The quote in the title comes from a Radio National interview (Australia) conducted on the Breakfast program on 24 November 2021 (Yes, I was walking). The interviewee was asked about women’s excessive workload, at home and elsewhere, during the pandemic. She replied: “The data is in”, and it revealed that women did the bulk of home schooling, etc.
Since being prompted to write about the topic of “data” and research by a participant in the recent “Kick-off” event for the planned international WPR symposium in August 2022 (https://www.kau.se/statsvetenskap/forskning/international-symposium-critical-policy-studies-exploring-premises-and I have paid particular attention to the ways in which the term “data” appears in our conversations and in public pronouncements, by government spokespeople and researchers (see last entry 29 April 2022). As with many of the other topics I’ve broached over the past four years in this Research Hub, enquiring into the status of “data” as research material opened a Pandora’s box.
I decided to explore the issue under two broad interconnected headings:
- The place of “data” in governing (this entry)
- The place of “data” in research (subsequent entry)
This distinction is drawn purely for heuristic purposes. From the outset I need to emphasize that how researchers “use” “data” needs to be conceived as a part of governing, as a governing practice. Drawing upon Foucault’s notion of governmentality, governing is seen to involve a broad array of agencies and groups, including professionals, experts and researchers.
On this point, it is helpful to keep in mind Annemarie Mol’s (2002: 155, emphasis in original) argument that: “Methods are not a way of opening a window on the world, but a way of interfering with it. They act, they mediate between an object and its representations”. Therefore, the production of “data”, through research, is understood to be a form of political practice that creates “realities” (see Bacchi 2012: 143). In this view, “data” operate as governmental technologies, mechanisms through which governing – understood broadly – takes place (Bacchi and Goodwin 2016: 44), as elaborated below.
With this backdrop, I advance two propositions:
first, that “data” is produced and hence it is important to reflect on the means of its production; and
second, that “data” is productive – that it produces “things” including “objects”, “subjects” and “places” (see Benozzo et al. 2013: 309).
Readers may recognize that here I am questioning (or problematizing) “data” much in the same way I problematize “problems” in WPR – where I point out that “problems” do not simply exist out there waiting to be “solved” but that they are produced in policies as particular sorts of problem (Bacchi and Goodwin 2016: 14). I also emphasize in WPR that how “problems” are produced involves the categorization (and making “real”) of “objects”, “subjects” and “places”. Both in reference to “problems” and to “data”, therefore, the emphasis is on the production of “reality” as a performative practice (see BACCHI KICK-OFF PRESENTATION). In poststructuralism, “a performative is that which enacts or brings about what it names” (de Goede 2006: 10)
“Data” and “problems” are linked in a second way. “Data” are generally put forward to solve “problems” – assumed to exist as clear and unquestionable states or conditions. As Valentine (2019: 365) describes, “Simply put, algorithms are mathematical processes for solving defined problems”. Hence, “data” form part of what I call the problem-solving knowledge that dominates the contemporary intellectual and policy landscape (Bacchi 2020; see also Edwards et al., 2021). The common use of “data” as “evidence” in evidence-based policy approaches confirms this link (Street et al. 2021: 1). Morozov (in Schüll 2013) labels this tendency “technological solutionism” – recasting complex social phenomena such as politics, public health, education and law enforcement as “neatly defined problems with definite, computable solutions”. As a contemporary example of “technological solutionism”, consider the former Australian Prime Minister, Scott Morrison’s, “action plan” for climate change which he characterized as reliant on “technology, not taxes” (https://www.abc.net.au/news/2021-04-23/pm-defends-his-technology-not-taxes-approach-to-emissions/13315616
Generally, “data” have a taken-for-granted status as unquestioned information (Gitelman and Jackson 2013: 10). In addition, “data” are treated as the basis of knowledge, understood as “truth”. The word “data” sits alongside allied terms such as “facts” and “evidence”. If you wish to “prove” something, you need to be able to provide back-up “data”.
