I have been prompted to produce this entry due to the large and increasing numbers of publications that adopt forms of WPR to engage critically with climate change. The questions I address include: does WPR contribute usefully to considerations of climate change policy? If so, what can it add to those conversations? Where can it be applied? How can it be applied? 

I have selected four articles to guide the discussion. I’ll tackle two in this entry: 

Christiansen, K. L. and Lund, J. F. 2024. Seeing the limits of voluntary corporate climate action in food and technology sustainability reports, Energy Research & Social Science 118 (2024) 103798, https://doi.org/10.1016/j.erss.2024.103798

Reimerson, E. et al. 2024. Local articulations of climate action in Swedish forest contexts. Environmental Science and Policy, 151 (2024) 103626

and two next time:

Olsson, D., Öjehag-Pettersson, A. & Granberg, M.2021. Building a Sustainable Society: Construction, Public Procurement Policy and “Best Practice” in the European Union. Sustainability, 13, 7142. https://doi.org/10.3390/su13137142.

Fischer, K. et al. 2024. Progress towards sustainable agriculture hampered by siloed scientific discourses. Nature Sustainability, https://doi.org/10.1038/s41893-024-01474-9 

I found all four articles useful and stimulating. Unfortunately, I had to cut content to make the entry size workable and therefore lose some of the nuance of the arguments. 

To be clear, the Fischer et al. (2024) article does not specifically mention WPR or list the WPR questions as they appear in Tables introducing the approach (Bacchi and Goodwin 2016, p. 20). As a result, I am unable to list it in the comprehensive list of WPR references (at this link:  Welcome to the WPR Network! | Karlstad University (kau.se); see this list for numerous other articles applying WPR to environmental politics and climate change). Still, the authors reference Analysing Policy (Bacchi 2009) in relation to the questions they produce (see Box 1). I intend to show the connection between their questions and WPR thinking in the next entry. 

Significantly, all four articles are produced by researchers in Scandinavia, one in Denmark (Christiansen and Lund 2024) and three in different parts of Sweden (Reimersen et al. 2024, Olsson et al. 2021 and Fischer et al. 2024). I say “significantly” because it is noteworthy to recognise Scandinavian countries leading the field in broadening the parameters of climate change discussions. The connection with Scandinavian countries raises questions about the context in which such discussions become possible – topics for another day. 

I intend to follow the lead of the four selected articles in accepting the relevance of a topic designated as “climate change” (though I probably would have preferred “global warming”). In other words, I accept the relevance of the category “climate change”. Instead of debating its existence (topics for elsewhere) I examine how “climate change” and “politics” are brought together using WPR. What do these articles hope to accomplish? Does WPR assist in the selected tasks? 

A distinction needs to be drawn between the articles by Christiansen and Lund, Reimerson et al. and Olsson et al. on the one side and Fischer et al. on the other. Specifically, the first three articles engage directly with policy interventions directed at climate change, while the last (Fischer et al. 2024) offers a form of systematic review of articles that target climate change or rather “sustainability” (discussed in more detail in next entry). By examining these four applications of WPR we can identify commonalities in the findings, specifically that it is relevant politically to consider deep-seated assumptions about the nature of the world, knowledge and political change in reflections on climate change. In this way the four articles help us to see the kind of thinking, analysis and politics associated with WPR as an analytic strategy. 

To begin, I tackle each article separately. I offer a brief summary of each article. Next, I indicate the variety of materials analysed in the article. The goal here is to indicate just where WPR can be applied. I then take some time to show how the selected article “reads” and applies WPR. Specifically, I want to draw attention to the ways in which the articles work with what I call “WPR thinking” – starting from proposals (of a variety of forms) and “working backwards” to identify how the “problem” is represented. As the discussion progresses, I will highlight relevant distinctions among the articles and the specific interpretations they offer of WPR. Finally (next entry) I hope to draw together the insights produced in the selected articles. This intervention should allow us to say something useful about WPR as a critical analytic strategy. To make my contribution usable, I keep the comments brief and hope that in the process I do not misrepresent the arguments.

Article 1: Christiansen, K. L. and Lund, J. F. 2024. Seeing the limits of voluntary corporate climate action in food and technology sustainability reports, Energy Research & Social Science 118 (2024) 103798, https://doi.org/10.1016/j.erss.2024.103798

Brief summary

The analytic target in this article is corporate climate action. The authors want to know how companies represent the “problem” of climate action. To this end they examine corporate sustainability reports. They find that companies, across the two sectors of agriculture and technology, emphasise solutions aimed at improving efficiency and substituting carbon-intensive inputs in production processes, yet remain largely silent on solutions which could transform or reduce current consumption and production patterns. The companies also frame their products and services as necessary for society, legitimizing their continuing production of emissions. 

Materials analysed 

Christiansen and Lund make an important contribution by highlighting the possibility of applying WPR to forms of material other than Government reports and legislation. Note: they target companies’ sustainability reports. They offer therefore an analysis that illustrates the relevance of a governmentality approach, broadening our conception of “governance” (Bacchi and Goodwin 2016, pp. 41-44). Corporate entities are clearly involved in governing practices. 

Applying WPR thinking

In Christiansen and Lund (2024) there is a strong and useful focus on “proposals”, designated as the starting point for a WPR analysis. Let us recall the key premise in WPR: what is proposed as an intervention (proposed solution) reveals a target for change and hence the way the “problem” is represented and produced (Bacchi 2022). The authors clearly articulate this point: 

“we examine corporate sustainability reports to illuminate how companies’ proposed solutionsboth constitute what they consider their own climate impacts and contribute to identifying and delimiting possible and reasonable corporate responses. (Christiansen and Lund 2024, p. 2; emphasis added)”

Here I offer a few examples of selected proposals and how problem representations are “read off” from them.

•  P. 3 – “All the food companies highlight investments in electrification, energy efficiencies and production of renewable energy, for example, from biogas and solar panels, as key avenues towards reducing their climate impact. JBS [a large meat-producing enterprise] writes: ‘Upgrades include switching out equipment with more energy-efficient alternatives, such as replacing gas boilers with heat pumps and optimizing refrigeration systems. The business also already has local solar projects at many of its sites’ (p. 58)”. Based on these proposals, the authors (p. 3) conclude: “the companies frame the problem of their climate impact, as one of inefficiency in energy use and use of fossil fuels”. Having identified the problem representation [as “inefficiency”], the authors proceed to examine underlying assumptions and the limitations of the problem representation.
•  In relation to the technology companies in their sample, the authors note: “The companies often emphasise the need to increase the awareness of and information about emissions, both for themselves, their suppliers and their clients”. It follows that “Emphasis on these solutions partly frames climate change as an information deficit problem, where we need to know more about our emissions and need new tech to address them” (p. 6). The authors proceed to query the grounds of this problem representation [information deficit], pointing out for example how such an approach can produce a “fixation on metrics and data” (p. 6).

For both the agriculture and technology sectors the authors stress the “unquestioned pursuit of efficiency” and how such a frame of reference “legitimises and depoliticises energy-intensive production and consumption patterns” (p. 7). The depoliticization of climate change identifies an important theme that arises in the other articles selected for this discussion.

The authors distinguish what they are doing in their WPR analysis from studies of “greenwashing”. They describe “greenwashing” as “strategic framings that obscure discrepancies between what actors say they do to address climate change and what they actually do” (p. 2). Rather, they want to consider how the communications of the targeted companies contribute to “producing perceptions of possibility and responsibility for mitigating climate change” (p. 2). This distinction between strategic framings and the unintended effects of corporate communications illustrates a useful understanding of WPR. The target is not deception on the part of companies but the operation of deep-seated assumptions in producing “problems” as particular sorts of problem. 

