors differs by gender and level of adversity exposure. Multiple adversity factors (e.g., home material deprivation, sexuality discrimination, bullying) were found to impact later wellbeing. Girls and white adolescents presented lower wellbeing resilience than their peers. Internal psychological factors (self-esteem, emotional regulation, optimism) consistently presented the strongest protective effects, but behavioural/activity factors (physical activity, sleep) also contributed to wellbeing resilience. Among external factors, friendships and peer support were the most salient. Physical activity yielded stronger protective effects among boys (compared to girls), while the reverse was true of self-esteem. Effects of protective factors were stronger among those at lower (compared to higher) levels of adversity exposure, which suggests that prevention and intervention efforts to reduce said exposure in the first place should be prioritised. These findings provide clear implications for policy and practice in terms of prevention (of adversity exposure) and intervention (to facilitate resilience).
Black, L., Humphrey, N., Panayiotou, M., & Marquez, J. (2023). Mental health and wellbeing measures for mean comparison and screening in adolescents: an assessment of unidimensionality and sex and age measurement invariance. Assessment
The #BeeWell survey examines several domains of young people's wellbeing, drawing on consultation from young people and experts in its development (see here for more information). For instance, it covers questions about autonomy, optimism, as well as about stress and negative affect. Developing good measures is a difficult, lengthy and resource-intensive process. While the #BeeWell survey is made up of those supported by good evidence, there is still work to be done in this field in general. We therefore examined how our different wellbeing measures performed from several perspectives through psychometric analyses. First, we considered how well questions hung together when considering total scores for a given measure (e.g., autonomy or stress). Second, we analysed whether the measures were interpreted in the same way by boys versus girls (sex), and by those in Year 8 versus Year 10 (age). We found five of the eight measures had good evidence for use as total scores. For these five, we did however find differences in how the measures worked across sex and age. This means complex models should ideally be used rather than basic mean differences when comparing across these groups to ensure valid conclusions are drawn. Score cut-offs (e.g., considering a score of 10 or more to be indicative of increased risk) tended to be applicable across groups except for negative affect, for which the same score across girls and boys likely substantially underestimates the amount of problems experienced by boys.
Marquez, J., Humphrey, N., Black, L., Cutts, M., & Khanna, D. (2023). Gender and sexual identity-based inequalities in adolescent wellbeing: findings from the #BeeWell study. BMC Public Health
In #BeeWell Evidence Briefing 1 we reported substantial inequalities in wellbeing affecting LGTBQI+ young people in Greater Manchester. These were routinely substantially greater than those concerning other characteristics (e.g. disparities across ethnic groups), meaning that LGTBQI+ young people reported much lower levels of wellbeing than their non-LGTBQI+ peers. In this paper, we extended our analysis to focus on a wider range of wellbeing domains, including life satisfaction, positive and negative affect. autonomy, self-esteem, optimism, positive relationships, symptoms of distress and mental wellbeing. For each of these, we found further evidence that wellbeing inequalities affecting LGTBQI+ young people were greater than those relating to other characteristics (e.g., socio-economic disadvantage). Young people whose identities transcend traditional binaries (e.g., non-binary) were subject to the most substantial inequalities when considering gender. Regarding sexual orientation, the greatest disparities were apparent for those who identified as gay/lesbian or bi/pansexual. Although these inequalities were substantial across all the wellbeing domains studied, they were particularly strong for negative affect (e.g. sadness, worry). We discuss the implications of these findings, including the importance of a whole-system response to tackle gender- and sexuality-based discrimination and prejudice, and provide safer and more inclusive cultures and spaces in which all young people can thrive.
Marquez, J., Humphrey, N., Black, L., & Wozmirska, S. (2023). This is the place: a multi-level analysis of neighbourhood correlates of young people’s wellbeing. Social Psychiatry and Psychiatric Epidemiology
The #BeeWell data provide an excellent opportunity to study how neighbourhoods influence young people’s wellbeing in Greater Manchester. To investigate neighbourhood characteristics, data collected by the Co-op’s Community Wellbeing Index (CWI) were linked to #BeeWell dataset. The CWI uses neighbourhood boundaries called ‘seamless locales’, which are based on user feedback, data on observed travel, work and shop patterns, and administrative data. We studied neighbourhood effects for 35,902 young people across 243 neighbourhoods in Greater Manchester using a statistical analysis technique known as multi-level regression. In line with previous research, neighbourhoods accounted for a small but significant proportion of the variation in young people’s life satisfaction (0.61%) and internalising symptoms (1.17%). Furthermore, we found evidence that wellbeing inequalities for some groups (e.g. gender diverse young people) varied across neighbourhoods. Finally, several neighbourhood characteristics were found to predict individual wellbeing outcomes. For example, higher levels of perceived wellbeing support from local people predicted lower internalising symptoms. On the basis of such findings, we make a number of recommendations for placed-based, hyper-local policy responses. We also provide guidance for future research in this area, including the need for longitudinal work that uses indices of neighbourhoods and their characteristics designed by young people.
