Keywords

Introduction

The previous chapter outlined the strategies and specific steps we took to code interview, ethnographic and documentary qualitative data collaboratively. This chapter explores the epistemological and methodological steps involved in interpreting qualitatively coded data in greater detail. For many scholars, interpretation ‘happens wherever and whenever meaning is made’ (Willig, 2014: 137) thereby ‘creating a new narrative from the data’ (Reyes et al., 2021: 6). When a researcher asks, ‘what does this mean?’, the number of answers can be unlimited as there exist variant ways to interpret the same thing. Thus, interpretation is always shaped by assumptions made by the researcher in the form of previous knowledge, personal preferences and the researcher’s experience and background in terms of class, race, gender, (dis)ability and sexuality. It is also conditioned by what is, or is not, available in the data—i.e., own limitations. Put more directly, how we interpret our data is not only shaped by our positionality, but also by the epistemological and ontological views we harbour prior to embarking on the project of interpreting (for more on positionality, see Ackerly & True, 2020). Whilst interpretation can be shaped by the above factors, it can also simultaneously generate different types of knowledge, which means that researchers are responsible for assessing, criticising and restricting the generalisability of their findings (Lewis et al., 2003; Schwartz-Shea & Yanow, 2012).

As stated in previous chapters, different stages of qualitative research—such as data gathering and data coding—may be conducted at the level of the broader research project or at the level of research articles. In that sense, Chapter 4 showed that the document data was mainly selected to match the research design and questions of specific research articles, whilst interview and ethnographic data was gathered according to the broader objectives of the project. Similarly, Chapter 5 explained how we coded the interview and ethnographic data collaboratively using a coding strategy designed for the whole project, whereas document data was coded with a coding scheme designed for research articles. This chapter is exclusively attentive to the interpretation of data relative to research questions developed for the purpose of research articles, some as single-authored and others with up to six co-authors.Footnote 1

Qualitative research is often based on analyses that are driven by research questions (Blaikie, 2010; Schwartz-Shea & Yanow, 2012). We applied this approach so that all the questions asked in individual research articles matched the broader questions of the project. Likewise, all research articles contributed to the empirical, theoretical, and methodological findings of the broader project. The broad initial research questions of the project were: ‘how does gender create fault lines between and within political groups?’; ‘how do gendered norms impact formal and informal practices in the EP?’; and ‘how are gendered policies and practices in the EP and the political groups shaped by political ideologies?’ Whilst these research questions informed particular articles (as overall research aims or goals), we sometimes had different—narrower or broader—aims and objectives in other publications.

Some research articles focused on the interpretation of frames and discourses used by political groups on specific policies. For example, when considering sexuality and human rights we asked, ‘how do the political groups in the EP understand and (re)frame human rights?’ (Ahrens et al., 2022); on social and economic policies we asked, ‘how is gender equality sidelined in the EP on EU economic governance’ (Elomäki, 2021); and ‘how do dominant ideas in the EU’s economic governance shape the constructions and frames of economic and social issues?’ (Elomäki & Gaweda, 2022); and on gendered violence we asked, ‘how are sexual harassment, and solutions to it, discursively constructed in the EP by the political groups?’ (Berthet & Kantola, 2021); and ‘how are support and opposition to the EU’s ratification of the Istanbul Convention on violence against women and domestic violence constructed in the EP by the political groups?’ (Berthet, 2022a).

Other publications, however, focused on interpreting the formal and informal practices of the political groups. For example, in relation to the formation of political groups we asked, ‘how are the formal and informal practices of political group formation gendered in the EP and what does it mean for democracy?’ (Ahrens & Kantola, 2022); in relation to Brexit we asked, ‘how was the impact of Brexit constructed in the EP, and how did it affect UK MEPs’ parliamentary work?’ (Kantola & Miller, 2023). When we considered the dynamics of intra-group policy formation, we asked, ‘how do political groups formulate group lines on policies and how does it impact democratic decision-making in the EP and intra-group democracy?’ (Elomäki et al., 2022); and in relation to national party delegations (NPDs) we asked, ‘how is the role of NPDs constructed within the political groups and what differences does this elicit between them? What formal and informal institutions are at play?’ (Elomäki et al., 2023). Finally, some research articles used the data in a different way, namely, not to interpret constructions, but rather to trace the development of a specific policy within the parliament. Good examples of this were the enquiry into the parliamentary process of the European Semester policy developments (Elomäki, 2021) and into those that led to the adoption of the Matić resolution on sexual and reproductive health and rights in Europe (Berthet, 2022b).