Insights generated 

Connected to the previous point, the authors produce a useful discussion of frame theory and include WPR as part of “the framing and discourse theories we build upon” (p. 2). They describe both frame theory (drawing on Entman 1993) and WPR as following “Foucauldian ideas of how discursive practices contribute towards defining and delimiting what is seen as legitimate and sensible to say”.

This postulated association between WPR and frame theory is debatable (see Research Hub entry 29 January 2025). I describe frame theory as more interested in interpretations of “problems” while WPR targets how specific interventions constitute or produce “problems” as particular sorts of problem. The authors recognise that in WPR “policies are constitutive of problem framings that, in turn, come to shape broader understandings of what constitutes societal problems (and not)” (p. 2; emphasis added). The contrast between interpretive and constitutive approach to “problems” is important and deserves more attention. It comes up again in the discussion of Reimerson et al. 2024 (below) and in Olsson et al. 2021 (next entry). 

Here I note that the authors state their decision to emphasise questions 1, 2 and 4 of the WPR approach because “Responding to questions 3, 5 and 6 requires engagement with other empirics” (p. 2). I suggest that this delineation indicates the interpretive focus in this article while the other “empirics” draw attention to the constitutive effects of problem representations. Since I stress that all the WPR questions are interconnected, I would advise against drawing this distinction. 

Article 2: Reimerson, E. et al. 2024. Local articulations of climate action in Swedish forest contexts. Environmental Science and Policy, 151 (2024) 103626

Brief summary

This article brings an innovative use of WPR to the study of “local articulations” of climate action. I say “innovative” because most common applications of WPR target forms of Government policy or the directives of other organisations (see the targeting of corporations and their reports in Christiansen et al 2024 above). In this article, the target group is “local actors”. The authors connected with this “broad range of stakeholders in two case study locations in Sweden to explore potential course of action for local climate action in relation to forests” (Abstract). The expressed hope is that “forests might offer a way to connect climate change as an abstract, global dilemma to concrete actions on the local level”. A grounding premise is that “local actors are recognized as key drivers for climate action”. 

Materials analysed

To study “local articulations” the authors conducted a series of four consecutive workshops with two groups of forest stakeholders in two different locations – one in northern Sweden and one in southern Sweden (p. 3). Thirty-one forest stakeholders participated in the process. The “pairs of workshops” aimed in the end at presenting “prioritized targets” for local decision-makers and public officials. This article focused on Workshop 2 and “participants’ development of local and societal pathways towards their envisioned futures”. The material from the workshops “was transcribed, translated from Swedish to English, and coded by the first author”. For the coding process, the authors “used a software for qualitative text analysis (NVivo 12) to enable detail and complexity in the coding and analysis, inductively developing themes and sub-themes following an analytical framework based on poststructural policy analysis (see Section 3.4 below)” (p. 4; emphasis added).

I have italicized the point about applying the software “following an analytical framework”. This analytic approach fits my suggestion that to use coding in relation to WPR it is necessary to apply the theory (WPR) first (see Research Hub entries 28 Dec 2023 and 29 January 2025). Also, the treatment of the transcripts as “text” may provide a way to use WPR in other scenarios based on group discussions. The suggested way forward here shares family resemblances with Postructural Interview Analysis (PIA) (Bacchi and Bonham 2016). Forecasting their clear understanding of a WPR analytic strategy, the authors note that the analyzed material “includes all proposals put forward by the participants” (p. 4; emphasis added). 

Applying WPR thinking

The authors state at the outset that the “aim of the paper is to critically analyse these local articulations of climate action”. To this end they ask: 

• What is the problem represented to be in local forest stakeholders’ articulations of climate action? 
• What assumptions underlie these problem representations? 
• What are their potential effects and consequences for local and forestry-related climate action? (p. 2)

Note here the uptake of Questions 1, 2 and 5 from the WPR template, compared to 1, 2 and 4 in Christiansen et al. 2024 above). I comment on the significance of this distinction below.

I was struck by how consistently and how effectively the authors identified “proposals” as starting places for the analysis. In other words, they clearly grasp and use WPR thinking in their analysis. They begin the analysis with recognition that 

“The participants presented both concrete, detailed proposals for forest and natural resource management (e.g., “triad forestry – intensive cultivation 33%, ‘normal’ management 33%, protect 33%”2) and abstract, visionary targets for the environment, the climate, or natural resources in general (e.g., ‘use our resources in the right way’; ‘climate- smart targets for the use of forests). (p. 6; emphasis added)”

They proceed to identify specific proposals and how they represented the “problem”. For example, 

“Among the proposals addressing landscape use and forestry on the local (municipal) and regional (county) levels, several targets indicate a desired move towards a different perspective on landscape use (including the creation of ‘new functional areas’, and a target stating that ‘[a] landscape perspective [is] needed in natural value assessments’). (p. 6; emphasis added)”

As one more example: “A number of proposals targeted resource use by addressing construction, consumption and waste, and energy”. (p. 6; emphasis added)

The authors thus use identified proposals to launch the analysis. Proposals provide access to the problem representations that require analytic scrutiny for their deep-seated assumptions and for their effects. 

Insights generated

Paralleling the argument in Christiansen et al. 2024, this article describes local climate action as “contingent on overarching discourses that affect understandings of possible actions, possible actors, and their opportunities and limitations on all levels of politics and administration” (p. 2). Their close analysis indicates that “Apart from some proposed targets indicating a more radical shift towards local autonomy and self-sufficiency, the participants’ proposals tended to stay close to current policies and practices”. (p. 7) They elaborate:

“For example, among the proposals concerning transportation, most seemed to assume continued high mobility of people and goods – just with different modes of transportation (public transport; railways) and/or different (fossil free) fuel sources. The detailed policy content of more radical transformation (e.g., significant decreases in mobility) remains rather quiet, if not completely silent, in the material. There is also a tendency of silence around concrete alternatives for land- and resource use. (p. 7)” 

And finally:

“Through proposals of, for example, more wood construction and increased use of bioenergy – and through the silences on concrete alternatives for land- and resource use – the participants’ pathways assume a continued use of forest resources. They seem to remain largely within the confines of – and further reiterate – dominating market-oriented and administrative- managerial rationalities.” 

Importantly, this article probes the “discursive effects” produced by these problem representations (Question 5 in WPR; Bacchi and Goodwin 2016, p. 20). That is, it moves beyond an analysis of competing interpretations (see Christiansen and Lund 2024, above) to consider the political fallout produced through the problem representations they identify. They emphasise “the discursive and positioning effects of assumptions of politically neutral (or apolitical) knowledge as necessary for action” and conclude on the need “for continued critical scrutiny of the underlying premises and choices of climate action on all levels (cf. Hulme, 2015) (p. 10; emphasis added). Through this perspective they demonstrate the usefulness of WPR as an analytic strategy.

I pursue articles 3 (Olsson et al. 2021) and 4 (Fischer et al. 2024) in the next entry. 

References

Bacchi, C. 2009. Analysing Policy: What’s the Problem Represented to be? Pearson Education.

Bacchi, C. 2022. Keynote Address, Karlstad International WPR Symposium. https://www.kau.se/en/political-science/forskningsprojekt/welcome-wpr-network

Bacchi, C. and Bonham, J. 2016. Appendix: Poststructural Interview Analysis: Politicizing “personhood”, in C. Bacchi and S. Goodwin, Poststructural Policy Analysis: A Guide to Practice, NY: Palgrave Macmillan, pp. 113-122.

Bacchi, C. and Goodwin, S. 2016. Poststructural Policy Analysis: A Guide to Practice. NY: Palgrave Macmillan. 

Entman, R. M. Framing: toward clarification of a fractured paradigm. Journal of Communication 43(4): 51-58.