Black, L., Farzinnia, R., Humphrey, N. & Marquez, J. (2023). Variation in global network properties across risk factors for adolescent internalizing symptoms: evidence of cumulative effects on structure and connectivity. Psychological Medicine
Teenagers can be particularly vulnerable to developing ‘internalizing symptoms’ (e.g. worry, sadness) and certain factors, such as perceived lack of family support, can heighten this risk further. The number of these risk factors to which a given young person is exposed may be important in predicting experiences of symptoms. However, previous studies have analysed symptoms in a simple way, treating mental health problems like a disease. More recently, researchers have begun to consider them as an interacting network of symptoms. For instance, worry might cause sleep problems, which might cause concentration issues, which might cause feelings of sadness, and so on. The idea is that symptoms reinforce each other to cause a distressed mental state. In this study, we assessed the effect of nine risk factors, and cumulative risk, on the structure and connectivity of internalizing symptom networks. By structure we mean which symptoms are interacting with one another, and by connectivity we mean how strongly they are related. Cumulative risk refers to the number of risk factors to which an individual is exposed. Network theory suggests that those at risk should show stronger connectivity and altered structure compared to those who are less vulnerable, and indeed, this is what we found in relation to individual risk factors and cumulative risk, though effects were mostly small. However, gender and sexual minority risk groups, as well those experiencing low parent/carer support, showed bigger differences for connectivity than the risk-absent groups. Furthermore, cumulative risk was a strong predictor of network connectivity. We conclude that network approaches to internalizing symptoms in adolescents can be useful to understand risks further. Gender and sexual minority groups, as well as those with low parent/carer support, may be particularly vulnerable and should be prioritized in prevention and intervention efforts. Our analysis also supports the idea that the number of risk factors experienced, regardless of their nature, predicts mental health outcomes. Young people experiencing multiple known risk factors should therefore receive additional support.
Marquez, J., Qualter, P., Petersen, K., Humphrey, N., & Black, L. (2023). Neighbourhood effects on loneliness among adolescents. Journal of Public Health
Loneliness is a growing public health concern, but little is known about how place affects loneliness, especially in adolescence. This is the first study to examine the influence of neighbourhoods and their characteristics on loneliness in early/middle adolescence. We used #BeeWell data and administrative data collected at the neighbourhood (Lower Super Output Area) level to study neighbourhood effects on loneliness for 36,141 young people across 1,590 neighbourhoods in Greater Manchester using a statistical analysis technique known as multi-level regression. Neighbourhoods accounted for a small but significant proportion of the variation in adolescents’ loneliness (1.18%). We also found higher levels of loneliness among older adolescents (vs younger), whites (vs ethnic minorities), girls and gender diverse individuals (vs boys), sexual minorities (vs heterosexual), and those eligible for free school meals (vs non-eligible). Moreover, we found evidence that ethnic, gender and sexual orientation inequalities in loneliness varied across neighbourhoods. Furthermore, loneliness was higher in neighbourhoods with higher skills deprivation among children and young people, lower skills deprivation among adults, higher geographical barriers, lower outdoor environment deprivation, and higher population density, as well as among those who have to travel further from home to school. Finally, more positive perceptions of the local area (feeling safe, trust in local people, feeling supported by local people, seeing neighbours as helpful, and the availability of good places to spend your free time) were associated with lower levels of loneliness. Our findings suggest that local-level interventions may be particularly helpful to tackle adolescent loneliness and reduce loneliness inequalities affecting vulnerable groups (e.g. those who identify as LGBTQ+).
Black, L., Humphrey, N. & Marquez, J. (2023). The influence of minority stress-related experiences on mental wellbeing for trans/gender-diverse and cisgender youth: a comparative longitudinal analysis. Royal Society Open Science
Trans and gender-diverse (TGD) young people are likely to experience poorer mental health and wellbeing than their cisgender peers. Minority stress theory suggests that this may be because increased exposure to factors such as bullying and discrimination lead to excess stress and reduced wellbeing. However, the evidence base remains limited. In this study we set out to examine these ideas using #BeeWell data. First, we examined differences in wellbeing between TGD young people and their peers. Second, we examined differences between these groups in exposure to minority-related stressors (bullying, gender discrimination, lack of family support, and feeling unsafe). Third, we examined the extent to which exposure to minority-related stressors predicted later wellbeing. Our analysis showed that compared to cisgender boys, TGD young people, cisgender girls, and those who preferred not to say their gender identity reported lower wellbeing (with the largest effect evident for the TGD group). TGD young people also reported higher rates of exposure to all minority-related stressors. We found that exposure to these stressors predicted later wellbeing (e.g., being bullied more in 2021 predicted lower wellbeing in 2022), but that this effect was not moderated by gender (e.g., the magnitude of the impact of bullying on later wellbeing was equivalent for all our gender groups). We discuss the implications of these findings in relation to things like coverage of gender identity in relationships and sex education guidance for schools, in addition to tailored services to support TGD young people.
