Three experiments on technology, inequality and ideology

Roldán-Monés, Antonio (2024) Three experiments on technology, inequality and ideology. PhD thesis, London School of Economics and Political Science.

Text - Submitted Version Download (6MB)

Abstract

This dissertation consists of three Randomized Controlled Trials analyzing the relationship between technology and inequality as well as the ideological constraints to evidence-based policy implementation. In the first chapter I investigate the impact of a Generative Artificial Intelligence system (GenAI) on productivity and inequality, examining a novel set of skills in a unique setting: a university debating competition. Early experimental research in the area highlights two key findings: GenAI significantly boosts productivity in a variety of tasks, and it disproportionately benefits lower-performing workers, potentially reducing work inequalities. However, most studies in this literature are focused on written tasks. Contrary to previous studies, my findings reveal that GenAI’s effects vary considerably when higher-order skills like persuasion or social perceptiveness are needed. My results show that GenAI exhibits stronger complementarities with top performers than with low performers at debating, suggesting that GenAI could reinforce inequalities depending on the context and the skills required. The second chapter explores an innovative pedagogic design to address educational inequalities through online learning in a post-pandemic context. While previous research has shown the effectiveness of in-person tutoring, evidence for 100% online environments was very limited. Our study demonstrates that a cost-effective 2-students-to-1-mentor online tutoring model significantly improves maths learning outcomes for children in vulnerable settings. The intervention significantly increased standardised test scores (+0.26 SD) and end-of-year maths grades (+0.49 SD), while reducing the probability of repeating the school year. The intervention also raised aspirations, as well as self-reported effort at school. The two-on-one design allows us to significantly reduce costs and improve scalability, while showing similar results as one-on-one tutoring programs. The third chapter studies political constraints to scientific communication. Utilising a nationwide experiment, the paper examines the role of political alignment between knowledge brokers (such as think tanks and media outlets) and real policymakers in the adoption of an evidence-based policy. Results show that policy adoption increases by over 65% when the informing institution aligns ideologically with policymakers, while opposite ideologies do not affect adoption. Both think tanks and newspapers are equally effective when aligned ideologically. Finally, a proposed three-stage framework reveals that ideological alignment influences belief updating but not the decision to get exposed to new evidence among politicians.

Item Type:	Thesis (PhD)
Additional Information:	© 2024 Antonio Roldán-Monés
Library of Congress subject classification:	H Social Sciences > H Social Sciences (General) H Social Sciences > HC Economic History and Conditions L Education > LB Theory and practice of education T Technology > T Technology (General)
Sets:	Departments > Social Policy
Supervisor:	Özcan, Berkay and Velasco, Andres and Oliver, Adam
URI:	http://etheses.lse.ac.uk/id/eprint/4842

Actions (login required)

Record administration - authorised staff only