This conception of “data” lines up with a correspondence view of knowledge. “Data” are “found” in the “real world” and hence provide “knowledge” of that world – assumed to exist as fixed and inert, waiting to be known. In other words, “data” is not an innocent term. It brings with it an ontology and an epistemology, a way of thinking about “reality” and how to “interact” with “it”.
Foucauldian-influenced research offers, as an alternative to realist knowledge, a focus on how knowledges (note the plural form; see below) are produced. As argued in an earlier entry (29 Nov. 2021), in Foucauldian-influenced perspectives, “truth” is always situated. There is no universal basis for “truth”. Rather, “truth” and “knowledge” are produced in “‘local centres’ of power-knowledge” (Foucault 1990: 98). The analytic task, therefore, involves seeking out and examining the multitudes of practices – the “processes, procedures and apparatuses” (Tamboukou 1999: 202) – involved in the production of “truth” (or “truths”), rather than (simply) to uncover what is concealed. The goal becomes showing how political practice takes part in the “conditions of emergence, insertion and functioning” of “regimes of truth” (Foucault 1972: 163).
What does it mean to track how “data” are produced? What sorts of processes do we need to identify? There are several levels at which this topic could be pursued. At one level it is possible to draw attention to where data come from (Fordyce and Jethani 2021: 3) – who asks the questions that produce data? What authority do they have? At this level, there are expressed concerns to regulate or oversee the practices involved in the production and control of “data” in order to monitor and regulate surveillance (see Raley 2013). Padden and Öjehag-Pettersson (2021) usefully subject one such declared attempt at regulation, the EU GDPR (General Data Protection Regulation), to critical analysis using WPR.
At a second level, “data” are produced as particular kinds of objects because they are organized in specific ways. “Data” do not speak for themselves. Through categorization and visualization, they are made to speak. Data are not just found; they are imagined (Gitelman and Jackson 2013: 3) and generated (Manovick 2020).
Furthermore, the forms of categorization and organization that produce “data” have productive (constitutive) effects – creating “subjects”, “objects” and “places” of specific kinds. Through these effects, “data” play an active governing role, forming what governmentality scholars refer to as “technologies of rule” (Bacchi and Goodwin 2016: 44). In this view governmental instruments such as censuses, league tables and performance “data” are involved in “the conduct of conduct” (Gordon 1991: 2).
For example, Rowse (2009) explores how the current Australian census arranges the “population” into categories (such as Indigenous and non-Indigenous) with repercussions for the sorts of political claims that can be made. In this account, numbers operate, not as a neutral, statistical register of the “real”, but as involved in shaping political possibilities. They are “integral to the problematisations that shape what is to be governed, to the programmes that seek to give effect to government, and to the unrelenting evaluation of the performance of government that characterises modern political culture” (Rose 1991: 175; see also Miller 2001).
Illustrating this point Harrington (2021) shows how the UN Secretary-General’s requirement to report on SEA (sexual exploitation and abuse) “data” shapes solutions: “The reports reflect managerial audit systems that produce a performance culture in which the goal becomes performing well for the audit”. Audit systems demand quantifiable progress in “solving” assumed “policy problems”: “Audits of not-readily quantifiable performances must contrive ways to quantify them, often with far-reaching consequences”.
Instead of reporting on the “real”, “data” – numbers and statistics – are involved in constituting “the real”. They are involved for example in the production of an “urban/rural” dichotomy, and in the distinction between “developed” and “developing” “places” (Bacchi and Goodwin 2016: 100-104). Walters (in Tietäväinen et al. 2008: 67) describes how “Europe” has become a “domain of statistics, calculation and projections”, firming up its existence as an entity.
Fernandez (2012: 57) reminds us that the focus on statistics transforms a political issue into a technical one, described above as “technological solutionism”. She illustrates how the BLP [Below the Poverty Line] census of “poor families” used by the Indian government constitutes poverty as “lack of income”, silencing the inequitable distribution of power and access to resources. The accompanying proposal to reduce poverty by organizing the “rural poor”, specifically women, into self-help groups, creates “subjects” held to be responsible for their own welfare.