The Role of Code Reports in Interpretation and Analysis

The first step towards interpretation was to export the code report from ATLAS.ti. The previous chapter covered the details of coding qualitative data in ATLAS.ti, the different tools and functions offered by the software to code and become familiar with the data and to begin analysing it by employing labels, categories and revealing patterns. We regarded the step of exporting code reports as the first step of interpretation. As we explained in Chapter 5, a code report can retrieve all quotations from one code or several codes, or all quotations at the intersection of one or more code(s) and of one or more document(s). Once retrieved, the reports consist of pages of quotations from interviews or ethnographic notes; its length will depend on the total amount of data and the level of fine-grained analysis the researcher exporting the report is seeking.

First, we began by reading the code report, either as a whole or by dividing it amongst co-authors. If the publication is to be co-authored, the reports can be divided either by categories (e.g., team A reads political groups X; team B reads political groups Y; or team A reads female respondents; team B reads male respondents), or by sections of analysis planned in the research article (e.g., team A is assigned the part on informal practices and will read the data through that lens). Whilst reading the report is an individual task, it is imperative that regular meetings are scheduled to fully discuss this first stage of analysing the data with all co-authors.

Even if we divided the analytical work between team members, we often read whole interviews to ensure we did not ‘take things out of context’ or mistake the participants’ meaning. This also depended on the research questions and led us to reflect on the limitations of team coding. With qualitative research, context and subtext are key for meaningful interpretation of the data and at times, limited quotations (coded by someone else) could be confusing. Our solution was for each team member to be as familiar as possible with all the relevant data that was used for any particular publication. We also used different types of data side-by-side (see Box 6.1 for an example), reinforcing our interpretation by using videos, debate transcripts and parliamentary documents. Depending on the type of article being produced, we could change the significance or weight of data we relied upon.

Box 6.1 Triangulating different data sources and using a different coding scheme

Elomäki and Gaweda (2022) developed a new coding scheme that corresponded to their specific research questions, because they used data that mostly consisted of EP committee documents. They triangulated the research material to juxtapose, compare and contrast the data from EP documents, debates and interviews, to enhance credibility and increase the validity of their interpretive outcomes. Specifically, they coded the document material in ATLAS.ti to structure the extensive data for discursive and interpretive textual analysis, and to allow for comparisons between committees and groups.

The coding system was developed deductively and inductively based on the literature attentive to the hierarchy between social and economic goals. They designed and discussed it in multiple meetings to ensure consistency and coherence between the coders. The coding system aimed to identify discursive constructions of the social-economic relationship (for example economy prioritised over social goals or the reverse), the specific social policy issues discussed (for example poverty, different care services, etc.) and discursive constructions of these issues (for example, as a cost, an investment, valuable in itself, applying labour market logic). In the case of this article, the authors also analysed videos of 20 committee debates corresponding to the reports, following a schedule of sensitising questions on political conflicts and policy content. The committee debates provided interpretive material for the analysis of the political and ideological context of the discourses. Finally, they only used the project interviews dataset to contextualise, explore meta-narratives and nuance within political group positions, and to gain insights into the policy-making processes (Elomäki & Gaweda, 2022).

