Using Facebook data to track migration
Can Facebook data be used to gain insight into international mobility patterns? This is the question the authors of a recently published have posed.
Quantifying international human mobility patterns using Facebook Network data looks at how the international community can improve the quality of international migration statistics.
The point authors Spyridon Spyratos, Michele Vespe, Fabrizio Natale, Ingmar Weber, Emilio Zagheni and Marzia Rango make is that non-traditional data sources can provide insights that improve migration governance and this can assist policy-makers.
It says estimates made using Facebook data can and do add value. The caveat however is that it must be possible to validate these using third party data.
The simplest way to do this is to cross-reference data gleaned on Facebook with official statistics and reliable data sets from sources like the Migration data portal or the International Organisation on Migration.
The authors have completed a comparative analysis of the number of estimated “Facebook migrants” from two separate time periods.
This was done on the assumption that the Facebook advertising platform’s classification of users as having “lived in country X” refers to a country other than the country they currently reside in. The reason for this is not too complicated. Facebook relies on the country of a user’s hometown to classify them as “having lived” in a country and this is usually where they were born.
They compared this data with statistics from the US Census Bureau’s annual American Community Survey, Eurostat migration statistics for 18 EU and four European Free Trade Association countries.
Other useful sources are the OECD Database on Immigrants in OECD Countries and the extended Database on Immigrants in OECD and non-OECD Countries.
All these data sets provide information related to population categorised according to age, sex, and country of birth - something they have in common with Facebook’s Marketing Application Programming Interface.
The authors found that when used rigorously Facebook estimates have the potential to assist policy-makers in three important areas.
First, they can help them anticipate emerging trends and offer much-needed support to early-warning mechanisms. They can also be used to facilitate the monitoring of migration at the national, sub-national, and local levels.
“Second, these estimates can provide a finer description of the characteristics of the migrant population. Third, these estimates can offer a wider definition of migrants that captures forms of migration that do not appear in the official statistics,” the study says.
How they tracked migration using Facebook data
The impact that distortions in data from Facebook’s Marketing Application Programming Interface can bring in cannot be overstated.
This is a problem the authors encountered very early in the study. They found that Facebook data on migrants from the US to Indonesia was overestimated by 11,520% compared to United Nations’ Department of Economic and Social Affairs or OECD figures.
“This overestimation may be due to fake accounts linked to “click farms” based in Indonesia,” the study observes.
When all the above pitfalls have been overcome the study found that the discrepancies seen when comparing Facebook estimates can plug critical information gaps. For one, the may represent trends not yet captured in official statistics.
They were certainly useful in identifying and quantifying an increase in migrants from Venezuela to Spain that had been reported in several sources but did not appear in official statistics.
Stocks of international migrants and Facebook migrants from Venezuela in Spain. Source: Spyridon Spyratos
The project used Facebook data to anticipate the increase in Venezuelan’s in Spain, and this trend had since been verified by the 2019 statistics from the National Statistical Office of Spain.
Can data from social media companies be trusted?
Social media companies that gather or provide data on demographics are commercial enterprises and questioning their intentions and autonomy is important.
In reality the issue boils down to how independent the data is and whether it is suitable for reuse outside of commercial interests.
Can the data be relied on when reused for input for policy-making rather than advertising?
This a concern the study shares and discusses openly. “At any time, social media companies may change the conditions for accessing the data or decide to change the types of available variables and their definitions,” it says.
In addition to commercial interests, third parties might also try to manipulate content to promote a hidden or even nefarious agenda. The study notes that anti-immigrant interest groups can influence public debate on migration.
One way they can do this is by increasing the number of “Facebook migrants” from specific backgrounds through fake accounts.
Politicians and nation-states have tried to interfere in election processes over the past few years, including the American election of 2016.
An observation the Freedom House’s Freedom on the Net 2019 report makes is that digital platforms have become a new battleground for democracy.
According to Freedom House, incumbents use both blunt and nuanced methods to deter opposition movements. They do this while still preserving a veneer of popular legitimacy, Freedom House said.
Twitter’s General Counsel Sean Edgett was also forced to admit to how troubling the misuse of the platform corrodes public faith in democratic processes.
It’s worth remembering that not so long ago Twitter’s General Counsel Sean Edgett was forced to admit to how troubling the misuse of the platform corrodes public faith in democratic processes.
The fact that leading politicians in India deployed bots to spread disinformation also shows how the subversion of social media platforms can be a major issue for big democracies with a digital savvy population.
Using social media data to track movement
The use of demographics data from non-traditional sources as an emerging theme is slowly gaining traction.
In 2015 a team of researchers analysed whether geotagged photographs uploaded on Flickr could be used to quantify international travel flows into the United Kingdom.
They found evidence of a correlation between the number of Flickr users visiting the UK from different countries and official visitor estimates calculated by the UK Office for National Statistics.
Moreover, the data on the time and location at which the photographs were taken can be used to generate more accurate estimates than data from conventional socio-economic indicators.
Another important revelation was that when done rigorously this is an important alternative to data gathering using expensive and rather time consuming tools like surveys.
Data from these non-traditional sources also has the potential to enrich debates on vexed policy reform questions.
According to the International Organisation on Migration, sensible analysis of the “digital crumbs” from internet-based platforms can offer important insights into societal phenomena.
One example is that data from LinkedIn lends itself to digital mapping of the global workforce that includes occupational profiles and migration patterns.