Cheng, Q., Mills-Webb, K., Marquez, J. & Humphrey, N. (2025). Longitudinal relationships across bullying victimization, friendship and social support, and internalizing symptoms in early-to-middle adolescence: A developmental cascades investigation. Journal of Youth and Adolescence
This study, published with Journal of Youth and Adolescence, 2025, used data from the #BeeWell survey in Greater Manchester to understand the complex relationships between bullying, friendships, social support, and mental health in young people. The researchers looked at information from over 26,000 adolescents, who were around 12-13 years old at the start, and followed them over three years. The main goal was to figure out how these factors influence each other over time, and whether the patterns differed for boys and girls. We used a "Random-Intercept Cross-Lagged Panel Model" (RI-CLPM), which separates two things: comparing different teens (e.g., Are teens who are bullied more, in general, more anxious?) and looking at changes within the same teen (e.g., If a teen is bullied more than usual one year, are they more anxious or depressed the next year?). The key findings revealed important differences between boys and girls. For girls, being bullied led to fewer friendships and less social support later on. It also led to more symptoms like anxiety, sadness, and loneliness. Girls with better friendships and more support from friends had fewer of these symptoms. For boys, the pattern was different. Being bullied led to more of these symptoms in the subsequent year. Also, boys who were already feeling anxious or depressed were more likely to be bullied later. The study highlights that bullying is harmful to young people's mental health and, for girls, their friendships. Because of these gender differences, we emphasize that anti-bullying programs and mental health support should be tailored to boys and girls, considering how they experience these challenges differently. It is worth noting that the study is based on what teens reported about themselves.
Cheng, Q., Panayiotou, M., Finserås, T. R., Andersen, A. I. O., & Humphrey, N. (2025). How do social media use, gaming frequency, and internalising symptoms influence each other over time in early-to-middle adolescence?
This paper, currently under review for publication, explored the complex relationships between social media use, gaming, and internalising symptoms like anxiety and depression among young adolescents. Using the #BeeWell survey in Greater Manchester, we tracked over 25,000 adolescents, starting from ages 12-13, across three years, asking: Does increased social media use or gaming frequency lead to increased anxiety and depression, or do adolescents experiencing these mental health challenges tend to spend more time on these platforms? We used a Random-Intercept Cross-Lagged Panel Model (RI-CLPM) to distinguish between comparing different teenagers and tracking changes within individuals over time. Our analysis revealed unexpected findings, challenging popular beliefs: We found no evidence that increased time spent on either social media or more frequent gaming predicted higher levels of emotional difficulties, such as anxiety and depression, in the following year. However, we uncovered gender-specific patterns; for girls, higher gaming frequency predicted decreased social media use a year later, while boys experiencing more symptoms like anxiety or depression tended to reduce their gaming time a year later, possibly due to loss of interest or increased parental restrictions. Whether social media use was "active" (posting and chatting) or "passive" (scrolling) had a very small impact on our overall findings, although a few minor differences were observed. These results suggest that the relationship between technology use and adolescent mental health is more nuanced than previously thought; rather than technology use driving mental health challenges, emotional well-being might influence how teenagers engage with technology, particularly among boys. These findings are based on self-reported data, which has limitations.
Stepanous, J., Irizar, P., Mills-Webb, K., Kapadia, D., Cheng, Q., Marquez, J. & Humphrey, N. (2025). Ethnic inequalities in adolescent mental wellbeing: An intersectional analysis of social identity markers, risk and protective factors.
This paper is under review, which means it is being considered for publication. This study looked at ethnic differences in mental wellbeing among #BeeWell adolescents aged 12-15 and analysed how various social identity factors (like gender) and life experiences (such as discrimination or support from parents) predict mental wellbeing across different ethnic groups. Key findings were that several minoritised ethnic groups, such as Black African and Indian young people, reported better mental wellbeing compared to White British young people, while no minoritised groups reported worse wellbeing. Some factors, like the traditional male/female gender gap in wellbeing and the positive impact of peer support, were consistent across all ethnic groups. Other factors interacted with ethnicity, such as racial or religious discrimination having a particularly negative impact on Indian and Pakistani youth, and the protective effect of living in areas with a higher density of people from the same ethnic background being beneficial only for certain groups. Overall, the study provides new insights into the complex nature and factors contributing to ethnic disparities in mental wellbeing.