Several factors combine to justify the need to co-author articles in qualitative research. Two of our research articles were co-authored by all six researchers involved in the project. One consisted of analysing the normative whiteness and racism in the European Parliament using the code ‘Racism’, and generated unforeseen research findings that exceeded the initial objectives of the project (Kantola et al., 2023). The other article involved the analysis of the power dynamics between national party delegations in the European Parliament. Co-authoring it as a whole team made sense, as the code ‘National Party Delegations’ was the biggest we had returning over 490 quotations. Thus, it was more efficient and logical to analyse the code with more researchers. For the latter publication, each researcher or group of two researchers was in charge of reading the ‘National Party Delegation’ code report for one big political group or two small political groups. After reading and extracting important parts of the text, joint discussions of findings (including patterns and differences) amongst all researchers were key to systematise and remain consistent with the analysis.

Interpreting Frames and Discourses

The analysis of qualitative data may involve the interpretation of frames and discourses prior to analysing policies and policy developments, formal and informal practices, and parliamentary processes. Epistemologically, we approached knowledge as constructed (Yanow, 2006a), thus the methodologies relevant to us were interpretivist and constructivist. The former relies on the belief that the analysis of human actions and practices is possible by interpreting the meanings that actors attribute to actions, practices, and the institutional environment in which they operate (Bevir, 2006: 283). Put simply, interpretative qualitative research is typically unconcerned with inferring or hypothesising the ‘one and only’ truth from the data, rather it is focused on analysing and attempting to understand the variant constructions that emerge from the data. For instance, when analysing the role and meaning of expertise in the European Parliament, researchers may not be interested in the subject and the object (i.e., the expert and the expertise), instead they will be attentive to who is constructed as an expert, what is constructed as expertise and how does it impact policy developments in parliaments (for an example of constructing expertise in the EP, see Elomäki & Haapala, 2023).

Similarly, when analysing the role and power of national delegations in the EP, our team was not interested in quantifying the power of national delegations, but rather in analysing which delegations were constructed as powerful and why (Elomäki et al., 2023). Thus, constructive and interpretative methodologies may involve questioning which discursive constructions are dominant, and which are silenced, in order to look for hegemonies, power dynamics and omissions. Rather than ‘truths’ or objective facts, interpretivist scholars tease out textual substance to be interpreted. With that in mind, scholars have argued that the attribution of meaning to actions, practices and to their institutional environment is best explored through an analysis of frames and discourses (Lindekilde, 2014)—which, in turn, are best understood using a qualitative toolkit (Bevir, 2006).

Although there exist a variety of ways to interpret frames and discourses, they are all concerned with how language (i.e., talk or text) constructs social realities (Willig, 2014), or to assert that discourses are socially constructed, and consequently play a predominant role in constituting the social (Blommaert & Bulcaen, 2000: 448). Interpretivist and constructivist methodologies acknowledge the constraints that broader discursive environments impose on individual discursive practices (Ferree et al., 2002). Language or discourse are not ‘transparent tools’ and thus require a significant degree of interpretation as to what is constructed and how (Bacchi, 1999). Therefore, one key site of analysis is the discursive battles over meanings that are played out between various actors, and the consideration of the constraints imposed by their institutional discursive environments (Lindekilde, 2014). In our case, this has meant, for instance, studying the discursive constructions around gendered policy issues, the differences and similarities between political groups, the fault lines within the groups and to embed/contrast those within broader discursive frames, such as the self-promoted narrative of championing gender equality in the European Parliament.

In our epistemological approach—which is interpretivist, constructivist but also often post-structural and feminist—we understand discourse as that which is infiltrated by power relations, because power is omnipresent and performative (Foucault, 1972, 1980). The post-structuralist approach in terms of discourse analysis emphasises both discursive and non-discursive elements of social reality, such as institutional practices and norms guiding behaviour (Panizza & Miorelli, 2013). Importantly, these are mutually constitutive, which allows for a fuller understanding of the role of discourse in creating power (and inequality). From this point of view, the social orders we observed in institutions were never fully structured, but they were open to political interventions and dislocations that made it possible to ground or subvert them (Panizza & Miorelli, 2013: 302). From our standpoint, social phenomena are not purely discursive or linguistic, but for things to be intelligible they must exist as part of a wider framework of meaning and discourse (cf. Fairclough, 1995; Jorgensen & Phillips, 2002; Panizza & Miorelli, 2013).

