Physical activity (PA) has been labelled the miracle drug (Pimlott, 2010) and participating in regular PA has ample physical and mental wellbeing benefits. However, physical inactivity remains a critical public health concern, particularly across adolescence. In England the proportion of adolescents aged 13-15 years meeting the recommended guidelines for PA decreased significantly from those at a younger age (Health and Social Care Information Centre, 2009; 2012; 2015). The adolescent years (13 18 years) have been identified as the age of greatest decline in PA, although it is possible that large declines can also be seen at younger ages (Sallis, 2000). Among girls the decline in PA is greater at younger ages (9 12 years old) and among boys it is greater at older ages (13 16 years old) (Dumith, Gigante, Domingues, & Kohl, 2011). Thus, examining behaviour of early adolescents (aged 11-13 years) is a primary focus of this thesis. Researchers have called for a more comprehensive grasp of PA correlates and determinants and their impact on behaviour (Biddle & Mutrie, 2001, 2008). This broader picture needs to incorporate longitudinal study designs to accurately portray developmental changes (Evenson & Mota, 2011). This thesis aims to work towards a better understanding of associations among variables across aspects of the ecological model in relation to PA behaviour during early adolescence. Early adolescents within their first year of secondary school (year 7, aged 11 12 years) were recruited through schools across the East Midlands, United Kingdom (UK). These participants completed various measures across an 18 month period to compile all data required for the thesis. The thesis begins with a focus on active transport as a means of commuting to school which can significantly contribute to overall PA levels (Aibar, Bois, Generelo, Bengoechea, & Paillard, 2015; Slingerland, Borghouts, & Hesselink, 2012). The distance from home to school is an important influence on the decision to use active transport; however, ecological perspectives would suggest this variable may interact with individual, interpersonal and environmental factors. Therefore, the first study of this thesis investigates whether the relationship between distance to school and active transport is moderated by (i) gender, (ii) biological maturation, (iii) perceived family support for PA and (iv) multiple deprivation. Cross-sectional results from the baseline data collected demonstrated that the relationship between distance to school and the likelihood to actively travel to school is moderated by biological maturation, multiple deprivation and family support of PA in adolescents. Further analysis revealed that late-maturing children, those from less socio-economically deprived backgrounds and children with low family support of PA are less likely to actively commute to school as distance to school increases. Due to the interaction between these variables described above, the second study focused on the variables collectively using a person-oriented approach, which aimed to classify distinct profiles of early adolescents based on correlates of PA. The outcome variables were also broadened to include active transport and overall PA across two time points. Findings from this second study illustrate that the highly supported, shortest commuters produced the highest levels of self-reported PA and that affluent, short commuters were the most likely to use active transport to travel to school. The affluent, short commuters lived a relatively short distance to school in areas of the lowest deprivation and had relative moderate family support of PA. The highly supported, shortest commuters were characterised by the highest family support of PA and lived the shortest distance to school in areas of low deprivation. Study 1 evidenced an association between biological maturation and PA behaviour; however, study 2 displayed that biological maturation did not meaningfully contribute to the class characteristics, and were not a predictor of PA. Previous evidence as to whether early, average or late maturing adolescents are more likely to disengage from PA is mixed and tends to focus on one gender only (Sherar, Cumming, Eisenmann, Baxter-Jones, & Malina 2010; Bacil, Mazzardo, Rech, Legnani, & Campos, 2015). Thus for the third study a more focused inspection of biological maturity was undertaken. Biological maturity status was investigated as a predictor of PA behaviour at two subsequent time points (6 9 months after baseline and 12 18 months after baseline) and whether there was variation across genders. Findings displayed that biological maturity status does not predict subsequent PA, with no distinction across genders. To conclude, the final study examined additional forms of PA behaviour. For children to develop and maintain healthy PA behaviours, their PA during the school day, particularly during physical education (P.E) classes is important (Owen, Smith, Lubans, Ng, & Lonsdale, 2014). Self-reported PA was divided into school-time PA (during P.E. lessons, break and lunchtimes) and leisure-time PA (after school, during evenings and weekends). The final study fully utilised the longitudinal data collected and utilised longitudinal growth modelling to describe the changes in PA behaviour across 12-18 months during early adolescence. Results displayed that school-time PA and leisure-time PA are distinct. Males; those from less deprived backgrounds and individuals with higher family support of PA all separately reported more school-time PA than their counterparts (females, those from higher deprived backgrounds and individuals with lower family support of PA) at baseline. Males and those with higher family support of PA also reported more leisure-time PA than their respective counterparts at baseline. On average, both genders decreased in school-time PA across 18 months yet for leisure-time PA, on average, there was no change over time and no significant difference in the rate of change between genders. There were no observed significant differences in the rate of change between multiple deprivation status and biological maturation across the 18 months for both behaviours. For family support, on average school-time PA decreased over time and results showed significant difference in the rate of change between individuals with lower or higher levels of family support of PA across the 18 months. On average, there was no change over time for leisure-time PA yet there was a significant difference in the rate of change between individuals and their family support of PA across 18 months. Further analysis demonstrated if an individual s family support increases, so does their leisure-time PA and vice versa. These overall key findings demonstrate the complexity of PA behaviour throughout early adolescence. This thesis works towards predicting individuals, correlates and determinants that may be susceptible to physical inactivity and/or a decrease in activity over time. Results can be used to target and direct PA intervention work.

Odds ratios and risk ratios are useful measures of effect size in 2-group studies in which the response variable is dichotomous. Confidence interval methods are proposed for combining and comparing odds ratios and risk ratios in multistudy designs. Unlike the traditional fixed-effect meta-analysis methods, the proposed varying coefficient methods do not require effect-size homogeneity, and unlike the randomeffects meta-analysis methods, the proposed varying coefficient methods do not assume that the effect sizes from the selected studies represent a random sample from a normally distributed superpopulation of effect sizes. The results of extensive simulation studies suggest that the proposed varying coefficient methods have excellent performance characteristics under realistic conditions and should provide useful alternatives to the currently used meta-analysis methods.

A handful of studies have shown that National Inter-religious Councils (NIRCs) contribute to electoral violence prevention. However, no quantitative study has evaluated the impact of their interventions and the conditions under which they lower electoral violence. Using data on African national elections from 1992-2019, I examine whether NIRCs' electoral-related peacebuilding interventions lower the severity of electoral violence and if their ability to do so depends on NIRCs' social power. The results in this thesis indicate that the presence of NIRCs' interventions during electoral rounds lowers the likelihood of severe electoral violence. The findings also reveal a significant disordinal interaction between NIRCs' interventions and their power, where the predicted probability of severe electoral violence reduces by 42% when powerful NIRCs implement peacebuilding interventions and, on the other hand, increases by 20% when less powerful NIRCs intervene. While this thesis fails to explain the disordinal interaction effect, it details this study's methodological tools and limitations and contributes an original dataset of NIRCs' interventions and power for future research.