Researching without asking actors directly: the value of interpretative methods in studying gender mainstreaming implementation in the European Parliament

ABSTRACT This article discusses interpretative methods as well-suited approaches to researching norm- and value-driven topics, such as gender mainstreaming (GM). Using the European Parliament as a case study, the article introduces the idea of ‘flipped expert interviews’, extending the interview population to non-experts, as an approach to study GM. It interrogates the advantages and limitations of this method in capturing power relations and the formal and informal practices and processes in the EP’s committees and political groups. The article suggests that going beyond ‘usual suspects’ when selecting the interview population helps to paint a broader picture of contextual factors and micro-dynamics that illuminate processes of politicization around contentious topics.


Introduction
Scholars using interpretative methods explore the 'centrality of meaning in human life in all its aspects' (Yanow and Schwartz-Shea 2013, xiv). In political science, this translates into studying actorness, practices, micro-politics, power relations and political institutions (Egeling and Wiesner this collection). Interpretative methods are well-suited to research norm-and value-driven topics, such as gender equality, as actors often negotiate these across formal rules and procedures on the one hand, and informal rules, personal positions and power relations on the other True 2010, 2018;Kenny 2007;Krook and Mackay 2011;Waylen 2017).
Interviews with laypersons, experts or political elites are one of the core methods to collect narratives and shared stories through listening (Lareau 2022;Yanow and Schwartz-Shea 2013). Extant literature on interviewing emphasizes the importance of reflexivity and positionality in interview practices as these impact the social dynamics at play, including gender aspects True 2010, 2018;Brown 2018;Fujii 2017;Lareau 2022;Yanow and Schwartz-Shea 2013). Aware of these social dynamics, we carefully considered our and interviewees' positionality (and potential interactions) in each step of the research process (Ackerly and True 2010;Fujii 2017;Soedirgo and Glas 2020).
We join the conversation in interpretative methods of studying formal and informal institutions (Krook and Mackay 2011;Holmes et al. 2019;Waylen 2017) by discussing our approach to examining GM implementation. The EP, understood as a social organization with a certain power structure affecting policies and polity, invites exploring the role of actors and their (micro-)political practices and processes around norm-driven policies. Herein, we discuss methodological challenges when conducting expert interviews on gender mainstreaming (GM) implementation in the European Parliament (EP).
GM is a political strategy to promote gender equality adopted in many countries (UN Women 2020) with most EU member states committed to it since the mid-1990s (Jacquot 2015). GM puts the responsibility for promoting gender equality on all actors and requires them to tackle structural gender inequalities and gendered institutional practices through considering gender aspects in all public policies and phases of policy-making (European Parliament 2022). 1 Yet, implementation has been patchy (Jacquot 2015;Minto and Mergaert 2018). Studies showed actors operate with everyday 'gender knowledge' instead of gender expertise (Cavaghan 2017), opposition turns GM into a contentious topic (Mergaert and Lombardo 2014;Kuhar and Paternotte 2017;Verloo 2018) and difficulties originate from constantly (re)negotiating policy goals and implementation processes (Cavaghan 2017;Lombardo, Meier, and Verloo 2009;Meier and Celis 2011). The understandings of what gender equality entails and what the goal of GM should differ often substantially between gender equality actors and 'mainstream' actors, and between individuals, interest groups and political parties (Cavaghan 2017;Kantola and Rolandsen Agustín 2019;Lombardo, Meier, and Verloo 2009;Mergaert and Lombardo 2014).
The rich research on GM implementation highlighted the limited reach and lack of institutional transformation despite the original political goals. Researchers often interviewed exclusively gender equality actors on GM, this means, persons either tasked with GM in their job portfolio or generally proactive in promoting gender equality. Given our own experience with GM implementation (Ahrens 2019;Elomäki and Ahrens 2022) and the premise that GM should be 'everyone's task' in the EP, we reconsidered this previous interview strategy. We assumedif implemented successfully -GM implementation should become detectable when talking to any actor of an institution implementing GM. Thus, we inquired about the subject in any interview we could get, regardless of a formal GM responsibility. This originates from the assumption in interpretative approaches that each interview provides useful data (Fujii 2017;Lareau 2022;Yanow and Schwartz-Shea 2013). We label this approach as 'flipped expert interviews' as it is aimed at interviewing non-experts on GM. By doing so, we hoped discovering whether GM 'trickled down' and was embedded in EP daily practices and EP bodies, thereby allowing us to capture a comprehensive overview of GM implementation.
The potential pitfalls we imagined for this specific 'mainstream' interview population was how to gather information from interviewees on a topic, GM, that is probably of no interest to them (or they even oppose) but they are supposed to keep in mind. Additionally, we expected that asking directly would produce prefabricated empty phrases instead of open-minded narratives because political actors usually discursively support gender equality as an important global norm (Cavaghan 2017).