This reasoning is especially important when feminist methodology is an ‘epistemology in action’ (Weldon, 2006). Typically, a feminist perspective implies a critical approach, one aimed at creating social change or exposing social injustice and inequality. The position of the researcher, and individual choices in terms of methods whilst reflecting one’s ontological and epistemological commitments, have implications in feminist research that differ markedly from traditional positivist social science. The approach moves beyond a determinist and traditionally positivist concept of causality, providing instead a reflexive perspective and a contextualised and dynamic way of interpreting meaning (Kulawik, 2009: 263). When combined, different forms of knowledge will arguably produce, not a claim to universal understanding, but rather a broader, albeit contingent, understanding of the nexus of gender, institutions, power and discourse in legislatures (Jorgensen & Phillips, 2002: 155).

In other words, making sense of how discourses maintain, challenge and transform (unequal) power relations within a given institution (Fairclough, 1995; Wodak, 1996), is a critical approach to the analysis of discourse (Willig, 2014). Discourse can be seen as a form of social practice that both constitutes the social world and is constituted by other social practices. ‘[T]he discursive constitution of society does not emanate from a free play of ideas in people’s heads, but from social practice which is firmly rooted in and oriented in real, material and social structures’ (Fairclough, 1992: 66). Thus, the ability to define social realities by making one discourse or frame dominant ‘is an act of power with important consequences for social practices’ (Lindekilde, 2014: 199).

In gender and politics scholarship, this has meant conceptualising social structures, such as gendered inequalities, as cemented by power relations (Kantola & Lombardo, 2017). A feminist approach to (critical) discourse analysis cannot remain descriptive and neutral, since the interests guiding it aim to uncover or make transparent processes and mechanisms that perpetuate injustice, inequality, manipulation, sexual discrimination in both overt and subtle, pernicious forms (Sunderland & Litosseliti, 2002: 20). Therefore, analysing the discursive layers embedded in institutions can help understand the processes through which power moulds these institutions in the form they take (Sunderland & Litosseliti, 2002). In this regard, discursive practices influence what can be said, achieved and reformed in an institution (Bacchi, 2009; Lombardo et al., 2009). Critical frame analysis is an additional tool for analysing discourses. It calls for reflection on both the discourses within which actors operate, and the active deployment of concepts and categories for political purposes (Verloo, 2005; Verloo & Lombardo, 2007). A framing methodology shows the ways in which the framing of a concept or policy affects how policymakers and legislators think about an issue (Forest & Lombardo, 2012). Such methodologies are useful for interpreting the discourses and frames around policies or policy developments in parliaments.

Whilst there are similarities between discourse and frame analysis, they can be used to answer different research questions by identifying the different meanings a concept holds (Bacchi, 2009; Lindekilde, 2014; Lombardo & Meier, 2008; Lombardo et al., 2009; Roggeband & Verloo, 2006; Verloo & Lombardo, 2007). This is best illustrated for our purposes, through the ways in which gender equality is a disputed concept. Born out of social movement studies, frame analysis seeks to identify ‘how particular ideas/ideology are used deliberately to mobilise supporters and demobilise adversaries vis-à-vis a particular goal’ (Lindekilde, 2014: 200; Snow & Benford, 1988). It is used in other scholarship to analyse the strategic framings of a particular problem. For Bacchi (2009), problematising a policy issue leads neither to an objective description of it, nor to objective solutions for solving it. Rather, it is part of creating the problem. A framing methodology shows the ways in which the framing of a concept affects how policymakers and legislators think about an issue.

Previous research on the European Parliament uses interviews as data to gain ‘objective’ information (as insights or direct records) on what had happened behind closed doors, such as in committee negotiations or in trilogues (Bressanelli & Chelotti, 2018; Ripoll Servant & Panning, 2019). Whilst our main focus was on discourses and framings, we also used the qualitative data to trace policy processes, analyse policy-making practices and identify obstacles for the promotion of gender equality within the political groups and in the European Parliament. For instance, some of our research was driven by the question why pro-gender-equality amendments made by progressive political groups and MEPs often disappeared throughout the committee negotiation process? Interviews with MEPs and staff from different political groups who were involved in negotiations about the specific reports, helped us to identify some of the dynamics that led to the sidelining of gender equality. Gender equality was not necessarily a priority for pro-equality groups in the negotiations, and it was opposed by some groups. Moreover, some of the political groups that made amendments about gender equality did not have enough leverage in the negotiations to push their views through, or may not have invested their resources on negotiation about reports they know will be voted against (Elomäki, 2021).