Khanna, D., Panayiotou, M., Black, L., Demkowicz, O. & Humphrey. N. (2024). The effects of learning-related and peer-related school experiences on adolescent wellbeing: a longitudinal structural equation model.
This paper is under review, which means it is being considered for publication. Schools are complex environments, with multiple experiences occurring within them at the same time. To improve students’ wellbeing, it is important to understand which of these experiences matter, and whether certain areas of school life – academic or social – are more or less strongly associated with wellbeing. This paper considers the effects of three learning-related school experiences (happiness with attainment, careers education, staff-student relationships) and four peer-related school experiences (school belonging, friendships and social support, relational bullying victimisation, physical bullying victimisation) on later wellbeing. Using a sample of 25,249 young people in Greater Manchester over two years, we fit a longitudinal structural model, an approach which allows us to account for the correlation between different school experiences with one another as well as account for prior wellbeing levels. We find that school belonging is most strongly related to wellbeing outcomes while staff-student relationships and physical bullying victimisation do not have significant effects. This has implications for which school experiences future structural interventions are targeted towards to improve students’ wellbeing.
Khanna, D., Panayiotou, M., Black, L., Demkowicz, O. & Humphrey. N. (2024). Reciprocal associations between school experiences and adolescent wellbeing: a developmental cascades approach.
This paper is under review, which means it is being considered for publication. This paper looks at how school belonging, staff-student relationships, relational bullying, and wellbeing are associated over time, that is, how do these factors ‘cascade’ in relation to one another over adolescents’ development. While previous studies has considered developmental cascades for certain school experiences and wellbeing, none have looked at these specific factors together while separating between-person differences (stable trait-like differences across individuals) from within-person changes (state-like changes in an individual over time). To better understand developmental cascades at an individual level, this separation is necessary. This paper uses a statistical technique known as a random intercepts cross lagged panel modelling to achieve this, studying these factors over three years in a sample of 27,623 young people. The study finds positive spirals between school belonging and wellbeing, negative spirals between bullying victimisation and wellbeing, and also that wellbeing strongly predicts all school experiences. This has implications for both the target area and timing of structural interventions that take place in schools to improve wellbeing.
Marquez, J., Farzinnia, R., Cheng, Q. & Humphrey, N. (2025). Longitudinal relationships across sleep, physical activity, and mental wellbeing in early- to-mid-adolescence: a developmental cascades investigation. Quality of Life Research
This study looks at how sleep (SL), physical activity (PA), and wellbeing (WB) are connected during adolescence. While some research suggests these factors influence each other, few studies have examined their reciprocal associations over time, distinguishing between between-person and within-person effects, and none have explored all three together. Because of this, we still don’t fully understand how SL, PA, and WB interact over time at an individual level. This study contributes to filling this gap by investigating these links during the important transition from early to mid-adolescence. Using #BeeWell dat, the study analysed responses from 27,949 adolescents (aged 12/13 at the start) in Greater Manchester. It tracked changes over three years (T1, T2, T3) to explore how SL, PA, and WB influence each other over time, both between different people and within individuals. The results showed differences between boys and girls. For girls, sleep and wellbeing were reciprocally associated between T1 and T2. For boys, sleep predicted better wellbeing at both T1→T2 and T2→T3, but wellbeing did not predict sleep. Additionally, for boys, wellbeing predicted increased physical activity at T2→T3, but this effect was smaller and varied depending on how the model specification. These findings highlight the crucial role of sleep in shaping adolescent wellbeing and suggest that gender plays a role in how these factors influence each other over time.
