This blog is part of the COMPAS Coronavirus and Mobility Forum.
The COVID-19 crisis could be a game-changer for big data analytics. Several posts in this blog series have already illustrated the many facets as well as the political and social implications of digitisation in contemporary migration and border regimes: be it the use of AI for immigration decision making, the establishment of digital warfare in border zones, or the power of new technologies of immobility.
In this blog post, I would like to draw attention to an aspect of this development that has so far gone almost unnoticed, but that may well turn out to have far-reaching implications for migration studies. In a nutshell, migration research is facing important methodological challenges. These challenges are linked to a still nascent, but clearly discernible epistemic reconfiguration: the launching of “big data alliances” between computational sciences, supranational organisations, and national statistics institutes. These alliances aim to explore and realize the potential of new data sources and novel analytical techniques in the field of migration.
Migration scholars should engage with these new tendencies proactively. I see two main reasons to do so: First, there is an urgent need for social science reflectivity to control algorithms and new technologies regarding their (very likely) detrimental effects; second, as risky and as challenging as new data worlds might be, they also hold important promises for migration research.
The development of these “migration data alliances” has been rapid over recent years. The European Commission and IOM launched “BD4M” as a collaborative platform in 2018 , following a number of earlier achievements such as the establishment of the Migration Data Portal (2017) or opening of IOM’s Global Migration Data Analysis Center (GMDAC) (2015).
The COVID-19-induced public hunger for big data may turn out to be a powerful catalyst for further escalation of these initiatives. This is partly due to the increased political pressure to use new technologies for monitoring and controlling international mobility on a large scale. Perhaps even more important is the strong link between epidemiology and mobility. This connection is, of course, not new. The closing of borders and the immobilization of parts of the population have been key (and contested) “public health measures” for centuries. But the new ability to track population movements and algorithmically classify individuals almost instantaneously opens completely new opportunities for controlling borders and mobilities. Data from mobile phone providers and social media are already used for these purposes. These showcases have very obviously gained momentum over the past few weeks.
Based on my own research on these new alliances, I think that it would be a misunderstanding to interpret them primarily as signs of methodological innovation. Rather, they mark a reconfiguration of politics, social theory, and epistemology. On closer inspection, the methodological role of big data and data analytics is actually really marginal. Even the most state-of-the- art data platforms rely mostly on traditional census and survey data, and the kind of questions asked and of statistical analyses done follows well-established patterns.
The crucial point seems to be another one: There is quite a sharp rupture between new data alliances and contemporary migration studies. Migration scholars rarely take part in big datavconferences and workshops, nor do these alliances refer to or engage with contemporary migration studies. The questions asked and answers given thus seem oddly out of sync with present social science debates. Most importantly in the current context of datafied migration politics, critical scholarship on border and migration regimes, on post-colonial racism or the like are completely neglected.
From a social science perspective, most existing computational science studies of migration issue may hence seem theoretically uninformed, politically naive, and sometimes trivial. They nonetheless resonate well with powerful political and institutional actors. To put it bluntly, their more or less atheoretical character makes big data initiatives quite compatible with political discourses. While social sciences have grown increasingly critical of how migration issues are politicized by dominant forces, computational scientists seem to find it easier to speak the language of political actors such as intergovernmental organisations and national governments.
This is the very real danger: That the locus of legitimate interpretation of migration issues shifts from research based in theoretical and empirical reflexivity to atheoretical and affirmative approaches. In other words, we might be facing a serious struggle over who is considered competent and responsible for producing knowledge on the social world. Of course there is nothing wrong with interdisciplinary perspectives and with computational sciences being engaged in studying social dynamics. But the risk is that decades of critical social research and theory building are lost sight of in the process.
Whether or not these data initiatives will gain momentum and influence, migration scholars need to start engaging with new data worlds. After all, migration is already linked to the digital world in many ways and this requires novel methodological approaches. The emerging digital migration studies offer some starting points and an increasing number of empirical and theoretical insights that deserve attention by a broad academic audience.
There is another reason for taking a proactive approach vis-à-vis the mentioned developments: big data and digitisation may actually help to overcome challenges that have burdened migration studies for decades. For example, many traditional (especially quantitative/statistical) data sources on migration issues are deeply marked by an inherent methodological nationalism. This is far less the case for most forms of big data. Big data could hence let us ask new research questions, or at least to answer old ones in new ways.
For example, statistical “mining” techniques (typical for big data analytics) could be employed for exploratory analyses of social media data in order to analyse ideas of cultural belonging or processes of social boundary-making from the bottom up (rather than forcing respondents to identify with predefined ethnic categories, as is commonly done in survey research and census questionnaires).
Such a research strategy would at the same time open new perspectives for integrating qualitative and quantitative methods. After all, the analysis of big data is not limited to number crunching. Starting with the need to actually understand the social contexts in which data are produced, big data require inductive research logics more akin to qualitative methodologies. In turn, new data worlds offer qualitative researchers completely new possibilities of linking their deep understanding of social phenomena to larger scale patterns. These opportunities come with risks and costs (from gaining data access and privacy protection issues to acquiring new methodological skills).
In the face of recent developments, the question is not whether but how to deal with these challenges.