Policy and Process Tracing

In the policy-focused articles, we used the interview data as information to trace policy processes and policy-making practices. Whilst not looking for causality, or indeed to make claims about it, we used policy process tracing as a complementary method, or an additional level, to fully understand and interpret the discursive and non-discursive elements. As such we traced, outlined and connected the stages of a particular process, which enabled us to identify the power hierarchies, interplay of different formal and informal norms and the contingent reasons for the emergence of gendered inequalities in both institutional policy and practice. For instance, since some interview participants were experts on economic issues, either as MEPs, political group staff or committee staff and represented a fair balance of political groups, we used this approach to study the development of economic policies in the parliament. For those experts, the interview questions were highly specific and delved deeper into the specificities of some policy processes. This approach elicited rich and multifaceted knowledge about parliamentary processes, boosting our within-case analysis. However, we also experienced the well-documented difficulties of using interviews as sources of ‘objective’ information. Indeed, in an interview narrative the unfolding of events may be influenced by memory effects, unwillingness to answer questions, strategic misrepresentation of events and the tendency of actors to under—or over-represent some events or their own role in them (e.g., Berry, 2002; Beyers et al., 2014; Fowler et al., 2011).

Not surprisingly, it became evident that most of our interviewees wanted to present themselves and their political groups in a favourable light. Concomitantly, they were not necessarily ready to provide sensitive information. For instance, our data related to the moment when two MEPs had come to an agreement for a committee position on a specific file and represented themselves as winners of negotiations. They omitted to mention, however, the dissatisfaction of the coordinators in the outcome—as stressed by other interviewees who took part in the policy process. Similarly, our data stressed a number of contradictory accounts amongst interviewees in relation to policy-making processes at play in the political groups, with some interviewees describing how everyone’s opinion is allowed, whilst others from the same group stressed the silencing of dissenting voices. Such contradictions show the difficulty of using interviews as accurate evidence about the unfolding of events.

Therefore, even when asking research questions about processes and practices, we acknowledge that interviews do not provide access to an ‘objective’ reality, but are always a construction based on perceptions. Interview data can certainly point one in a certain direction to find out more, but it needs to be complemented with other sources to obtain a fuller picture whilst simultaneously addressing possible biases in the data (Natow, 2020). At the same time, our commitment to interpretivist, constructivist and feminist qualitative epistemological research, ensured that claims or assumptions of reaching ‘objective’ truths or determining causality, were never made (Yanow, 2006b). Unlike positivist empirical methods designed to generate results that can be replicated by different scholars, interpretivist, constructivist and feminist can yield different outcomes in the hands of different researchers. This highlights the collective self-reflective and deliberative nature of such approaches (Ackerly et al., 2006: 7).

To counterbalance the limitations of interview data, we triangulated it with document data, such as parliamentary debates, amendments and political group documents—thus providing a more complete grasp on specific policy processes (see Box 6.1 above for an example; for more on triangulation, see Natow, 2020). This helped with cross-validating and interpreting the evidence obtained through interviews, and allowed for the integration of additional information (Beyers et al., 2014). In turn, ethnographic data provided an additional important tool to analyse the practices of political groups, enriching our knowledge of parliamentary practices and policy processes.

Interpreting Ethnographic Data

Because ethnographic fieldnotes are subjective, even personal, and sometimes written down in conditions not always conducive to note-taking, the generated data is by its nature difficult to share with other researchers for analysis. For the ethnographer, this meant several difficult and thorough rounds of rewriting fieldnotes to make them intelligible to others (Jarzabkowski et al., 2014). In this respect, it was important to establish a protocol, shared with all other researchers in the team, that structured the process of taking field notes (see Chapter 3 for the template), and which allowed us to co-author research articles with our ethnographer. Overall, we took a structured approach to the analysis of fieldnotes.

