Digital Innovations and Education: Assessing the Potential of Digital Technology in Learning, Well-Being, and Research

This dissertation examines how digital technology can be used for monitoring and intervention in three domains: education, mental health, and the scientific process itself. Chapters 2 and 3 report results from a nationwide RCT in Ecuadorian high schools during the COVID-19 pandemic. Chapter 2 evaluates interventions at the student, teacher, and system levels and finds a clear hierarchy of effectiveness: a centrally managed digital learning system increased study time and academic performance. Chapter 3 evaluates digitally delivered soft-skills and hard-skills courses, finding sizable gains in course-related knowledge and some psychological outcomes, but limited evidence of effects on post-graduation outcomes. Chapter 4 uses the project's survey and administrative data to show that a growth mindset predicts each step of the university admission process. Chapters 5 and 6 are based on a digital mental health support project after the 2022 full-scale invasion of Ukraine. Chapter 5 uses an online platform to rapidly assess the mental health of Ukrainians seeking psychological help, documenting very low well-being overall and particularly high distress among refugees and internally displaced people. Chapter 6 presents a multi-country RCT of "Self-Help Online," a digital adaptation of the WHO's Self-Help+ stress-management program. The intervention reduced psychological distress and improved well-being among refugees. Chapters 7 and 8 study doctoral research output in Germany. Chapter 7 shows that cumulative, English-language, and empirical dissertations are more likely to result in peer-reviewed publications with higher total citation counts. Chapter 8, using ~70,000 manually collected test statistics from dissertations and ~26,000 from their follow-up papers, finds little evidence of p-hacking and that early-career researchers report fewer tests in papers than in dissertations without evidence of selection on statistical significance. Together, the dissertation demonstrates that system-level digital monitoring can quickly generate actionable data, digital interventions can deliver knowledge and psychological first aid at scale, and that representative digital records enable a more unbiased view of research practices. Its central policy implication is that rigorous evaluation, ideally through randomized trials, should precede the adoption of digital tools.