The Programme for International Students Assessment (PISA) provides another example of how governing by numbers takes place. The Programme tabulates and compares national literacy proficiency indicators in the areas of reading, numeracy and science, indicators which translate into assessments of students, their “skills”, and labour-market performance. The Foreword to the summary of PISA 2012 results on “Creative Problem Solving” makes this connection explicit:
“highly skilled adults are twice as likely to be employed and almost three times more likely to earn an above-median salary than poorly skilled adults.” (OECD, 2014: 3; emphasis added)
What “skills” entail and how they are conceptualized are taken for granted, as is the computation of “success” (an above-median salary). This decision to classify students by their standardized achievement and aptitude tests “valorizes some kinds of knowledge skills and renders other kinds invisible” (Bowker and Star 2000: 5-6). As Stephen Ball (2020) describes, “the relationships of truth and power are articulated and operationalized more and more in terms of forms of performance, effects or outputs and outcomes, all expressed in the reductive form of numbers”, resulting in the “numericisation of politics” (Legg 2005: 143).
Numbers and statistics also play a key role in the production of risk categories and statistical projections. As Dean (2010: 206) reminds us:
“There is no such thing as risk in reality …Risk is a way, or rather a set of different ways, of ordering reality, of rendering it into a calculable form. It is a way of representing events in a certain form so they might be made governable in particular ways, with particular techniques and particular goals”.
For example, Lancaster et al. (2020) describe how “Numbering and calculation practices have a key role in the creation, monitoring and regulation of risk populations and risk factors in public health (Castel 1991; Petersen 1997), linked to the rise of modern epidemiology and ‘surveillance medicine’ (Armstrong 1995)”. They highlight the importance of reflecting on the “forms of action” (Rowse 2009: 45; emphasis in Rowse) made possible by these calculative practices (and one might add the “forms of action” rendered irrelevant). The prominent role played by modelling in the COVID-19 pandemic provides a recent example of an intervention where such reflection is called for (Rhodes and Lancaster 2021; Rhodes et al. 2020). We need to ask: what “forms of action” have been rendered acceptable through the models produced to govern COVID-19? What possible “forms of action” go undiscussed?
Edwards et al. (2021) show how risk categories play an increasing role in welfare governing, which they describe as an exercise in the “datafication” of citizens. People are turned into data: “identifying and categorising them to predict future behaviour, allocate resources, and determine eligibility for services and interventions” (Edwards et al. 2021: 2). As the authors explain, “The data are regularly updated, integrated and subject to application of algorithmic tools and predictive risk modelling”. One noteworthy outcome is the way in which particular families are targeted as “problematic”. Peterson and Lupton (1996) describe how this form of targeting is individualizing, “given the tendency to identify ‘risks’ in people’s lives while leaving them responsible to reshape their lives to meet those risks”.
Statistical risk assessments also function prominently in the criminal justice system. For example, Padden and Öjehag-Pettersson (2021) highlight the role of algorithmic profiling
“to detect welfare and tax fraud, assess the likely truthfulness of police complaints, assist in sentencing decisions, and make predictions or trigger ‘risk alerts’ in child welfare, policing, eldercare and psychiatric services (Dencik et al. 2018; AlgorithmWatch 2019)”.
“Predictive policing” algorithms are now being used in law enforcement to determine areas where crime is likely to occur. “Re-offending algorithms” are also being used as a part of sentencing. Elliott Smith (2019: 8-10) reminds us that such algorithms, which necessarily require simplification in order to work, are reductive and reflect prevailing prejudices, “making bigots of us all” (see Street et al. 2021: 159). For example, Edwards et al. (2021: 15) show how the predictive risk modelling used in child protection obscures the in-built equation of socio-economic disadvantage with risk, building in discrimination against the poor.