There are different ways of interpreting and presenting evidence from ethnographic data (Cerwonka & Malkki, 2007; Schatz, 2009; Shore et al., 2011), and this is further reflected in the variant scholarship that employs parliamentary ethnography (Abélès, 1993; Crewe, 2018; Miller, 2022). Like interview data, the analysis of ethnographic data occurs across different research stages and is influenced by decisions made at various moments. For example, designing the research idea, formulating the questions, deciding on a protocol to record fieldwork interactions and writing down fieldnotes. Whilst it is important to avoid making ‘instant interpretations’ in order ‘to remain as reflexive as possible’ (Niemi, 2010: 89), Ackerly and True note that the process of ethnographic data production and analysis are inextricably linked (2020: 190). Sometimes, the goal is to open up the black boxes of what is little known or understudied to produce a thick description.

In our case, the analysis of ethnographic data was abductive and nonlinear. It can be referred to as both a formal and an informal process. The process was formal because it took shape in analytic notes and memoranda, and it was informal because it was embodied in the ethnographer’s ‘ideas and hunches’ (Hammersley & Atkinson, 2019: 167). For instance, we used informal strategies, such as ‘hanging around’ with research participants, to discuss research ideas and test whether they resonated with them. Such informality can generate shared understandings with research participants, bring about new perspectives, or be shut down at once.

We included ethnographic analytic notes or memos into the fieldnotes, either as appendages at the bottom, or weaved them through the text when writing them up, but we acknowledge that they can be written in a separate document. In terms of the formal placing of the analysis amongst the fieldnotes, we eventually wrote the analytic notes and fieldwork diary in conjunction and merged them as ‘raw’ data. The observation protocol included a section for ‘reflections’, in which the ethnographer reflected on the observation just made (see Box 6.2 for examples). The ethnographer noted their positionality in the observation protocol, though they were inevitably present in the whole field note as they were noting the dynamics that they saw as relevant. In other political ethnographies, ethnographers have one column of ‘raw’ fieldnotes and then an analysis column, or if handwritten they write with a different pen.

Box 6.2 Ideas for prompts in analytical ethnographic notes

  1. (1)

    Using references to literature and ideas from the ‘raw’ data: for instance when a participant mentions topics that make the researcher think, or academic literature they read on the issue, even using random associations.

  2. (2)

    Discussing emerging interpretative ideas with research participants, for instance ‘running them by’ field members to see if they resonate in informal conversations or settings.

  3. (3)

    Documenting one’s own surprises about observations in the field: noting down researcher’s own emotional reactions and reflecting later why that happened.

  4. (4)

    Documenting the research process in terms of theoretical sampling: for instance, trying to get an interview with interpreters or other field participants, who might otherwise be seen as ‘bystanders’ in the political processes. This is useful to discuss affective dynamics, since the interpreters, for example, implied to us that they sensed the mood of group meetings. Interestingly, some interpreters interpret empathetically, becoming key agents in palpably gauging the mood in the room, which is valuable when looking into affects and affective atmospheres.

  5. (5)

    Comparing shadowing experiences from the locations of two differently situated participants that might be working within the same group and committee to discern patterns and divergences.

  6. (6)

    Marking links to other sources to triangulate ethnographic fieldnotes, either using the interview dataset or own observations in the field—notes, for example, about the location, posters, images and embodiment.

  7. (7)

    Comparing unusual (crisis) situational contexts with ‘normal’ contexts: for instance, when unexpected circumstances affect regular institutional procedures. In our case, it was the onset of the Covid-19 pandemic and announced restrictions to the EP President’s conference. We noted how exceptionally open Sassoli’s press conference was for ordinary people in the parliament.

  8. (8)

    Situating parliamentary powers: in the notes it is useful to reflect on the institutional context the researcher is studying, and to compare it with other political institutions they know along power relationship lines, especially in similar circumstances—how, for example, does the president of the EP behave relative to a national parliamentary speaker?

