|
|
Victoria Prieto Rosas , Programa de Población, Universidad de la República
Camila Montiel, Centre d'Estudis Demogràfics
Martin Koolhaas, Programa de Población, Universidad de la República
Mariana Fernández-Soto, Programa de Población-Universidad de la República
Francis Silvera, Programa de Población, Universidad de la República
Julieta Bengochea, Programa de Población, Universidad de la República
This paper assesses the correspondence between Meta web social media data (Facebook, Instagram, and Messenger) and traditional demographic sources—population censuses and household surveys—in estimating total and migrant populations in Latin America. Using data extracted from the Meta Marketing API aligned with census and survey periods for Argentina, Ecuador, Mexico, Paraguay, and Uruguay, the study evaluates the strength and consistency of associations between Meta daily active users and official counts. Multivariate regressions were estimated for both total and migrant populations, controlling for origin, destination, sex, and Meta app penetration rates derived from the Latinobarometer survey. Results reveal that Meta user estimates are a strong and consistent predictor of both census- and survey-based measures. However, the fit is notably higher with census data than with household surveys, both for total populations and for migrant stocks. Among migrant populations, the relationship remains significant even when accounting for Meta penetration at origin and destination, suggesting that coverage and digital habits at either end do not fully explain the association. These findings indicate that Meta data track census figures closely and provide valuable, timely information for migration research in contexts where traditional data are limited or delayed. The results also challenge the assumption of censuses and household surveys as “gold standards,” given their growing omissions and inconsistencies, particularly in measuring foreign-born populations.
Presented in Session 26. Flash Session Emerging Data Sources in Demography: Digital Traces, AI and Mobile Phone Data