Hulme, M., 2015. (Still) disagreeing about climate change: Which way forward? Zygon® 50, 893–905

It is difficult, if not impossible, to keep up with the pace of change to do with AI. Some months ago, I shared with you my first experiments with ChatGPT (29 Sept 2023). A question arises as to the usefulness of WPR in relation to these developments. In this entry I introduce a recent article that tackles European policy regulating AI and that draws upon WPR to produce a cogent analysis (Pham and Davies 2024). 

I am frequently asked for exemplars of how to work with WPR. The Research Hub provides an ideal venue to pursue this project. Hence, you can expect to find several subsequent entries that will bring to your attention useful WPR applications. The entries will include comments on the kinds of materials that can be used for a WPR analysis and on theoretical issues that require further consideration.

Article: Bao-Chau Pham & Sarah R. Davies (02 Jul 2024): What problems is the AI act solving? Technological solutionism, fundamental rights, and trustworthiness in European AI policy, Critical Policy Studies, DOI: 10.1080/19460171.2024.2373786 

Brief summary: 

The authors critically explore the policy “responses” to AI produced by the European Commission, with a particular focus on the AI Act (the EU’s Proposal for a Regulation: Laying Down Harmonised Rules on AI), first published in April 2021. They find that this policy constitutes the technology as both opportunity and threat, and that these problematisations are made to “cohere” through risk-based categorizations. A particular point of interest is how these problematisations enact an “exceptionalist notion of Europe as a policy actor and coherent political community” (Abstract).

Materials used:

Pham and Davies (2024) offer a clear guide to the sources they draw upon for their analysis. Table I (p. 6) lists the AI policy documents published by the European Commission (EC) and the High-Level Expert Group on AI (HLEG AI) between April 2018 and April 2021. The authors explain that, while the policy “response” is most explicitly spelt out in the AI Act, they consider this wider “corpus”, 

“as the documents are all constitutive of the policy discourse that underpins the AI Act and wider EU AI policy discussion. These additional documents allow us to understand how the proposal and justification for the AI Act became intelligible. (p. 7)”

This use of related texts is often necessary to “build up a fuller picture of a problem representation” (Bacchi 2009, p. 20). 

Applying WPR

The authors display and state clearly that they see their work as extending “prior analysis of what the AI Act is doing (and especially Krarup and Horst 2023; Paul 2023)” (p. 15 Note 1). They consider how the WPR approach extends “the discussion on the performative politics of European AI policy” (p. 15). Their analysis offers readers insights into the usefulness of approaching policy through problematisations. 

The authors produce a clear guide to how to apply WPR. I proceed to summarize some key points:

The WPR approach recommends starting from “proposals” or “proposed solutions” in policy texts and “working backwards” to see how these produce the “problem” as a particular sort of problem. 

Pham and Davies (p. 2) follow this analytic strategy. They note: “Our starting point is the policy solution presented in EU documents, namely the AI Act and the regulatory strategy it proposes, that of risk-based tiers”. The authors provide a useful overview of the literature that examines the place of “risk technologies” in governing practices (p. 3). They also link their analysis to other scholarship that understands “policy as constitutive and as producing the entities that it refers to”. This point, that WPR targets what policies produce as real, is critically important: 

“Policy discussions of AI and related digital technologies are thus not neutral but enact particular visions and imaginations of these technologies and the societies in which they are situated (af Malmborg 2022; Bareis and Katzenbach 2022).(p.4)” 

By “working backwards” from the initial risk-based proposal, Pham and Davies (2024) are able to identify two problematisations: first, that AI is necessary to productivity; and second; that (some) AI is “risky” and poses threats to the “rights” of European citizens. The notion of “risk” allows these two apparently conflicting problematisations to “cohere”. 

Mitchell Dean (1999, p. 177) elaborates the governmentality position that, in effect, “There is no such thing as risk in reality”:

“Risk is a way – or rather, a set of different ways – of ordering reality, or 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 for particular goals.”

The task for researchers, therefore, is to draw attention to the practices involved in the production of “risk” thinking and “risk technologies”. The notion of “risk technology” highlights the role of “risk” categories as governing mechanisms. Pham and Davies’ (2024) analysis provides insights into the place of “risk” thinking in governing AI.

To elaborate on the two identified problematisations, the authors “mine” the “related texts” listed in Table 1. They describe their analytical approach as abductive “in that we were sensitized to the seven questions outlined by Bacchi”. They elaborate: 

“We considered Bacchi and Goodwin’s (2016) suggestion that it might be necessary to go through the WPR steps multiple times. In this vein, we read the documents repeatedly and conducted multiple rounds of iterative coding, revisiting the WPR framework and its questions throughout the analytical process.” (p. 7; emphasis added)

Through this practice Pham and Davies (2024) illustrate how it may be possible to develop methods to use WPR in relation to large bodies of material if you apply the theory (WPR) throughout the analytic process (an argument I make in the previous Research Hub entry, 27 Feb 2025).

The authors effectively use quotes from their primary material (see reference to Table 1 above) to support their arguments. For example, they quote this 2020 EU White Paper to firm up the problematisation of AI as necessary for productivity: AI will facilitate “gains that can strengthen the competitiveness of European industry and improve the well- being of citizens (European Commission 2020b, 25)”. I would point out that quotes such as this one could be described as “proposals” in the WPR sense of the term as they promote a particular vision that describes what needs to be done. Hence, they serve as entry-points for identifying and interrogating problematisations/problem representations. 

In elaborating the second problematisation, the need to protect European values from “risky” AI, the same White Paper states: 

“It is more important than ever to promote, strengthen and defend the EU’s values and rules, and in particular the rights that citizens derive from EU law. These efforts undoubtedly also extend to the high-risk AI applications marketed and used in the EU under consideration here. (European Commission 2020b, 18) “

This quote, again, is a clear proposal (in the WPR sense of the term) about what needs to be done. I make this point to illustrate the analytical fruitfulness of approaching primary textual material through the lens of “proposals” and their problematisations (this topic forms the basis of a forthcoming Research Hub entry). In other words, the supplementary policies listed in Table 1 are replete with proposals that could provide the focus for analysis. 

• Pham and Davies (2024) are particularly interested in the effects of the identified problematisations (Question 5 in WPR; Bacchi and Goodwin 2016, p. 16). Among the three sub-categories of interconnected effects introduced in Analysing Policy(Bacchi 2009, p. 15-18) (discursive, subjectification and lived), they zoom in on subjectification effects. Through this topic, they illustrate how it is possible to foreground some parts of a WPR analysis if space constraints impose limits on what can be covered. 

The authors describe subjectification effects as “how certain subject positions (in this case that of “Europe” itself) are constituted through problematizations and their proposed solutions” (p. 11 ff). They stress that “European AI” is imagined to “safeguard ‘core societal values’” and in doing so to “carve out a distinctive trademark”. The notion of a “(more) trustworthy, ethical ‘AI made in Europe’) (European Commission 2018b, 1)” is “repeatedly stated as a desirable goal”. 

• In terms of critique (thinking of Question 4 in WPR and the need to identify “silences”, p. 7) the authors endorse analyses that query the notion of “trustworthy AI”, describing it as ambiguous (Stix 2022) or as a buzzword (Reinhardt 2023), and that draw attention to the limitations of “techno-solutionism” (Katzenbach 2021; Paul 2022). They note two specific issues that emerge from their analysis: “the way in which Europe emerges as an exceptionalist policy actor and the related question of who is included in the AI Act’s efforts to protect ‘citizens’” (p 13). They raise questions concerning “whose values and rights are not being included in EU policy”, whether the “dignity and rights of people on the move” are included, and if the global impact of AI technology features in the policy statements. Finally, they emphasise that the constitution of AI technology as economically productive equates citizenship with participation in markets in ways that reflect “at the very least, an incomplete imagination” (p. 14):

“the policy documents produce a version of European citizenship that is perhaps better aligned with Homo economicus than with other versions of the citizen (Brown 2015).” 

