Séminaires LEMMA


"Ideology and Information Cascades: Evidence from French Politicians' Tweets "

Margherita COMOLA (Université Paris-Saclay), joint with S. Shabayek

Tuesday, 14 October 2025, 11h-12h

Lemma - 4 rue Blaise Desgoffe, 75006 Paris. Salle Maurice Desplas

Margherita is a Professor of Economics at Université Paris-Saclay, Junior Member of the Institut Universitaire de France (2021–2026), Affiliate Professor at PSE, and Research Fellow at IZA. Her research spans networks, econometrics, experimental, digital, and development economics.

AbstractThis paper investigates the diffusion of political messages within the French online political community by combining a novel administrative dataset with complete Twitter data for politicians active between 2021 and 2022. We focus on the role of ideology in shaping retweet cascades, distinguishing between identity bias (systematic retransmission within and across political affiliations) and topic bias (systematic retransmission depending on message content). Using a duration model of retweeting behavior, we show that both identity and topic biases strongly drive information diffusion, though with heterogeneous patterns across political communities. 

 

"Learning Through Living Rooms Screens: Experimental Evidence from Kenya"

Nicolas BOTTAN (Cornell University)

FRIDAY, 10 October 2025, 11h-12h

Lemma - 4 rue Blaise Desgoffe, 75006 Paris. Salle Maurice Desplas

Nicolas is an assistant professor in economics at Cornell. His research interests are in the fields of Public, Behavioral and Development Economics.

AbstractWe present experimental evidence on the effects of over-the-air educational TV on literacy and socio-emotional skills. In a randomized encouragement design with 4,300 Kenyan children, watching a literacy-focused show increased reading comprehension by 9% and curiosity by 12% of a standard deviation, with stronger effects among English-speaking households. Studying mechanisms, we find that the effects are driven by the show's content itself, not by changes in time use, preferences, or parental inputs. However, the program inadvertently reinforced traditional gender norms. At $18.48 per standard deviation, it is highly cost-effective. Results highlight both the promise and limits for educational media.