Anales del Congreso ALAP
Anales del X Congreso ALAP

The predictive capacity of statistics models using parameters derived from Facebook data for estimating migrant stocks has been largely documented for destinations including the US and the European countries. However, less is known about migrant stock estimates using Facebook data for destination countries in the Global South where internet penetration rates tend to be modest and truth ground timely data for comparisons scares. This paper provides two proofs of concept that respectively discuss the pros and cons of using Facebook data on users too i) approximate the magnitude and demographic profile of migrant stocks in the Latin American and Caribbean region; and ii) to sample and outreach migrant populations in the same region with surveying purposes. In the first case, we assess the correlation between Facebook number of monthly active users tagged as EXPATs and the United Nations estimates on the migrant stocks by country of origin and destination, sex, age and year of observation for 2019 and 2020. This analysis focuses on regional migration including 13 countries of origin and 21 destinations. In the second case we examine the accuracy of Facebook EXPATS tagging to outreach actual migrants by presenting the results on a survey conducted in Uruguay during August 2020 to check whether people tagged as EXPATS from country i by the app were actually recent migrants coming from i. Results show high correlation between Facebook data and UN migration stocks and a collection of parameters that grasp the systematic biases of Facebook which differ by origin, destination, age groups and sex. In addition, the results from the survey applied to EXPAT Facebook users from Cuban, Dominican, Peruvian and Venezuelan origin in Uruguay provide further evidence supporting the use of web social media data in predicting migrant stocks from the most numerous communities of origin while warning on the limits of such approach for examining smaller communities (Peruvians and Dominicans). Finally, the discussion section of the paper will review the opportunities and limits for estimation and sampling of regional migrant populations using Facebook data.