Importantly, Gitelman and Jackson (2013: 8) point out that “data aren’t only or always numerical”. However, while they do not always exist in number, data are treated as particulate – in the form of separate particles. For example, “qualitative data” and “interview data” involve attention to specific bits of “information”, raising the suggestion that references to “data” in such contexts may well be metaphorical (pursued in next entry) (Gitelman and Jackson 2013: 10).
Because data are particulate, data practices are necessarily aggregative – “They [data] are collected in assortments of individual, homologous data entries and are accumulated into larger or smaller data sets” (Gitelman and Jackson 2013: 8; emphasis in original). It is these practices of organization, classification and deployment that need to be tracked. The goal, according to Bowker and Star (2000: 4), is to identify the origins and consequences of a range of social categories and practices.
You may have noticed that throughout this entry I have used “data” as a plural noun – referring for example to “data are” rather than to “data is” (unless I am quoting someone who uses the singular form as in the title). I apologize to those of you who find grammar uninteresting and hope you may come to see that there is a great deal at stake in whether “data” is treated as “it” or as “they”. I should caution that adopting “data” as a plural is not always easy to implement, given the common singular usage. I had to go back to correct this entry in several places where I had slipped inadvertently into talking about data as “it”.
There appear to be two camps of thought on the topic. The Oxford English Dictionary accepts that the concept of “data” is generally not treated as a plural but as a mass noun (like “information”), an “uncountable noun” that takes a singular verb (i.e., data is) (https://www.lexico.com/definition/data). By contrast, the Publication Guidelines for the American Medical Association state that the singular noun is “datum”, and the plural is “data”, the latter necessitating a plural verb (i.e. data are) (https://amastyleinsider.com/tag/clear-and-concise/).
I prefer and promote the second plural option on the grounds that treating “data” as a singular item, a “mass noun”, produces “it” as a “thing”, with some sort of fixed parameters. In the process the practices involved in “its” production and organization go unobserved and unanalysed. Using “data” in the plural provides a simple way to ensure that how “data” are produced becomes a necessary feature of one’s analysis. Poststructuralists use a similar stratagem in references to knowledgesrather than to knowledge, which is commonly characterized as a “mass noun”.
Clearly, “mass nouns” are not innocent. They assume a specific epistemology and have political effects. They reflect a trend among users of the English language to “focus more on categories and classifications which define a thing and fix its nature for all time” with less concern for “processes, movement and change” (Hoagland 1988: 224). A poststructural approach to concepts and categories aims to return attention to those processes.
Other attempts to highlight the fluid and constructed character of “data” include references to “datafication” (see above regarding “datafication of citizens”) and “datafying”. However, these terms have already been co-opted by data analysis companies. “Datafication” is now commonly adopted to refer to “the collective tools, technologies and processes used to transform an organization to a data-driven enterprise”(https://www.techopedia.com/definition/30203/datafication). In this understanding an organization that implements datafication is said to be “datafied” (https://datafying.co).
Elsewhere I have written about the possible use of gerunds (turning a noun into a verb form by adding “ing”) as a poststructural conceptual strategy to direct attention to the ongoing formation of “things” deemed to be fixed in place and/or time. The concepts of “gender-ing” and “border-ing” are put forward as a means of directing attention to the practices involved in the becoming of “things”, rather than simply accepting them as unchanging states of being (Bacchi 2017; Bacchi and Goodwin 2016: 100). For data, then, we might try referring to “data-ing” – though I suspect that such an awkward term is unlikely to become popular.
Clearly, then, researchers face many dilemmas in relation to “data”, including the rather basic task of how to refer to data without inadvertently producing “it” as a politically neutral singularity. More broadly, returning to the question/comment in the “Kick-off” event, there is an ongoing presumption in mainstream social science studies that research arguments will be “backed up” by data. At the same time and in dramatic contrast, recent debates around the need for “post-qualitative” research put into question the use of all forms of “humanist” research (St Pierre 2013: 226), including research that relies on data. How are researchers to negotiate this minefield? I reflect on this topic in the next entry.