• Pham and Davies (2024) conclude their analysis with a section on “self”-problematisation, examining “the situatedness and limits of our own analysis” (p. 13). It was refreshing to see this topic included as it is so often ignored. Step 7 (Bacchi and Goodwin 2016, p. 20) signals the need to recognise that researchers are inevitably located in cultural and political logics that could well affect their analyses The authors clearly understand the significance of being willing to subject one’s own proposals to the WPR questions. They acknowledge the limits of the corpus they draw upon. The focus on policy documents and on Europe means “ignoring the many other actors and processes that are influential: industry activity, media discussion, national policy in both Europe and across the world, NGOs, and civil society” (p. 14). It follows that their argument about how AI and European identity are realized is necessarily incomplete.  

Theoretical Issues:

The WPR approach asks you to leave behind conventional notions of “problem” and “solution”. There are no “problems” per se. There are only problem representations. In addition, references to “solutions” take on a new meaning. A “solution” is not the resolution of a difficulty; rather, it provides the starting place for identifying problem representations, remembering that proposals (postulated solutions) indicate what is targeted as in need of change and hence what is represented as problematic, i.e., as the “problem”. “Solutions”, therefore, provide the starting points for analysis, not the end points. To keep this argument to the fore, I suggest the need for quotation marks when you use these terms except in cases where it is clear that “problems” and “solutions” are treated as some sort of presumed pre-existing state or entity. This topic is pursued further in forthcoming Research Hub entries. 

Conclusion: 

The usefulness of WPR can best be gauged by examining what researchers achieve through its adoption. I would highly recommend reading Pham and Davies (2023) to see how they craft their argument. For the next several entries, I will bring to your attention insightful and challenging contributions on climate change, menstruation, unemployment and Universal Basic Income. Please let me know which particular topics you would like to see discussed (carol.bacchi@adelaide.edu.au).

For other applications of WPR in relation to AI and related topics, see the following suggestions. Please let me know if I have missed any: 

Kallioinen, E. 2022. The Making of Trustworthy and Competitive Artificial Intelligence: A Critical Analysis of the Problem Representations of AI in the European Commission’s AI Policy. MA thesis, Faculty of Social Sciences, University of Helsinki.

Lindt, M 2022. The Geopolitics of Artificial Intelligence: A Comparative Policy Analysis of French and Chinese Artificial Intelligence Policy Discourse. MA thesis, Malmö University, Faculty of Culture and Society (KS), Department of Global Political Studies (GPS).

Padden, M. 2023. The transformation of surveillance in the digitalisation discourse of the OECD: A brief genealogy. Internet Policy Review, 12(3). https://doi.org/10.14763/2023.3.1720

Padden, M. and Öjehag-Pettersson, A. 2021. Protected how? Problem representations of risk in the General Data Protection Regulation (GDPR). Critical Policy Studies, https://doi.org/10.1080/19460171.2021.1927776

Padden, M. and Öjehag-Pettersson, A. 2024. Digitalisation, democracy and the GDPR: The efforts of DPAs to defend democratic principles despite the limitations of the GDPR. Big Data & Society, 1-13, DOI: 10.1177/20539517241291815 

Puukko, O. 2024. Rethinking digital rights through systemic problems of communication. Revista Latina de Comunicación Social, 82, 01-19. https://www.doi.org/10.4185/RLCS-2024-2044  

Rahm, L. and Rahm-Skågeby, J. 2023. Imaginaries and problematisations: A heuristic lens in the age of artificial intelligence in education.  British Journal of Educational Technology, pp. 1-13. DOI: 10.1111/bjet.13319 

Sundberg, L. 2019. If Digitalization is the Solution, What is the Problem? In T. Kaya (Ed.) ECDG 2019 19th European Conference on Digital Government. Academic Conferences and Publishing Ltd. pp. 136-143.

Wong-Toropainen, S. 2024. Problematising User Control in the Context of Digital Identity Wallets and European Digital Identity Framework. In: Prifti, K., Demir, E., Krämer, J., Heine, K., Stamhuis, E. (eds) Digital Governance. Information Technology and Law Series, vol 39. T.M.C. Asser Press, The Hague, pp. 115-136. https://doi.org/10.1007/978-94-6265-639-0_6 

References

af Malmborg, F. 2022. “Narrative Dynamics in European Commission AI Policy—Sensemaking, Agency Construction, and Anchoring.” Review of Policy Research 40 (5): 757–780. advanced online publication. https://doi.org/10.1111/ropr.12529.

Bacchi, C. 2009. Analysing Policy: What’s the Problem Represented to be. Frenchs Forest: Pearson Education. 

Bacchi, C. and Goodwin, S. 2016. Poststructural Policy Analysis: A Guide to Practice.New York: Palgrave Macmillan.

Bareis, J., and C. Katzenbach. 2022. “Talking AI into Being: The Narratives and Imaginaries of National AI Strategies and Their Performative Politics.” Science, Technology, & Human Values 47 (5): 855–881. https://doi.org/10.1177/01622439211030007.

Brown, W. 2015. Undoing the Demos: Neoliberalism’s Stealth Revolution. Cambridge, MA: MIT Press. 

Dean, M 1999, Governmentality: Power and rule in modern society, Sage, London.

European Commission. 2018b. “Coordinated Plan on Artificial Intelligence.” COM (2018) 795 

Final.

European Commission. 2020b. “White Paper: On Artificial Intelligence – a European Approach to 

Excellence and Trust.” COM (2020) 65 Final.

Katzenbach, C. 2021. “’AI Will Fix This’ – the Technical, Discursive, and Political Turn to AI in Governing Communication.” Big Data & Society 8 (2): 1–8. https://doi.org/10.1177/ 20539517211046182. 

Krarup, T., and M. Horst. 2023. “European Artificial Intelligence Policy as Digital Single Market Making.” Big Data & Society 10 (1): 1–14. https://doi.org/10.1177/20539517231153811.

Paul, R. 2022. “Can Critical Policy Studies Outsmart AI? Research Agenda on Artificial Intelligence Technologies and Public Policy.” Critical Policy Studies 16 (4): 497–509. advance online publication. https://doi.org/10.1080/19460171.2022.2123018.

Paul, R. 2023. “European Artificial Intelligence “Trusted Throughout the World”: Risk-Based Regulation and the Fashioning of a Competitive Common AI Market.” Regulation & Governance. https://doi.org/10.1111/rego.12563.

Pham, B-C & Davies, S. R. (02 Jul 2024): What problems is the AI act solving? Technological solutionism, fundamental rights, and trustworthiness in European AI policy, Critical Policy Studies, DOI: 10.1080/19460171.2024.2373786  

Reinhardt, K. 2023. “Trust and Trustworthiness in AI Ethics.” AI and Ethics 3 (3): 735–744. https:// doi.org/10.1007/s43681-022-00200-5. Stix, C. 2022. “Artificial Intelligence by Any Other Name: A Brief History of the Conceptualization of ‘Trustworthy Artificial intelligence.” Discover Artificial Intelligence 2 (1), advance online publication. https://doi.org/10.1007/s44163-022-00041-5. 

This entry follows through on the preceding one in which I endeavour to clarify distinctions between the theoretical approaches offered in WPR, RTA (Reflexive Thematic Analysis) and CFA (Critical Frame Analysis). To begin I offer a summary of the key points raised in the preceding entry and, at the end, I consider a way to combine approaches that may prove useful. I refer you also to the Research Hub entry on 23 December, 2023, entitled: “Oops, I said ‘themes’: WPR and RTA (Reflexive Thematic Analysis)”. 

The key distinctions among approaches can be tackled through three interconnected discussion points:

First, the kinds of assumptions interrogated;

Second, approaches to knowledge;

Third, approaches to materials used in analysis.

