Official migration statistics are developed and published by national offices of statistics and collated by international organisations. These statistics are based on rigorous internationally harmonised principles, but they come with a considerable time lag. New data sources offer opportunities to complement traditional sources for migration statistics. In particular, availability of high quantities of individual geo-located data from social media has opened new opportunities. In this research, we develop probabilistic methods to combine provide traditional and social media bilateral migration data to estimate timelier and potentially more accurate migration statistics by accounting for measurement parameters of each source. Bayesian methods offer a powerful mechanism to combine data sources. Previously models have been developed for solely combining traditional migration data sources using the prior models for measurement parameters. We adapt the basic methodologies of these former models to combine migration data from both traditional and new data sources derived from social media.
Presented in Session 103. Innovative Approaches, Data, and Analytical Strategies in the Study of Migration