As mentioned previously, ethnographic scholars debate the possibility of sharing ethnographic data with other researchers for interpretation (Murphy et al., 2021; Reyes, 2018; Tsai et al., 2016). When ethnographers of parliaments consider sharing their fieldnotes data, two difficulties are often raised: (1) the long-standing feeling that data is ‘hard won and the result of personal, trusting relationships’ built over years—which requires time away from the desk and less opportunities to craft findings into publications; and (2) the lack of full control over confidentiality and anonymity required when citing ethnographic data in research articles, when this was key to the trust built with participants (Murphy et al., 2021; Tsai et al., 2016). For this reason, our ethical statement specifically included the requirement to not harm research participants—which in ethnography with elite participants may mean not harming their political career. Furthermore, although metadata about the fieldnotes was included, we shared the fieldnotes with the team and co-authors, though not more widely. Murphy et al. (2021) do offer practical solutions, such as placing an embargo on data, and going back to subjects to ask them if they consent to archiving their stories.

With regard to the practicalities of presenting ethnographic data, we often used it as ‘raw data’ in research articles, but other ways of presenting ethnographic evidence that conveys dynamics of parliamentary worlds include vignettes, composite narratives that link together several actors and interactions from the field, and more temporal ‘process’ narratives (Jarzabkowski et al., 2014). In the course of our analyses, we also searched the ethnographic field notes for dissenting voices. This was especially important in the case of publications on more ‘invisible’ and problematic issues, like the article on institutional racism and normative whiteness in the European Parliament (Kantola et al., 2023).

Ethnographic data allowed us also to interpret wider facets of parliamentary institutional behaviour and norms than documents or interviews alone would have permitted. Thanks to specific codes we developed to capture affects and emotions, we were able to interpret the use of strong language and figures, ‘affective atmospheres’ (as more collective emotional entities rather than just a person expressing a feeling), as well as the observation of ‘tense’ or ‘businesslike’ interactions in meeting or encounters in our research articles (Kantola & Miller, 2021: 788). For instance, in their article on the affective impact of Brexit in the EP, Kantola and Miller (2023) did not measure the effectiveness of parliamentary work or politicians’ motivations and performance, but rather analysed its dimensions. Specifically, they covered the influence of emotion and affect on the constructions of parliamentary work, finding that these constructions were charged with emotions including sadness, joy, hope, civilised jubilation, relief, resolve and vigilance and that these were expressed and controlled (Kantola & Miller, 2023). The findings were largely based on the ethnographic material and contribute to studies on parliamentary work by moving away from more positivist and rational choice versions of role theory and categorising the roles of parliamentarians, to mapping how they construct different dimensions of parliamentary work (Kantola & Miller, 2023), thereby demonstrating the ‘added value’ of utilising ethnographic data.

Conclusion

This chapter finalised our section on analysing qualitative data by casting a critical eye over the epistemological and methodological steps of interpretation. Whilst many different ways to analyse qualitative material exist, we were attentive to the important aspects involved with the interpretation of qualitative data, in particular, how it is shaped by the researcher’s own positionality and epistemological and ontological disposition. These factors serve to generate different types of knowledge, meaning that qualitative researchers are required to assess, criticise and restrict the generalisability of their findings. The interpretation of qualitative data is typically driven by research questions. In our case, this has meant interpreting frames and discourses, analysing formal and informal practices, tracing policy processes and policy-making practices and interpreting ethnographic data. Whilst interpreting frames and discourses involved identifying different constructions that emerged from the data, and analysing them via interpretivist, constructivist, post-structuralist and feminist methodologies, tracing policy processes and policy-making practices significantly helped to identify power hierarchies, the interplay of different formal and informal norms as well as the emergence of gendered inequalities both in policies and practices. Finally, we have discussed the benefits and limitations of sharing ethnographic material and interpreting it for analysis. A difficult process eased by the use of ethnographic analytic notes or memos, fieldwork diaries and an observation protocol that reflected on the ethnographer’s positionality.