209 General Discussion could be assessed in this way but are not. The reasons are multiple and vary according to the discipline being discussed. Chapter 6 reflects on many of these challenges, including unpredictability associated with the implementation of the intervention (e.g. timing, intensity and reach), challenges with data linkage, and the need for disciplines such as public and global health to embrace the potential of quasi-experiments better. Some of the perspectives shared in this Chapter do not necessarily apply to the study of quasi-experiments in the context of applied economics. Fields such as health economics, economics of education, and development economics have specialised in using econometric methods to gauge causal effects. Other challenges arise in the case of these fields, namely a strong emphasis on defending the exogeneity of the intervention being studied of all potential threats. This means the ability to exclude all possible violations of identifying assumptions that researchers and peers might think of, despite their potential to change the results. These strong concerns with potential endogeneity reduce significantly the set of exposures eligible for quasi-experimental evaluations, particularly when aiming to publish in peer-reviewed economics outlets. In other words, having no evidence about the interventions is preferred over having evidence that might gauge causal effects at the risk of bias, and both quantitatively and qualitatively assessing these biases. This conservative approach widens the gap between research and policy and limits quasi-experiment potential. In summary, the experiences gathered through conducting and publishing the different chapters in this thesis suggest that boosting the role of quasi-experiments requires researchers from various disciplines and policymakers to better acknowledge the continuum towards causal evidence better. And to be trained in interpreting and discussing it. This continuum can be illustrated by looking at the chapters of this thesis. Chapters 3 and 4 correspond to robust quasi-experiments in which identifying assumption violations were extensively studied and excluded. Chapter 5, however, is an example of a causal research question in which the quasi-experimental design initially planned did not succeed due to the lack of detailed data to measure the exposure, and the small magnitude of exposure itself (e.g. number of psychologists hired during the study). As an alternative, the authors decided to pursue the study and use a method that partially addresses some endogenous concerns, positioning findings in the continuum from association to causal effects. REFLECTIONS ON CONDUCTING INTERDISCIPLINARY RESEARCH WITH SOCIETAL IMPACT Interdisciplinary research and the challenges of bridging academic disciplines In the methodological considerations section, I have reflected on different approaches to quasi-experiments by the health sciences and economics disciplines. The differences 7
RkJQdWJsaXNoZXIy MjY0ODMw