REFERENCES
AlgorithmWatch. 2019. “Automating Society – Taking Stock of Automated Decision-Making in the EU.” https://algorithmwatch.org/wp-content/uploads/2019/02/Automating_Society_ Report_2019.pdf
Armstrong, D. 1995. The rise of surveillance medicine. Sociology of Health & Illness, 17(3): 393-404.
Bacchi, C. 2012. Strategic interventions and ontological politics: Research as political practice”. In A. Bletsas and C. Beasley (Eds) Engaging with Carol Bacchi: Strategic Interventions and Exchanges. Adelaide: University of Adelaide Press. Available as a free download from the publisher’s website: https://www.adelaide.edu.au/press/
Bacchi, C. and Goodwin, S. 2016. Poststructural Policy Analysis: A Guide to Practice. New York: Palgrave Macmillan.
Bacchi, C. 2017. Policies as Gendering Practices: Re-Viewing Categorical Distinctions. Journal of Women, Politics and Policy, 38(1): 20-41.
Bacchi, C. 2020. Problem-Solving as a Governing Knowledge: “Skills”-Testing in PISA and PIAAC. Open Journal of Political Science, 10, 82-105.
Ball, S. J. 2020. Foreword. In E. Grimaldi, An Archaeology of Educational Evaluation: Epistemological Spaces and Political Paradoxes. London and New York: Routledge.
Benozzo, A., Bell, H. and Koro-Ljungberg, M. 2013. Moving Between Nuisance, Secrets, and Splinters as Data. Cultural Studies, Critical Methodologies, 13(4): 309-315.
Bowker, G. C. and Star, S. L. 2000. Sorting Things Out: Classification and its Consequences. MIT Press.
Castel, R. 1991. From dangerousness to risk. In G. Burchell, C. Gordon and P. Miller (Eds) The Foucault Effect: Studies in Governmentality.Hemel Hempstead: Harvester Wheatsheaf
Dean, M. 2010. Governmentality: Power and Rule in Modern Society. Second ed. New York: Sage.
de Goede, M. 2006. International political economy and the promises of poststructuralism. In M. de Goede (Ed.), International political economy and poststructural politics. Houndmills: Palgrave Macmillan.
Dencik, L., A. Hintz, J. Redden, and H. Warne., 2018. “Data Scores as Governance: Investigating Uses of Citizen Scoring in Public Services. Project Report.” Data Justice Lab. Cardiff University, UK. December. https://datajustice.files.wordpress.com/2018/12/data-scores-as-governance- project-report2.pdf
Edwards, R., Gillies, V. and Gorin, S. 2021. Problem-solving for problem-solving: Data analytics to identify families for service intervention. Critical Social Policy, DttOpsI:/1/d0o.i1.o1rg7/71/0.12167170/0128631201813021012092042jo9u4rnals.sagepub.com/home/csp. Rosalind Edwards, SSPC, University of Southampton, SO17 1BJ, England, UK.
Elliot Smith, R. 2019. Rage Inside the Machine: The Prejudice of Algorithms, AND How to Step the Internet Making Bigots of Us All.London: Bloomsbury Business.
Fernandez, B. (2012). Transformative policy for poor women: A new feminist framework. Surrey: Ashgate.
Fordyce, R. and Jethani, S. 2021. Critical data provenance as a methodology for studying how language conceals data ethics. Continuum: Journal of Media & Cultural Studies.https://doi.org/10.1080/10304312.2021.1983259
Foucault, M. 1972. The archaeology of knowledge, and the discourse on language
(trans: Sheridan Smith, A.M.). New York: Pantheon Books.
Foucault, M. 1990. The history of sexuality. Volume I: An introduction
(trans: Hurley, R.). New York: Vintage Books.
Gitelman, L. and Jackson, V. 2013. Introduction. In L. Gitelman and V. Jackson (Eds) “Raw data” is an oxymoron. Cambridge, Massachusetts: MIT Press. pp. 1-14.