First, the kinds of assumptions interrogated:

Alvesson and Sandberg (2013) usefully identify contrasting “levels of analysis” in academic theorising based on the kinds of assumptions they examine. They mention five kinds of assumption but for our purposes I zoom in on the contrast between “in-house assumptions” and “field assumptions”. 

Put simply, “in-house assumptions” are those accepted by the discipline (e.g. psychology, economics, etc.) within which you find yourself. Alvesson and Sandberg offer the example of “trait theory” in psychology. Someone who wishes to question the traits commonly assigned a leader, for example, still accepts that it is a worthwhile project to examine something called “leadership traits”. Trait theory is taken for granted; it is not questioned. 

By contrast, if a researcher examines “field assumptions” the whole notion of a “trait” would come up for questioning. What does it mean to assign “traits” to people? What assumptions about human psychology underpin the notion of “trait”? These are the sorts of questions that would be raised in Foucauldian-influenced analysis. Hence, such analysis questions “field assumptions”. 

WPR operates at the level of “field assumptions”. Question 2 specifies the need to identify the epistemological and ontological presuppositions underpinning identified problem representations – e.g., the meanings required for something called a “trait” to exist. By contrast, RTA appears to work at the level of “in-house assumptions”.

Consider, for example, how “evidence” is treated. As an in-house assumption, “evidence” informs the use of coding, commonly adopted in RTA studies. The premise is that bits of knowledge can be identified and labelled, and henceforth manipulated for research purposes. 

WPR, on the other hand, challenges the whole idea of “evidence” – therefore it works at a different level of analysis – at the level of questioning deep-seated assumptions or field assumptions such as “evidence”. I hope you can see how disagreements about the level of analysis that is required prove problematic in blending WPR and RTA.

Second, approaches to “knowledge”: 

Closely related to the first discussion point, WPR brings a sceptical approach to “knowledge”. This scepticism becomes clear if we consider the analytic strategy it endorses. As a commencing premise in that strategy, WPR makes the case that: what one proposes to do about something indicates what is identified as needing to change and hence what is deemed to be problematic (“the problem”). The analysis therefore starts from what is proposed (e.g., recommendations, aims, implied goals, etc.) with the argument that it is possible to “work backwards” from proposals to identify what the “problem” is represented to be (problem representations). I use the example of training programs for women to illustrate this argument:  if training programs for women (to increase their representation in positions of influence) is the proposal, the “problem” is represented to be women’s lack of training.

While this point may not appear at first glance to advance the discussion in a useful manner, WPR proceeds to interrogate this proposition. To open identified problem representations to critical scrutiny, we examine the “knowledges” they rely upon (Question 2 in WPR). Note the plural use of “knowledges” instead of common references to “knowledge”. The whole point of the analysis is to question taken-for-granted knowledges commonly accepted as true. 

For example, training programs for women accept and work within assumptions that “training” increases “skills”. These “in-house” assumptions, associated with developmental psychology, provide the grounds for the proposal. Challenging this reliance on psychology as a form of “truth”, in WPR, the assumed “knowledge” of “psychology” becomes something to question as one form of (competing) “knowledges”. 

By contrast RTA offers a psychological theory. Its very premise is the usefulness of psychology as an approach to knowledge. We can see this premise operating, for example, in the work of Byrne (2022) who examines “opinions” and “attitudes”, a fairly conventional psychological approach. For WPR “opinions” and “attitudes” are “field assumptions” that require questioning. 

Closely related to these different approaches, how the “subject” is conceptualised in RTA, as a being with “attitudes” and “opinions”, sits in sharp contrast to the position on “subjectification” in WPR. On the RTA side, we see a version of a sovereign subject with “attitudes”; on the WPR side we work with a provisional subject produced in practices (Bacchi and Goodwin 2016, pp. 30, 40). 

Third, the kinds of material used: 

RTA relies on interviews and focus groups as offering access to “opinions”. As in the point just covered, this research approach presumes subjects as kinds of being who express “views” that reflect an interior consciousness. By contrast, as just noted, WPR adopts a Foucauldian perspective that questions the sovereign subject presumed in RTA. Hence, for WPR researchers, there is a need for a new approach to interviews as research materials. 

In the Appendix to Poststructural Policy Analysis, Jennifer Bonham and I (2016) develop an approach to interviews as research materials that builds on the theoretical premises informing WPR. Called Poststructural Interview Analysis (PIA) it highlights the meanings that need to be in place for certain statements to be possible. As in WPR, the focus is on underlying conceptual logics and taken for granted assumptions in interview texts rather than on a presumption that interviewees can access the “truth” of their experience. 

A possible way forward

Researchers who turn to RTA in their WPR analysis tend to want to find a way to engage large amounts of material. The common focus in WPR studies on a single piece of legislation or report as a starting point for the analysis seems to miss the opportunity to provide a wider picture of relevant texts. I should note that in Analysing Policy (Bacchi 2009: 20) I point out that “It is often, even usually, necessary to examine related texts … to build up a fuller picture of a particular problem representation”. I still find this approach useful (Bacchi 2023) but remain open to other suggestions for bringing WPR to large bodies of material. In this spirit, I introduce the innovative study produced by Stoor et al. (2021) on suicide among the Sami. 

The authors started their analysis from “proposals” in a large number of texts (40) and “work backwards” to identify problem representations – providing a highly effective example of WPR as analytic strategy. Their 40 texts produced 40 different, but related, problem representations. They proceeded to organise the material in a useful way. They introduce five categories to refer to more specific kinds of problem representation. The proposals to “address” suicide cluster around these five themes, “pertaining to shortcomings on individual (5), relational (15), community/cultural (3), societal (14) and health systems levels (3)”. 

The suggestion in this example is that it may be possible to apply the theory first (ask the WPR questions) and subsequently to organise the identified problem representations into “themes” (using a version of RTA). Here we need to remember that themes capture “some level of patterned response or meaning” (Braun and Clarke 2021: 341). Clearly, Stoor et al.’s five categories offer five forms of explanation for suicide among the Sami that emerged in their material: they found representations that problematised individual behaviour, relational situations, community/cultural factors, societal factors and the problematisation of health systems. 

This useful organisation of the material by these five themes is not the only possible organisation. By producing supplementary material, Stoor et al. keep open the possibility that problem representations could be clustered in some other way. Hence, they refuse to impose one true meaning on the material, instead acknowledging the complexity of the situation. This willingness to embrace ambiguity and contingency results in a thought-provoking analysis.

Conclusion

I realize there may be pressures applied to research students and other researchers to produce analyses that can be described as “rigorous” or “systematic”. Large bodies of material, coding and “themes” appear to fit the bill. The goal is commonly described as “producing new knowledge”. The poststructural stance associated with WPR questions “knowledge” as assumed “truth”, referring instead to “knowledges” as plural conceptions of “truth”. I believe it is necessary to reflect on these contrasting perspectives in any attempt to bring together WPR and some other theoretical framework. 

References

Alvesson, M., & Sandberg, J. 2013. Constructing research questions: Doing interesting research. London: Sage.

Bacchi, C. 2009. Analysing Policy: What’s the Problem Represented to be?Frenchs Forest: Pearson Education.

Bacchi, C. 2023. Bringing a ‘What’s the problem represented to be?’ approach to music education: a national plan for music education 2022, Music Education Research, DOI: 10.1080/14613808.2023.2223220 .

Bacchi, C. and Bonham, J. 2016. Appendix: Poststructural Interview Analysis: Politicizing “personhood”, in C. Bacchi and S. Goodwin, Poststructural Policy Analysis: A Guide to Practice. NY: Palgrave Macmillan. pp. 113-122.

Bacchi, C. and Goodwin, S. 2016. Poststructural Policy Analysis: A Guide to Practice. NY: Palgrave Macmillan. 