Knowles, C., Thornton, E., Petersen, K., Lereya, S. T. & Humphrey, N. (2024). Latent classes of adolescent Health behaviour, social covariates and mental wellbeing: a longitudinal birth cohort study. BMC Public Health
This study used data from the first two annual waves of #BeeWell data. The aims were to evaluate the extent to which adolescent health behaviours (i.e., physical activity, sleep, fruit and vegetable consumption, and confectionary consumption) cluster together into distinct patterns; 2) whether a broad range of social and demographic covariates influence the likelihood of engaging in healthy or unhealthy patterns of behaviour; and 3) whether any of these patterns could be linked to mental wellbeing one year later. Using a statistical approach known as Latent Class Analysis we were able to sort adolescents into one of three distinct groups with a high degree of accuracy. The Wellness Weary (15% of the sample) were characterised as the least likely to be active or get sufficient sleep, and ate fruit and vegetables relatively infrequently (i.e., the unhealthiest group). The Balanced Bunch (40% of the sample) exhibited a moderate likelihood of being active, getting sufficient sleep and eating fruit and vegetables. Finally, the Green and Dream Team (45% of the sample) were most likely to be active and get sufficient sleep, and ate substantially more fruit and vegetables than the other two groups. Confectionary consumption was largely equivocal across all groups although the Wellness Weary did consume sweets, chocolate, crisps and fizzy drinks slightly more often than the rest. Several social and demographic factors influenced the health behaviour patterns adolescents displayed. Most notably, socio-economic disadvantage, Black or Asian ethnicity, and social media use were all linked with less healthy patterns. Conversely, cisgender heterosexual girls were more likely to have favourable health behaviour patterns. After controlling for the effects of these covariates among others, there remained a significant difference between the Green and Dream Team and the Balanced Bunch one year later whereby the Green and Dream Team reported greater wellbeing. Findings illustrate the tendency for adolescents to engage in multiple health behaviours simultaneously, and that a well-rounded healthy lifestyle may have a positive impact on long-term wellbeing. Efforts to reduce both social disparity and social media use could encourage adolescents to engage in health behaviour patterns classically considered beneficial for both physical and mental wellbeing.
Thornton, E., & Humphrey, N. (2024). Prevalence, inequalities, and sequelae of bullying in adolescence: insights from the #BeeWell study. International Journal of Bullying Prevention
Being bullied can lead to mental health difficulties, as well as other negative outcomes for young people. Although it is thought to be common in childhood and adolescence, estimates vary across studies due to differences in how bullying is measured. It is also important to identify any inequalities in exposure among different groups, as some groups (for example, those with special educational needs) may be more at risk of being bullied and the associated consequences than others. We therefore investigated whether changes in how we qualify someone as being bullied, or the specific type (physical, relational, or cyber) of bullying experienced, influences what our analyses tell us about how many young people are bullied, who is more likely to be a victim of bullying, and the extent to which exposure to bullying associated with mental health difficulties, using a range of statistical analyses of the #BeeWell dataset. The extent of bullying exposure ranged between 5% and 16%, depending on how this was defined, with young people who reported being bullied frequently on at least one of physical, relational, or cyber bullying being the most common (16%). Inequalities in exposure to bullying were consistently found among LGBTQ+ young people (compared to cisgender heterosexual males); those with special educational needs (compared to those without); younger students (compared with older students); and those from more disadvantaged neighbourhoods (compared to more advantaged neighbourhoods). However, inequalities among cisgender heterosexual females (compared to cisgender heterosexual males) and ethnic minority groups (compared to those of a White ethnicity) varied based on how bullying exposure was defined and the type of bullying being considered. Our findings also indicated that those who are bullied (regardless of how this was defined) are more likely to experience significant mental health difficulties. One of our analyses indicated that if bullying could be effectively prevented, we could reduce the rate of such difficulties by nearly 20%. Finally, the impact of being bullied on mental health difficulties was stable across sociodemographic groups, with the exceptions of being more strongly associated among LGBTQ+ young people and cisgender heterosexual females (compared to cisgender heterosexual males), and less strongly associated among Black students (compared to White students). We discuss the implications of our findings, for both research and practice, including the potential of targeting bullying to reduce rates of mental health difficulties among young people, and the importance of explicitly defining how bullying has been measured, so that any findings can be interpreted within that specific context.
Cheng, Q. & Humphrey, N. (2024). Preliminary evaluation of a targeted, school-based social and emotional learning intervention for at risk youth: Football Beyond Borders.
This paper is under review meaning it is currently being considered for publication. Football Beyond Borders (FBB) is a targeted, school-based social and emotional learning (SEL) intervention. Young people who participate in the programme are considered at-risk by their school, passive learners, or role models. The programme is run throughout the school year and includes: (i) weekly project-based classroom SEL sessions, implemented alongside weekly pitch-based football activities by a youth practitioner; (ii) weekly 1:1 sessions for those requiring therapeutic support led by a counsellor; (iii) engagement activities with parents/carers and, (iv) reward trips, including visits, residentials, and work experience opportunities. Our aim was to provide a preliminary robust independent evaluation of the efficacy of FBB in improving the mental wellbeing of at-risk young people, drawing on the #BeeWell dataset so that we had a comparison group who hadn’t participated in the intervention. To do so, we used a framework for analysis that can be used to test the effects of an intervention when it isn’t possible to conduct a randomised trial (difference-in-differences estimation) alongside a statistical technique that can be used to ensure that an intervention and comparison group in an evaluation are well matched (propensity score matching). We found a positive, statistically significant impact of FBB on the mental wellbeing of at-risk young people, equivalent to just under 2 and a half points on the SWEMWBS scale (an effect size of d=0.44, or a 17-percentile point increase). To check the robustness of this effect, we ran a series of additional checks (e.g. using different PSM approaches to the one we had chosen for our main analysis). In most cases, this did not substantively change the intervention effect we had identified, leading us to conclude that FBB improves the mental wellbeing of at-risk young people. This evaluation showcases the utility of the #BeeWell dataset in helping to understand how best to support young people’s wellbeing. We hope to undertake similar evaluations of other interventions in the future.