In the following, we present methodological lessons learned during the research process. We contribute to the exchange on interpretative methods by illustrating (a) how to interview experts on a topic they formally need to attend to but probably disregard, while (b) not asking them directly about the topic as this could induce political window-dressing. We start by providing details of the research context and broader research project, including on previous GM implementation studies using interpretative methods. Then, we reconstruct the research process considering interpretative methods and how the idea of 'flipped expert interviews' evolved and was employed. We then review its suitability and challenges, also regarding data analysis. We conclude with discussing how useful such interviews are for other research questions, particularly for other norm-driven politically contentious issues which tend to produce prefabricated empty phrases.

EP context and research setting
Researching GM implementation in the unique institutional context of the EP is a methodically demanding case in many regards. The EP is one of the rare parliaments worldwide which committed to GM in its own internal processes (Ahrens 2019;Elomäki and Ahrens 2022;Sawer 2020) and played an important role in integrating a gender perspective in the EU's policies (van der Vleuten 2019). Gender equality and GM feed party-political conflict between value-conservative and value-liberal parties (GAL-TAN) (Ahrens, Gaweda, and Kantola 2021;Elomäki and Ahrens 2022;Kantola and Rolandsen Agustín 2019), with growing numbers of anti-gender MEPs directly opposing gender equality (Zacharenko 2020). Conflicts about GM implementation also appear between the Women's Rights and Gender Equality Committee (FEMM) and other EP bodies, notably other committees (Lodovici et al. 2018). The process of GM institutionalization in the EP is long and still ongoing requiring to consider this history when interviewing people who only recently started working in the EP. Furthermore, the multi-language and multi-national setting of the EP required tailoring data collection to this specific setting (Sarikakis 2003).
Problems with GM implementationor comparable mainstreaming policies such as sustainability or new public managementoccur in any organization and thus pose similar methodological challenges to researchers using qualitative expert interviews. Our specific interest is in formal and informal rules shaping the everyday practices around GM in the EP and '[t]hrough listening to the stories of individuals tasked with implementing a particular institutional policy we find out something about the broader institutional value placed on it' (Holmes et al. 2019, 215). Meaning-making stories are 'central to understand significant dimensions of causality' (Yanow and Schwartz-Shea 2013, xiv), in our case for measuring the formal and informal rules of GM (non-)implementation.
Early GM studies focused on the adoption process, particularly women's movements mobilization around it, and took a macro-level perspective, for instance, on norm diffusion to member state level, yet often disregarded follow-up processes within institutions (Cavaghan 2017, 27-29). Other GM studies used critical frame analysis and country-or institution-comparisons, including public policy analysis, to examine different understandings of GM and institutional mechanisms in implementation (Lombardo, Meier, and Verloo 2009;cf Cavaghan 2017 for an overview). Cavaghan (2017, 38), in contrast, developed a gender knowledge contestation analysis to trace 'how policy is always collectively interpreted through pre-existing practices' and revealed contradictions between everyday 'gender knowledge' and gender expertise as a barrier to GM implementation. Studies on GM in parliaments are a research gap, potentially due to the scarcity of parliaments' self-commitment to this strategy (but see Sawer 2020).
Compared to previous GM implementation studies, our research benefitted from being embedded in the EUGenDem-project studying the EP's political groups and their gendered policies and practices. 2 Thereby, instead of focusing exclusively on GM implementation, our research evolved as one cross-cutting aspect threaded into the research process. We presumed many insights about the role of gender in policies and practices are, in fact, insights about GM, even when they are not explicitly named as such. Methods and data collection comprised three broad venues: (1) qualitative semi-structured expert interviews with MEPs and EP staff; (2) parliamentary ethnography 3 ; and (3) content analysis of EP documents, political group documents, rules and statutes, plenary debates and parliamentary questions and press-related output. The project generated an extensive qualitative dataset of 140 interviews with MEPs, their assistants, political group staff and parliamentary staff, conducted over the course of the 2014-2019 and the 2019-2024 terms in Brussels, Strasbourg and online. The interviews cover all political groups although to varying extent (less interviews with radical rights groups opposing gender equality). Most of the interview questions were about the political groups' work, leadership, gender equality and policy-making practices; one interview question directly addressed GM.
For examining GM implementation, we used interviews, combined with EP and political group documents related to GM. Interviews on social and economic policy (the selected policy focuses of the broader research project), and interviews with individuals in charge of GM, addressed the topic of GM in more depth. Given this focus, coverage was most extensive for the FEMM Committee, the Committee on Employment and Social Affairs (EMPL) and the Committee on Economic and Monetary Affairs (ECON). The interviewees included individuals involved in GM (e.g. FEMM members, MEPs and committee staff in charge of GM in committees, and MEPs committed to advancing gender equality) and 'mainstream' actors without a specific interest in gender equality or GM.
