A recent article in Demographic Research uses Yandex as a novel data source for informing migration and mobility research in Russia. During a visit to the Leverhulme Centre for Demographic Science, Athina Anastasiadou explores the articles origins in Oxford and its main findings.
Athina Anastasiadou is a third-year PhD student from the Max Planck Institute for Demographic Research, and collaborated on this article with the Centre’s Dr Douglas Leasure and Artem Volgin from the University of Manchester.
Dr Douglas Leasure, Senior Researcher and Data Scientist at the Leverhulme Centre for Demographic Science and Demographic Science Unit said, ‘This project is a great example of talented students applying their passion to an important research topic, leveraging innovative data, and ultimately making a valuable contribution to the academic literature. It has been a pleasure to be involved in the project, and we are already looking forward to our next collaboration.’
In this new story, we ask Athina Anastasiadou how the project came about, what the role of Yandex is, and what the main findings of the article were.
How did the project come about?
Five months after Russia’s invasion of Ukraine, I arrived in Oxford for the Summer Institute in Computational Social Science (SICSS) 2022 with an idea to estimate the movement of Russian’s using digital trace data.
At SICSS 2022, Dr Douglas Leasure presented his work on using social media activity to monitor population displacement in Ukraine which inspired me to develop this idea further alongside Artem Volgin. Artem and I presented our project idea at SICSS 2022.
We were particularly interested in how the Yandex search engine could be used to predict the intentions of Russian nationals leaving the country following the start of the war in Ukraine. This idea continued to evolve after the conference, following regular conversations with Doug and Artem.
It has been great to see this project evolve from an idea at SICSS 2022 in Oxford to a fully-formed article published earlier this month in Demographic Research.
Athina and Artem presenting their initial idea at SICSS 2022
How did you use Yandex to predict movement intentions of Russian nationals?
We started by comparing mobility-related searches on Yandex with city-level data on sociodemographic and geographic characteristics three weeks after the start of Russia’s invasion of Ukraine. There was a noticeable spike of mobility-related searches on Yandex during this time. Whilst we do not know for sure whether this is people leaving the country, we did observe a correlation between Yandex searches and selected border crossings.
There was also a massive spike in mobility-related searches on Yandex three weeks after mobilisation was announced in Russia in September 2022. During this time, there was more movement and mobility-related searches than when compared to the start of the invasion.
Study figure 1: Mobility-related internet searches across Russia. Note: Mobility-related searches (share of total weekly searches) performed at the city level (grey lines) in Russia and their average (black line). Red dots indicate February 24, 2022 (beginning of the full-scale invasion) and September 21 (beginning of mobilization).
Interestingly, countries that were closer and easier for Russians to travel to, such as Turkey and Georgia, received way more Yandex searches during mobilisation than the invasion. We also found that the wage levels of Russian nationals were less important during mobilisation than invasion.
These findings support dominant media narratives about a potential Russian brain drain and the scale of movement outside Russia, after the start of the invasion and after the start of mobilisation.
Study figure 2: Geographic variation in the number of mobility-related web searches and links between origin cities and potential destination countries for the three-week periods following the Russian invasion of Ukraine (February 2022) and the announcement of military mobilization (September 2022). The size of the red circles indicates the overall increase in the number of mobility-related searches for each city, and the darkness of the blue lines indicates the increase in the number of searches from a city that specified each potential destination country. The increase is calculated as the difference between the number of queries three weeks after and three weeks before invasion or mobilization.
What was the outcome of this article?
Prior studies have found online search data can be a reasonable predictor for migration intentions. In our article, we make first steps in exploring Yandex online searches to hopefully enrich the scarce data landscape and open new avenues of research that utilise this predictor for mobility intentions.
By combining spatial data on socio-demographic characteristics with Yandex search engine data, we were able to better understand the mobility patterns of Russian nationals from Russia after the invasion of Ukraine. Our findings confirm the potential of Yandex search engine data as a novel data source for informing migration and mobility research.
Athina’s bio
Athina is a third-year PhD student from the Max Planck Institute for Demographic Research where she works at the Lab for Migration and Mobility with Emilio Zagheni. She is also affiliated with the Netherlands Interdisciplinary Demographic Institute and the University of Groningen. Her research is placed in the field of Digital and Computational Demography. In particular, she is interested in the application of Digital Trace Data for monitoring sudden or conflict-induced migration events. Her PhD research focuses on gender- and sex-based differences in the migration process. Therefore, she is exploring gender- and sex-induced biases of statistical models for migration predictions as well as methods to address the lack of sex- and gender-disaggregated migration data. Athina visited Nuffield College as a SICSS participant in July 2022 and is currently an Academic Visitor at the Leverhulme Centre for Demographic Science.