Thornton, E., Petersen, K., Marquez, J. & Humphrey, N. (2024). Do patterns of adolescent participation in arts, culture and entertainment activities predict later wellbeing? a latent class analysis. Journal of Youth and Adolescence.
The ways in which we spend our free time (i.e., leisure) can benefit us in terms of our mental wellbeing. This includes participation in arts, culture, and entertainment (PACE) activities (e.g. creative hobbies, playing videogames, reading for pleasure). We used a statistical method called Latent Class Analysis (LCA) to identify common profiles (groups) of PACE among young people in Greater Manchester, based on the amount of time they reported spending doing each activity. We could then assess whether certain socio-demographic characteristics (e.g., socio-economic disadvantage) predicted membership of a given PACE group, and whether PACE group membership predicted later wellbeing. Patterns of PACE fell into four distinct groups, which we called 1) the Dynamic Doers, who had high probabilities of taking part in a wide range of activities; 2) the Mind and Body Crew, who were likely to spend their time reading for enjoyment, participating in arts and crafts or other creative hobbies, doing exercise or other physical activities, and playing video games; 3) the Game and Gain Squad, who were highly likely to play sports, do exercise or other physical activities, and play video games; and 4) the Activity Free Association, who were unlikely to take part in any PACE activities. Associations between socio-demographic characteristics and PACE classification were established (e.g., socio-economic disadvantage increased the likelihood of Activity Free Association classification, compared to Game and Gain Squad classification). We also found that PACE classification predicted later wellbeing (e.g., Dynamic Doers reported significantly higher wellbeing than the Activity Free Association). These findings emphasise the role of PACE in supporting young people’s wellbeing and underscore the importance of making PACE activities appealing and accessible to all.
Khanna, D, Black, L., Panayiotou, M, Demkowicz O & Humphrey, N. (2024). Conceptualising and measuring adolescents’ hedonic and eudemonic wellbeing: discriminant validity and dimensionality concerns. Child Indicators Research
‘Wellbeing’ can be defined and measured in many different ways. Commonly, there are two ways of thinking about wellbeing as ‘hedonic’, which considers wellbeing associated with immediate pleasure, moods and more cognitive aspects, or ‘eudaimonic’, which considers wellbeing associated with ideas of flourishing, functioning and even self-esteem and positive social relationships. This paper looks at some of the wellbeing domains used in #BeeWell (positive affect, negative affect, life satisfaction, autonomy, self-esteem, optimism, friendships and social support) to see whether they can be grouped together into categories of hedonia and eudaimonia, or whether the questions asked in the survey are better thought of as encompassed by a more general idea of wellbeing. We test three models, one that looks at each domain together at one level (correlated model), one that groups them into separate higher-level hedonic and eudemonic factors (higher-order model), and one that considers each of the individual items also aligned with a general wellbeing factor (bifactor model). We find that the traditional ways of separating wellbeing as hedonic versus eudemonic does not apply; there is a disconnect between the theoretical ideas and the empirical reality. This work is therefore useful in thinking about how established theories and conceptualisations often guide the design of new measures and analyses of wellbeing, without the preliminary conceptualisation first being tested in different samples eg. that of adolescents.
Marquez, J., Francis-Hew, L., & Humphrey, N. (2023). Protective factors for wellbeing resilience in adolescence: a longitudinal analysis using the residuals approach. Child and Adolescent Psychiatry and Mental Health
We used an innovative approach (the residuals approach) to advance our understanding of wellbeing resilience processes in adolescence. We analysed longitudinal #BeeWell data on 12, 130 young people to examine how adversity exposure impacts later wellbeing (life satisfaction, and internalising mental health difficulties) in the transition from year 8 (age 12/13) to year 9 (age 13/14); whether gender and ethnic differences in wellbeing resilience exist; which internal and external factors confer protective effects for wellbeing resilience; and whether the protective effect of these factors differs by gender and level of adversity exposure. Multiple adversity factors (e.g., home material deprivation, sexuality discrimination, bullying) were found to impact later wellbeing. Girls and white adolescents presented lower wellbeing resilience than their peers. Internal psychological factors (self-esteem, emotional regulation, optimism) consistently presented the strongest protective effects, but behavioural/activity factors (physical activity, sleep) also contributed to wellbeing resilience. Among external factors, friendships and peer support were the most salient. Physical activity yielded stronger protective effects among boys (compared to girls), while the reverse was true of self-esteem. Effects of protective factors were stronger among those at lower (compared to higher) levels of adversity exposure, which suggests that prevention and intervention efforts to reduce said exposure in the first place should be prioritised. These findings provide clear implications for policy and practice in terms of prevention (of adversity exposure) and intervention (to facilitate resilience).
