Search

How do mental models guide behaviour? Although a great deal of attention has been paid to mental models in psychology, relatively little is understood about how they influence incentivised decision-making or their relevance for policy. In this dissertation, I explore this question over the course of four papers. In the first paper, I synthesize the literature on mental models and aim to situate it in standard decision theory. In paper two, I look at how people use mental models of social groups to determine strategies of action under incentivized conditions. In paper three, I look at how unemployed young people's mental model of their agency within the world influences their decisions to search for work. In paper four, I look at how mental models of technology influence vaccine hesitancy.

This thesis consists of three standalone chapters that speak to the broad theme of responding to shocks in low- and middle-income countries. Chapter 1 investigates the impact of receiving cash transfers in anticipation of an extreme climate disaster, thereby enabling low-income households to cope with the shock on their own terms. We present quasi-experimental evidence showing the effectiveness of a timelier forecast-based cash transfer delivered to low-income households just before the peak of severe flooding in Bangladesh. We find that the `anticipatory' cash transfer enabled households to change their behaviour and improve their welfare, both during the flood and three months after the intervention. Chapter 2 explores the role of mental imagery in overcoming trauma - largely induced by exposure to violence - by fostering the ability to imagine the future. We design and test a ten-session entrepreneurship training curriculum that teaches mental imagery in business decision making through a randomised control trial in Colombia. We find that mental imagery can be taught as a skill and those who receive imagery training improve their earnings, relative to those who receive a placebo entrepreneurship training. Chapter 3 explores the interaction between public and private cash transfers, recognising that social protection may interact with existing coping strategies. I estimate the extent to which receiving an old-age pension crowds out remittances in post-Apartheid South Africa, using a regression discontinuity design. I find evidence of substantial crowding out in a middle-income country context: for every 1 rand of pension received, 0.22 rand of remittances are displaced on average.

Depression and risky behaviours often co-occur and represent two critical -but often overlooked - public health concerns. This co-occurrence is particularly important during adolescence, a time when individuals make vital decisions with often long-lasting consequences for their health, overall well-being, and life chances. The literature has explored the relationship between depression and risky behaviours mostly among adolescents in high-income countries. Therefore, it is an open question whether these findings can be generalised to low- and middle-income countries, where resources and conditions differ vastly. Examining this relationship is particularly important in low-resource settings, where poverty and everyday stressors may make adolescents even more vulnerable to the negative impacts of depression and risky behaviours. This thesis aimed to contribute to closing this research gap by pursuing four research objectives across five empirical studies. The first objective was to examine predictors of depression in a socio-economically disadvantaged rural community in South Africa, where most of the studies in this thesis are situated. Using a large cohort study of adolescents, Chapter 2 found significant and persistent effects of several risk factors on depressive symptoms. Many of these risk factors are common in low-resource settings, implying that a higher share of adolescents might be at risk of developing depression. This enhanced understanding of the contextual influences on depression led me to my second objective, which was to explore the relationship between depression and risky behaviours from different angles. To achieve this objective, I conducted two empirical studies. Chapter 3 presents the results from a systematic review and meta-analysis of the effect of depression on risky behaviours among adolescents in low- and middle-income countries. In line with the evidence from high-income countries, I found that adolescents with depression are more likely to engage in risky behaviours than non-depressed adolescents. Related to this objective, and given the evidence for a potential bidirectional relationship, Chapter 4 aimed to disentangle the direction of effect between depression and risky behaviours. The results from this chapter suggested that the relationship is complex and possibly dependent on specific individual characteristics. These findings informed my third objective, which moved beyond questions of associations to examining possible mechanisms underlying this relationship. To achieve this objective, Chapter 5 used a six-wave cohort study of adolescents to explore whether the relationship between depressive symptoms and risky behaviours is mediated through executive function. Results from the longitudinal mediation model showed that executive function did not mediate the association between depression and risky behaviours. However, I found that persistent depressive symptoms over four years were associated with poorer executive function in the next year, with the largest effect sizes observed for working memory. The fourth and final objective was to bring the evidence from all chapters together and explore possibilities for interventions targeting depression and risky behaviours among adolescents in low-resource settings. To this end, a smartphone application called the Kuamsha app was developed. Chapter 6 described the development of this app, which was co-constructed with more than 170 adolescents from rural South Africa and Uganda. While more evidence is needed to test the intervention's efficacy in reducing depressive symptoms and risky behaviours, preliminary results of early user testing showed that the Kuamsha app was acceptable in terms of usability and engagement. Overall, the findings in this thesis highlight the importance of studying depression and risky behaviours jointly, a relationship that has so far remained relatively unaddressed among adolescents in low- and middle-income countries. The results from this thesis also provide hope that interventions targeting depression can result in a virtuous cycle of increasing returns and promote adolescents' health in low-resource settings - a critical goal for global health research.