Gordon, C. 1991. Governmental Rationality: An Introduction. In Burchell, G., Gordon, C. and Miller, P. (eds) The Foucault Effect: Studies in Governmentality. Chicago: University of Chicago Press. pp. 1-51.
Harrington, C. 2021. United Nations Policy on Sexual Exploitation and Abuse: Problematizations and Performances. Critical Social Policy, 1-21.
Hoagland, S.L. 1988. Lesbian ethics: Toward new value. Palo Alto, CA: Institute
of Lesbian Studies.
Lancaster, K., Rhodes, T. and Rance, J. 2020. “Towards eliminating viral hepatitis”: Examining the productive capacity and constitutive effects of global policy on hepatitis C elimination. International Journal of Drug Policy, 80: 102419.
Legg, S. 2005. Foucault’s population geographies: Classifications, biopolitics and govern- mental spaces, Population Space and Place, 11 (3), 137–156.
Manovich, L. 2020. Data. In P. Heike, Critical Terms in Futures Studies. Palgrave.
Miller, P. 2001. Governing by numbers: Why calculative practices matter. Social Research, 68(2): 379-396.
Mol, A. 2002. The body multiple: Ontology in medical practice. Durham and London: Duke University Press.
OECD 2014. PISA 2012 Results: Creative Problem Solving: Students’ Skills in Tackling Real-Life Problems (Volume V). Paris: PISA, OECD Publishing.
Padden, M. and Öjehag-Pettersson, A. 2021. Protected how? Problem representations of risk in the General Data Protection Regulation (GDPR). Critical Policy Studies, DOI: 10.1080/19460171.2021.1927776
Petersen, A. 1997. Risk, governance and the new public health. In A. Petersen & R. Bunton (Eds) Foucault, health and medicine. London: Routledge. pp. 189-206.
Petersen, A., & Lupton, D. 1996. The new public health: Health and self in the age of risk. London, England: Sage.
Rhodes, T., Lancaster, K., Lees, S. & Parker, M. 2020. Modelling the pandemic: Attuning models to their contexts. BMJ Global Health, 5:e002914, doi:10.1136/ bmjgh-2020-002914.
Rhodes, T. & Lancaster, K. 2021. Excitable models: Enumerated projections and the making of futures without disease. Sociology of Health and Illness, (Advance online publication), DOI: 10.1111/1467-9566.13263.
Rose, N. 1991. Governing by numbers: Figuring out democracy. Accounting Organizations and Society, 16(7): 673-692.
Rowse, T. (2009), The Ontological Politics of “Closing the Gaps”, Journal of Cultural Economy, 2 (1-2): 33-49.
Schüll, N. D. 2013. The folly of technological solutionism: An Interview with Evgeny Morozov. Public Books: A Magazine of Ideas, Arts and Scholarship. Available at: https://www.publicbooks.org/the-folly-of-technological-solutionism-an-interview-with-evgeny-morozov/
St Pierre, E. 2013. The appearance of data. Cultural Studies, Critical Methodologies, 13(4): 2230227.
Street, C. et al. 2021. Do numbers speak for themselves? Exploring the use of quantitative data to measure policy ‘success’ in historical Indigenous higher education in the Northern Territory, Australia. Race Ethnicity and Education, https://doi.org/10.1080/13613324.2021.2019003
Tamboukou, M. 1999. Writing genealogies: An exploration of Foucault’s strategies for doing research. Discourse: Studies in the Cultural Politics of Education, 20 (2), 201–217.
Tietäväinen, A., Pyykkönene, M., Kaisto, J. 2008. Globalization and power – governmentalization of Europe? An interview with William Walters. Foucault Studies, 5, 63–73. Valentine, S. 2019. Impoverished Algorithms: Misguided governments, flawed technologies, and social control. Fordham Urban Law Journal, 46(2): 364-425
Carol Bacchi 