Braun, V. & Clarke, V. 2021. One size fits all? What counts as quality practice in (reflexive) thematic analysis?, Qualitative Research in Psychology, 18:3, 328-352, DOI: 10.1080/14780887.2020.1769238 

Byrne, D. 2022. A worked example of Braun and Clarke’s approach to reflexive thematic analysis. Quality & Quantity, 56: 1391-1412. 

Stoor, J. P. A, Eriksen, H. A. and Silviden, A. C. 2021. Making suicide prevention initiatives targeting Sámi in Nordic countries. BMC Public Health, 21:2035. https://doi.org/10.1186/s12889-021-12111-x  

This entry is prompted by the increasing attention directed to the possibility of blending, or integrating, WPR with other analytic approaches. The two analytic approaches I consider include Reflexive Thematic Analysis (RTA) and Critical Frame Analysis (CFA). These approaches share a commitment to use large bodies of material and to find ways to organise that material. They also share some common terminologies, including “themes”, “coding” and “evidence”. There is, in these approaches, a tendency to focus on policy actors, though Walsh (2022: 5) points to an interest in “hegemonic discourses” in CFA. 

I have considered these topics elsewhere. Specifically in a Research Hub entry in 2023 called “Oops, I said ‘themes’” (23 December) I focus on the presuppositions underpinning Reflexive Thematic Analysis. In April 2018 I made some preliminary comments on the relationship between WPR and frame theory. In other writing (Bacchi 2009b) I begin to probe some possible links between WPR and CFA. Here I offer reflections on a conference presentation by Walsh (2022) that suggests the usefulness of integrating WPR and CFA (see also Walsh 2024). I refer in passing to other recent articles that propose blending WPR and RTA (Olvik 2024, Dinmore et al. 2024). 

Starting places: problematisations

In the earlier Research Hub entry on Reflexive Thematic Analysis and WPR (23 Dec. 2023), I stress the importance of considering the form of research question brought to one’s research material. Research questions provide starting places for thinking. Both RTA and CFA tend to ask why something happens. Walsh (2022: 28-29), for example, asks, in her selected policy areas, “why the assumption of a clash between culture and women’s rights dominates public debate and the academy”. WPR does not ask “why” questions. It asks “how possible” questions – how was it possible for women’s rights to be conceptualised in a way that sets them in opposition to assumed cultural standards? 

The boundary between “why” and “how possible” questions may seem to be blurred. If you ask how something is possible, could you not say that inadvertently you are explaining why something happened. However, the term “why” evokes a sense of causality that is not present in “how possible” questions. In the latter position, we are looking to examine thinking at a different level of analysis (see below). 

Problematisations provide a vital linchpin in opening up this level of analysis. Problematisation is a tricky term. Elsewhere, I talk about the two most common uses of problematisation (Bacchi 2012). First, there is the use of the term to signal an analytic approach that questions (or problematises) things. This usage has become part of the vernacular. Think, for example, of references to one’s desire to problematise an issue. 

Second, the term is used to refer to how “things” are problematised – these are the forms of problematisation themselves. Think for example of Foucault’s work (1980) on how different eras have problematized “sexuality” and thus made “sexuality” a particular kind of object for thought in different sites, either as a biological imperative or as part of a moral code.  

This second usage of problematisation is much less common in contemporary theory despite its centrality in Foucault’s work. Foucault talks about a desire to get “inside thinking” and he sees problematisations (the forms themselves; note the use of the plural) as a way to do this. We look to identify and examine how things are problematised. WPR applies both meanings of problematisation. It problematises (asks critical questions about) identified problematisations (the forms themselves). The study of the forms of problematisation themselves, the problematisations (plural), entails examining the presuppositions that make these problematisations possible (see “how possible” questions above). 

Walsh (2022: 28) identifies the importance of this task in WPR. She characterises WPR as concerned with “background knowledges”. This concern, as she says, is the characteristic that most clearly distinguishes WPR from CFA: “WPR offers methods for uncovering the power of background knowledges underpinning policy controversies that an upgraded version of CFA lacks”. 

The meaning of “background knowledges”, the term Walsh adopts to describe WPR, needs elaboration. WPR specifies a concern with the underlying ontological and epistemological premises that make what is said and what is done possible. Consider for example the policy of offering training programs to women to increase their access to positions of influence. A particular understanding of the individual (an ontological presupposition) underpins the proposal, one in which human beings are seen as “skill-acquiring” creatures. This understanding depends upon psychological theories of human development (epistemological assumption/presupposition). There is also an assumption that women are deficient, lacking in some ways, e.g., a gendered assumption about women’s lack of abilities in certain areas. Both psychology and gender are forms of social (“background”) knowledge (discourses). Note that in Foucault and in WPR discourses are knowledges rather than forms of language use.

With Foucault, the goal in identifying these underlying forms of knowledge (“background knowledges”) is to open them up for questioning. We ask: are there ways to think about human beings other than as “skill-acquiring” creatures? Do “skills” exist outside human beings waiting to be acquired? What are the epistemological and ontological assumptions upon which such a proposal relies? And what would happen if we challenged these assumptions, if the “problem” were thought about differently? We could do the same thing with “gender”, asking: are there ways to think about human beings other than as men or women with assumed, differentiated “skill” sets, etc. 

Walsh (2022: 19) describes CFA as a complement to the WPR focus on “background knowledges”. She identifies two different analytic processes: “framing and the (re)production of dominant knowledges”. In her account, the “constructivist” CFA position allows access to framing, described as “what is said”, while “constructionist” WPR focuses on “background knowledges”. The key distinction, for Walsh, is that “an upgraded CFA ensures researchers become intimately familiar with what was said, who said it, and what it means” (Walsh 2022: 20). As an example of what CFA offers, Walsh (pp. 15-16) mentions how centrist and radical right Dutch politicians silence Dutch Muslim women. The example of “cultural essentialism” is put forward as a “background knowledge” (what WPR offers) informing her selected policies – e.g., banning the burka in France. 

A difficulty here, I would suggest, is Walsh’s separation between “what is said” and “background knowledges”. In WPR such a separation does not exist. Recall the focus in WPR on “how possible” questions – how is it possible for certain things to be said, or for certain things to be done? To answer these questions, WPR (Question 2) examines the “background knowledges”, using Walsh’s term (see below), that make it possible to say certain things. The suggestion that we can separate “what is said” from the kind of epistemological and ontological analysis offered by WPR would need to be defended. This point links to the common characterisation of RTA and CFA as somehow more objective and rigorous than WPR (pursued later).

Starting places: Proposals

If we accept the distinctive contribution in WPR just described – a focus on how things said and things done become possible – , we need to see how this approach can be “operationalised”. The key analytic premise in WPR is that proposals to institute forms of change reveal what is presented as needing to change and hence what is produced as problematic. It follows that the starting place for analysis is precisely proposals for change. This argument is a nuanced one. For example, an endorsement of the need for greater social cohesion can be read as a proposal in which lack of social cohesion is constituted as a “problem”. The need to identify proposals leads to the frequent use of legislation or reports as starting places for analysis. From these it is possible to extract recommendations (proposals for change) and then to consider how a particular issue/practice is being problematised within them. To initiate a critical approach to these “proposals”, we move to Question 2 in WPR (see Bacchi and Goodwin 2016: 20) and ask about the presuppositions that make identified proposals possible (see above on “how possible” questions). 

With Sue Goodwin I have developed this way of thinking to embrace a wide range of practices and “entities” beyond public policies (Bacchi and Goodwin 2016). In an earlier entry (14 January 2018), I describe how it is possible to see buildings as proposals. There I explain that buildings and other artefacts can be seen as proposals in the sense that they commit to particular ways of organizing the world. It follows that it is possible to ask: “If this building [or some part of a building, e.g. a purpose-built room or facility] is a statement about how things ought to be, what is seen as needing to change and hence as ‘the problem’?” Bottrell and Goodwin (2011, p. 4) use the example of modern schools with their “uni-purpose facilities located on enclosed land, fenced and gated” and how they reflect a “hidden curriculum” that problematises the moral and cognitive training of young people. 