Black, L., Humphrey, N., Panayiotou, M., & Marquez, J. (2023). Mental health and wellbeing measures for mean comparison and screening in adolescents: An assessment of unidimensionality and sex and age measurement invariance. Assessment
The #BeeWell survey examines several domains of young people's wellbeing, drawing on consultation from young people and experts in its development (see here for more information). For instance, it covers questions about autonomy, optimism, as well as about stress and negative affect. Developing good measures is a difficult, lengthy and resource-intensive process. While the #BeeWell survey is made up of those supported by good evidence, there is still work to be done in this field in general. We therefore examined how our different wellbeing measures performed from several perspectives through psychometric analyses. First, we considered how well questions hung together when considering total scores for a given measure (e.g., autonomy or stress). Second, we analysed whether the measures were interpreted in the same way by boys versus girls (sex), and by those in Year 8 versus Year 10 (age). We found five of the eight measures had good evidence for use as total scores. For these five, we did however find differences in how the measures worked across sex and age. This means complex models should ideally be used rather than basic mean differences when comparing across these groups to ensure valid conclusions are drawn. Score cut-offs (e.g., considering a score of 10 or more to be indicative of increased risk) tended to be applicable across groups except for negative affect, for which the same score across girls and boys likely substantially underestimates the amount of problems experienced by boys.
Marquez, J., Humphrey, N., Black, L., Cutts, M., & Khanna, D. (2023). Gender and sexual identity-based inequalities in adolescent wellbeing: findings from the #BeeWell study. BMC Public Health
In #BeeWell Evidence Briefing 1 we reported substantial inequalities in wellbeing affecting LGTBQI+ young people in Greater Manchester. These were routinely substantially greater than those concerning other characteristics (e.g. disparities across ethnic groups), meaning that LGTBQI+ young people reported much lower levels of wellbeing than their non-LGTBQI+ peers. In this paper, we extended our analysis to focus on a wider range of wellbeing domains, including life satisfaction, positive and negative affect. autonomy, self-esteem, optimism, positive relationships, symptoms of distress and mental wellbeing. For each of these, we found further evidence that wellbeing inequalities affecting LGTBQI+ young people were greater than those relating to other characteristics (e.g., socio-economic disadvantage). Young people whose identities transcend traditional binaries (e.g., non-binary) were subject to the most substantial inequalities when considering gender. Regarding sexual orientation, the greatest disparities were apparent for those who identified as gay/lesbian or bi/pansexual. Although these inequalities were substantial across all the wellbeing domains studied, they were particularly strong for negative affect (e.g. sadness, worry). We discuss the implications of these findings, including the importance of a whole-system response to tackle gender- and sexuality-based discrimination and prejudice, and provide safer and more inclusive cultures and spaces in which all young people can thrive.
Marquez, J., Humphrey, N., Black, L., & Wozmirska, S. (2023). This is the place: a multi-level analysis of neighbourhood correlates of young people’s wellbeing. Social Psychiatry and Psychiatric Epidemiology
The #BeeWell data provide an excellent opportunity to study how neighbourhoods influence young people’s wellbeing in Greater Manchester. To investigate neighbourhood characteristics, data collected by the Co-op’s Community Wellbeing Index (CWI) were linked to #BeeWell dataset. The CWI uses neighbourhood boundaries called ‘seamless locales’, which are based on user feedback, data on observed travel, work and shop patterns, and administrative data. We studied neighbourhood effects for 35,902 young people across 243 neighbourhoods in Greater Manchester using a statistical analysis technique known as multi-level regression. In line with previous research, neighbourhoods accounted for a small but significant proportion of the variation in young people’s life satisfaction (0.61%) and internalising symptoms (1.17%). Furthermore, we found evidence that wellbeing inequalities for some groups (e.g. gender diverse young people) varied across neighbourhoods. Finally, several neighbourhood characteristics were found to predict individual wellbeing outcomes. For example, higher levels of perceived wellbeing support from local people predicted lower internalising symptoms. On the basis of such findings, we make a number of recommendations for placed-based, hyper-local policy responses. We also provide guidance for future research in this area, including the need for longitudinal work that uses indices of neighbourhoods and their characteristics designed by young people.