Proposals provide a starting place for probing problematisations (problem representations) and the (background) knowledges upon which they rely. Problem representations are treated as implicit within proposals. If a program is put forward to train women, women’s lack of training can be identified as what is problematised (“the problem”). This analytic strategy stands at a distance from CFA and RTA where “themes” and “frames” are identified in forms of text, as elaborated in the next section. 

Levels of analysis

In the description above, the “steps” in WPR thinking hopefully become clearer. We start from what is done or what is said. We probe what is problematised in these practices or “statements” (work backwards to identify the implicit problem representations within them). We interrogate the “background knowledges”/presuppositions that make these problem representations possible. These distinctive research questions provide a unique level of analysis. To suggest blending WPR with RTA and/or CFA means asking if they operate at the same or at different levels of analysis. I suggest the latter.

Alvesson and Sandberg’s (2013: 53-56) work on how to develop interesting research questions assists in distinguishing different “levels” of analysis in varied research approaches. They identify five categories of assumptions underpinning specific theoretical stances: in-house, root metaphor, paradigmatic, ideological and “field assumptions”, associating the last with the work of Foucault. Studies that work at the level of “in-house” assumptions tend to accept the frames of reference adopted in their field. By contrast, Foucauldian-influenced research puts these terms of reference in question, unsurprising given the commitment to probe deep-seated presuppositions in “background knowledges”. As Alvesson and Sandberg note, Foucauldian-influenced theory questions “field assumptions” – that is, the assumptions taken for granted across many fields. 

Such an approach puts in question the commonly adopted categories of analysis in mainstream political and social theory. One example is the category of evidence. WPR brings a sceptical approach to presumptions of accessible “evidence”. Such “evidence” is associated with a positivist paradigm and a correspondence view of knowledge. In contrast with the critical approach to evidence in WPR, evidence tends to be referenced as taken-for-granted “knowledge” in those who support CFA (Walsh 2022: 4) and RTA (Dinmore et al. 2024: abstract and passim), suggesting deep-seated epistemological distinctions between WPR and these approaches. 

The acceptance of evidence in CFA and RTA corresponds with the terms mentioned earlier – “coding” and “themes”. According to Walsh (2024: 4), the “coding template” in CFA makes “evidence” “retrievable”, improving “intertextuality” and “accuracy”. The premise in this approach is that bits of knowledge can be identified and labelled and, henceforth, manipulated for research purposes. It is this perspective that leads to the enthusiasm for and defence of RTA and CFA as more objective than WPR. Olvik (2024: 6) develops such a contrast. She distinguishes between WPR, where she argues that “it is important to acknowledge the potential for researcher subjectivity in interpreting” policy documents, and RTA research practices, which ensure “consistency and reflexivity and mitigates researcher bias”. Contra this view, in a WPR approach to “knowledge”, “researcher subjectivity” is taken to be inevitable across the board (i.e. in WPR and in supposedly more objective RTA). For WPR researchers, however, “subjectivity” operates, not as a handicap, but as a research resource. 

Braun and Clarke (2006), who developed RTA, are more careful in their claims about “objectivity”. They do not position RTA researchers as independent of interpretation. Themes, they explain, do not emerge from thin air (Braun and Clarke 2021: 343). Still, they emphasise that RTA involves “Coding interestingfeatures of the data in a systematic fashion across the entire data set, collating data relevant to each code” (Braun and Clarke 2006: 87; emphasis added). In RTA, “the theory and method need to be applied rigorously” where “rigour lies in devising a systematic method whose assumptions are congruent with the way one conceptualises the subject matter (Reicher and Taylor 2005: 549)” (Braun and Clarke 2006: 96; emphasis added; see also Walsh 2024: 4 on “rigor” in CFA). It could be argued that current enthusiasm for RTA and CFA is in part due to the need to satisfy a research community wedded to “rigour” and “systematic” analysis, positions linked to positivist principles.

Braun and Clarke make a key point in recognising the central importance of checking to see if your research assumptions are “congruent with the way one conceptualises the subject matter”. It is at this level that distinctions among research approaches become clearest. While Braun and Clarke (2006: 77 Abstract) argue that RTA is a useful strategy in disciplines “beyond psychology”, the kinds of questions asked in RTA analyses tend to presume the existence of states of being or states of mind (i.e. premises within psychology). Byrne (2022), for example, focuses on deciphering the “opinions” and “attitudes” of research subjects. 

Key premises in this approach sit uncomfortably with a WPR analytic strategy. The presumption of sovereign subjects with “attitudes” is put in question in a WPR approach to subjectivity (Bacchi and Goodwin 2016: 49-53). Indeed, WPR treats “psychology” as a governing (background) knowledge and a contingent historical creation that needs “to be interrogated rather than enshrined as ‘truth’” (Bacchi and Goodwin 2016: 5). On this point Walsh (2022: 22) usefully disclaims “truth” as her standard for assessment, indicating a possible link with WPR’s endorsement of “self”-problematisation. Marshalling empirical material in poststructural accounts becomes possible if truth claims are refused. 

Coming back to “problems”

Clearly, a way needs to be found to clarify distinctions among theoretical approaches. I tend to concentrate on conceptions of “problems”. In Analysing Policy (2009a: 251-252) I produce a “Guide to paradigms in major policy approaches” based on competing conceptions of “problems”. I offer a similar analysis of “paradigms in health policy” in a 2016 article: “Problematizations in Health Policy: Questioning How ‘Problems’ Are Constituted in Policies” (Bacchi 2016). The sub-title signals my intervention target – how “problems” are constituted as particular sorts of problems. 

At the most basic level, a contrast needs to be drawn between positivist endorsements of forms of problems-that-exist and the troubling of “problems” in WPR. Hence, it follows that, to check if different research approaches are congruent, it is necessary to check how “problems” are constituted. 

The contrast between CFA and WPR on this point (how “problems” are conceptualised) is clear in the “sensitizing questions” in Verloo’s (2005) version of Critical Frame Analysis (see Appendix B in Walsh 2022; see also Lombardo et al. 2009). The first questions in the CFA template under the heading “Diagnosis” read: “What is represented as the problem?” and “Why is it seen as a problem?”. 

I wish to highlight the significance of altering the opening question in WPR – “what’s the problem represented to be?” – to read “What is represented as the problem?”, as occurs in CFA’s (Verloo’s) questions. As we saw above, a WPR approach starts from proposals and works backwards to see how the “problem” is represented within them. There is no separation between “problem” and “solution” (or “diagnosis” and “prognosis” as in CFA); they are mutually imbricated. Hence, there is no such “thing” as a problem pure and simple.

The prompt in the CFA sensitizing questions: “What is represented as the problem?” initiates a different mode of analysis, examining how social actors talk about an assumed issue/problem – how they represent something (an issue). It follows that the focus in CFA becomes voice and rhetoric. 

If we examine the other CFA-based templates in Walsh (2022), it becomes clear that problems are presumed (simply) to exist at some level (I drop quotation marks around the term “problem” when it is used in ways that accept problems as existing states of some sort). I offer a few examples. Consider in Appendix D: Template 1 Example, Sensitizing Questions for Collecting the Evidence, S.A.S. v France, 201415 (Walsh 2022, p. 40; emphasis added): “What is the problem(s) identifiedby the speaker? Why is it a problem? Is inequality a problem? If so, what is the inequality?”. 

Consider the contrast between these questions and the many applications of WPR to the question: “What is the problem of ‘gender inequality’ represented to be?” (Bacchi 1999; Bacchi and Eveline 2010). Note, for example, in the selected sensitizing questions, the interest in how speakers identify problems (assumed to exist). By contrast, WPR probes specific proposals in policies and other practices to consider how “gender inequality” is produced as a particular sort of problem in specific contexts (Bacchi and Goodwin 2016: 64-68).

Is it possible to somehow blend these perspectives? Walsh (2022: 22) constructs two templates: a coding template and an analysis template, and “toggles” between them. She recommends “adding WPR questions to the analysis template but keeping them separate from CFA questions”. The goal here is to assist “scholars to maintain clear distinctions between frames and framing on the one hand and background knowledges (read WPR) on the other hand” (Walsh 2022: 22-23; emphasis added). Avoiding a mechanistic form of analysis, Walsh recommends an iterative process that “includes CFA coding with CFA and WPR analysis in conjunction”. 

While I applaud the innovation and clear thinking in Walsh’s intervention, as mentioned above, it is the very idea of a distinction between frames and “background knowledges” that WPR challenges. “Background knowledges” informframes; it could not be otherwise. Walsh (2022: 10; emphasis added) admits as much; however, there is a lack of specificity of just what “background knowledges” entail. In one place, they become “worldviews”: “I found that in each case every frame was informed by the speaker’s worldview and policy position, and that both informed the speaker’s rhetoric and reasoning”. In another, they become biases: “those unquestioned knowledges are likely to include racialized-sexist biases about Dutch Muslim women as oppressed and lacking political agency and hence who cannot speak for themselves” (also Walsh 2024: 5). 

Knowledges in a WPR analysis go deeper than biases which appear to be no more than opinions. Nor is the interest in “worldviews” that inform speakers’ “rhetoric and reasoning”. Rather, the focus is on the deep-seated presuppositions that make different proposals possible. To identify and question these presuppositions, the analytic process needs to start from the WPR questions. 

Conclusion

The examples in this entry indicate the fertile thinking sparked by considerations of how to develop useful political theory. I encourage such thinking and experimentation. Walsh’s (2022) detailed investigation of the premises underpinning CFA and WPR is useful and insightful, partly because it allows us to draw attention to distinctions among analytic approaches. 

Walsh is attentive to these distinctions. As quoted earlier, she (2022: 28) concludes that “WPR offers methods for uncovering the power of background knowledges underpinning policy controversies that an upgraded version of CFA lacks”. In her view, constructionists (WPR users) would benefit from “applying CFA not least because it ensures they escape the accusation of determinism and can help them to uncover resistant perspectives and resistant stories”. This is doubtless a worthwhile goal. There is certainly a need to develop our theories of resistance. I have hesitations, however, about the suggestion that the best way to do this is to construct a dichotomy (distinction) between what is said and “background knowledges”. 

As I develop above, what is said needs to be examined in terms of what could be said, what it is possible to say – what is “sayable” (Foucault 1991, pp. 59, 63). For elaboration of this point, I refer you to a chapter I wrote with Jennifer Bonham (2016), applying this thinking to interview transcripts. Asking “what is sayable” means rethinking conventional approaches to interviewing as a research method that treats “things said” as providing access to people’s interior thoughts and experience. Rather, in interviews and more broadly, “things said” are analysed in terms of the practices, including the knowledge practices, that give rise to them (Bacchi and Bonham 2016: 116; see next entry). This analytic strategy raises questions about the possibility of identifying “frames” (“what is said”) as a separate analytic category, distinct from “background knowledges”.

I look forward to continuing these conversations and exchanges. All the best for your research and writing in 2025. Hoping to hear from you via the WPR list ( Welcome to the WPR Network! | Karlstad University (kau.se)

References

Alvesson, M., & Sandberg, J. 2013. Constructing research questions: Doing interesting research. London: Sage.

Bacchi, C. 2009a. Analysing Policy: What’s the Problem Represented to be?Frenchs Forest: Pearson Education.

Bacchi, C. 2009b. “The issue of intentionality in frame theory: The need for reflexive framing,” in E. Lombardo, P. Meier and M. Verloo (eds) (2009) The Discursive Politics of Gender Equality: Stretching, bending and policymaking. NY: Routledge. pp. 19-35.

Bacchi, C. (2012). Why Study Problematizations? Making politics visible. Open Journal of Political Science, 2(1): 1-8.

Bacchi, C. 2016. Problematizations in Health Policy: Questioning How “Problems” are Constituted in Policies, Sage Open, April-June. DOI: 10.1177/2158244016653986 

Bacchi, C. and Bonham, J. 2016. Appendix: Poststructural Interview Analysis: Politicizing “personhood”, in C. Bacchi and S. Goodwin, Poststructural Policy Analysis: A Guide to Practice. NY: Palgrave Macmillan. pp. 113-122.

Bacchi, C. and Eveline, J. 2010. Approaches to gender mainstreaming: What’s the problem represented to be? in C. Bacchi and J. Eveline (eds) Mainstreaming Politics: Gendering practices and feminist theory. Adelaide: University of Adelaidee Press.

Bacchi, C. and Goodwin, S. 2016. Poststructural Policy Analysis: A Guide to Practice. NY: Palgrave Macmillan. 

Bottrell, D., & Goodwin, S. (2011). Contextualising schools and communities. In D. Bottrell, & S. Goodwin (Eds.), Schools, communities and social inclusion. South Yarra: Palgrave Macmillan. 

Braun, V. & Clarke, V. 2006. Using thematic analysis in psychology, Qualitative Research in Psychology, 3:2, 77-101, DOI: 10.1191/1478088706qp063oa 

Braun, V. & Clarke, V. 2021. One size fits all? What counts as quality practice in (reflexive) thematic analysis?, Qualitative Research in Psychology, 18:3, 328-352, DOI: 10.1080/14780887.2020.1769238 

Byrne, D. 2022. A worked example of Braun and Clarke’s approach to reflexive thematic analysis. Quality & Quantity, 56: 1391-1412. 

Dinmore, H., Beer, A., Baker, E. and Bentley, R. 2024. Advancing a healthy housing policy agenda: how do policy makers problematise housing-related health issues? Journal of Social Policy, doi:10.1017/S0047279424000138.

Foucault, M. 1980b. The history of sexuality, Vol. I. An introduction. New York: Vintage Books.

Foucault, M 1991, “Politics and the study of discourse”, in G Burchell, C Gordon and P Miller (eds) The Foucault Effect: Studies in Governmentality, University of Chicago Press, Chicago, pp. 53-72, ISBN: 9780226080451.

Lombardo, E., Meier, P. and Verloo, M. (eds) (2009) The Discursive Politics of Gender Equality: Stretching, bending and policymaking. NY: Routledge.

Olvik, J. 2024. Navigating Complexity: Potentials and Paradoxes Within Governance Strategies for Public and Third Sector Collaboration. Voluntas, https://doi.org/10.1007/s11266-024-00682-3.

Reicher, S. and Taylor, S. 2005: Similarities and differences between traditions. Psychologist 18, 547/49. 

Stoor, J. P. A, Eriksen, H. A. and Silviden, A. C. 2021. Making suicide prevention initiatives targeting Sámi in Nordic countries. BMC Public Health, 21:2035. https://doi.org/10.1186/s12889-021-12111-x  

Verloo, M. 2005. Studying Gender Equality in Europe. European Studies Newsletter, 24(3/4): 8-10.

Walsh, D 2022. An Integrated Approach for Doing Comparative Discursive Policy Analysis. Prepared for presentation at the American Political Science Association Conference September 14-18, https://preprints.apsanet.org/engage/api-gateway/apsa/assets/orp/resource/item/634b14dc3e8d9944361b3d76/original/an-integrated-approach-for-doing-comparative-discursive-policy-analysis.pdf

Walsh, D. 2024. A complementary approach to Critical Frame Analysis and “what is the Problem represented to be?”, Critical Policy Studies, https://doi.org/10.1080/19460171.2024.2383194