Black, L., Farzinnia, R., Humphrey, N. & Marquez, J. (2023). Variation in global network properties across risk factors for adolescent internalizing symptoms: evidence of cumulative effects on structure and connectivity. Psychological Medicine
Teenagers can be particularly vulnerable to developing ‘internalizing symptoms’ (e.g. worry, sadness) and certain factors, such as perceived lack of family support, can heighten this risk further. The number of these risk factors to which a given young person is exposed may be important in predicting experiences of symptoms. However, previous studies have analysed symptoms in a simple way, treating mental health problems like a disease. More recently, researchers have begun to consider them as an interacting network of symptoms. For instance, worry might cause sleep problems, which might cause concentration issues, which might cause feelings of sadness, and so on. The idea is that symptoms reinforce each other to cause a distressed mental state. In this study, we assessed the effect of nine risk factors, and cumulative risk, on the structure and connectivity of internalizing symptom networks. By structure we mean which symptoms are interacting with one another, and by connectivity we mean how strongly they are related. Cumulative risk refers to the number of risk factors to which an individual is exposed. Network theory suggests that those at risk should show stronger connectivity and altered structure compared to those who are less vulnerable, and indeed, this is what we found in relation to individual risk factors and cumulative risk, though effects were mostly small. However, gender and sexual minority risk groups, as well those experiencing low parent/carer support, showed bigger differences for connectivity than the risk-absent groups. Furthermore, cumulative risk was a strong predictor of network connectivity. We conclude that network approaches to internalizing symptoms in adolescents can be useful to understand risks further. Gender and sexual minority groups, as well as those with low parent/carer support, may be particularly vulnerable and should be prioritized in prevention and intervention efforts. Our analysis also supports the idea that the number of risk factors experienced, regardless of their nature, predicts mental health outcomes. Young people experiencing multiple known risk factors should therefore receive additional support.
Marquez, J., Qualter, P., Petersen, K., Humphrey, N., & Black, L. (2023). Neighbourhood effects on loneliness among adolescents. Journal of Public Health
Loneliness is a growing public health concern, but little is known about how place affects loneliness, especially in adolescence. This is the first study to examine the influence of neighbourhoods and their characteristics on loneliness in early/middle adolescence. We used #BeeWell data and administrative data collected at the neighbourhood (Lower Super Output Area) level to study neighbourhood effects on loneliness for 36,141 young people across 1,590 neighbourhoods in Greater Manchester using a statistical analysis technique known as multi-level regression. Neighbourhoods accounted for a small but significant proportion of the variation in adolescents’ loneliness (1.18%). We also found higher levels of loneliness among older adolescents (vs younger), whites (vs ethnic minorities), girls and gender diverse individuals (vs boys), sexual minorities (vs heterosexual), and those eligible for free school meals (vs non-eligible). Moreover, we found evidence that ethnic, gender and sexual orientation inequalities in loneliness varied across neighbourhoods. Furthermore, loneliness was higher in neighbourhoods with higher skills deprivation among children and young people, lower skills deprivation among adults, higher geographical barriers, lower outdoor environment deprivation, and higher population density, as well as among those who have to travel further from home to school. Finally, more positive perceptions of the local area (feeling safe, trust in local people, feeling supported by local people, seeing neighbours as helpful, and the availability of good places to spend your free time) were associated with lower levels of loneliness. Our findings suggest that local-level interventions may be particularly helpful to tackle adolescent loneliness and reduce loneliness inequalities affecting vulnerable groups (e.g. those who identify as LGBTQ+).
Black, L., Humphrey, N. & Marquez, J. (2023). The influence of minority stress-related experiences on mental wellbeing for trans/gender-diverse and cisgender youth: a comparative longitudinal analysis. Royal Society Open Science
Trans and gender-diverse (TGD) young people are likely to experience poorer mental health and wellbeing than their cisgender peers. Minority stress theory suggests that this may be because increased exposure to factors such as bullying and discrimination lead to excess stress and reduced wellbeing. However, the evidence base remains limited. In this study we set out to examine these ideas using #BeeWell data. First, we examined differences in wellbeing between TGD young people and their peers. Second, we examined differences between these groups in exposure to minority-related stressors (bullying, gender discrimination, lack of family support, and feeling unsafe). Third, we examined the extent to which exposure to minority-related stressors predicted later wellbeing. Our analysis showed that compared to cisgender boys, TGD young people, cisgender girls, and those who preferred not to say their gender identity reported lower wellbeing (with the largest effect evident for the TGD group). TGD young people also reported higher rates of exposure to all minority-related stressors. We found that exposure to these stressors predicted later wellbeing (e.g., being bullied more in 2021 predicted lower wellbeing in 2022), but that this effect was not moderated by gender (e.g., the magnitude of the impact of bullying on later wellbeing was equivalent for all our gender groups). We discuss the implications of these findings in relation to things like coverage of gender identity in relationships and sex education guidance for schools, in addition to tailored services to support TGD young people.