Researching without asking directly: data collection with 'flipped expert interviews' We now turn to methodological reflections on interviewing 'mainstream' actors next to gender experts as an innovative strategy for studying GM implementation in the EPor any other organization committed to GM. When designing the research and conducting interviews, we discussed the selection and recruitment of interviewees, questionnaire design and the interview situation. Even when well-thought through, qualitative research designs are ideal types facing limitations in their application (Ackerly and True 2010; Lareau 2022). We discuss both benefits and limitations of our research design and additional challenges we faced. 4 As our accounts of planning each research process demonstrate, reflexivity and considering our own positionality as feminist researchers were core to our research process True 2010, 2018;Brown 2018). Wethe authors and the broader research teamcontinuously reflected on our assumptions about GM implementation, interviewees' potential positionality, or expectable researcher-interviewee interactions; a posture speaking to 'active reflexivity' (Soedirgo and Glas 2020).

Designing the questionnaire
We aimed at simultaneously collecting data on knowledge about the (non)compliance with the EP's formal GM rules, and the informal rules and practices either hindering or facilitating implementation. Regarding formal rules, the EP constantly extended GM provisions (European Parliament 2022). The EP's specialized gender equality body, the FEMM Committee, initiated seven resolutions since 2003, each proposing steps to further institutionalize GM (Ahrens 2019; European Parliament 2022). The resolutions' formal GM rules signal a comprehensive approach covering EP bodies like the standing committees, delegations, and EP administration. For example, the committees, core to the EP's legislative work, are expected to adopt gender action plans and appoint a member to oversee GM. In 2019, GM was integrated into the EP Rules of Procedure, followed by the adoption of an EP gender action plan with concrete actions and targets regarding GM. The political groups 5 , conglomerates of national party delegations negotiating EP's policies, are also important for successful GM implementation. Despite lacking formal EP rules for them, some nevertheless apply GM in their internal work (Elomäki and Ahrens 2022).
Informal practices are difficult to access through direct questions, as interviewees might lack a reflective understanding of everyday norms influencing their behaviour and the working of the institution (Ackerly and True 2018; Mackay 2014; Waylen 2017). We aimed to access informal rules through general questions that prompted narratives describing the daily work in committees and political groups (e.g. 'if you were to advise a new MEP on how to be effective in your political group, what would you stress'; 'how would you describe your everyday involvement with the political group;' 'how do you negotiate a compromise with other groups'). As stipulated by Feminist Institutionalism (FI), informal rules are routine processes not explicated or written down, but 'learned' by the relevant actors to navigate their routines (Kenny 2007;Waylen 2017). This navigation is what we wanted our interviewees to describe, including who has the power to decide what regarding GM.
With GM as a potentially unknown or contentious topic, we considered it crucial to remain receptive to different understandings of GM. Based on earlier studies, we expected to be confronted with different understandings and changing framings over time (Ahrens 2019;Cavaghan 2017). We also expected challenges with 'mainstream' actors. Cavaghan (2017, 82) highlighted the difficulties in securing interviews with these actors and that questions about GM often generated very short conversations, thereby partly challenging the assumption in interpretative approaches that any interview provides useful data (Lareau 2022). Furthermore, studying practices, particularly related to contentious topics, requires a coherent and vigilant reflexive stance of the tools, assumptions and positions in the field by the researcher (Nicolini 2009).
Our challenge, particularly regarding 'mainstream' actors, was thus to design interview questions asking about the topic but do not name it directly. We considered this important to avoid spill-over effects: interviewees talking about GM because they are asked about it, not because of self-motivated narratives on the subject. Such an approach, we expected, would also avoid halo effects: interviewees adopting our GM understanding and answering in a socially accepted way with prefabricated empty phrases. If GM in the EP would be fully institutionalized, GM practices should be detectable in any committee and any MEP and staff member should be able to talk about it.
To this end, researching GM in the context of the EUGenDem-project was useful. The project's general interview questionnaire included several questions about the role of gender equality in the political groups' policies, practices and processes (e.g. 'How would you describe your party group from the point of view of gender equality more generally?'; 'Which equality questions are easy to negotiate within the group?'). The interview questionnaire included one direct question about GM, but to avoid confusing interviewees not familiar with the concept, we did not use the term. Instead, we inquired about the substance of GM ('to what extent is gender equality discussed in relation to key policy issues in your group AND/OR in the XX committee'). We hoped these questions would allow the interviewees to mention GMeither as a positive or negative strategywithout being directly asked about it. Such narratives would allow us to reconstruct the factual implementation of GM in different EP bodies. In case interviewees mentioned GM as a term, we had additional questions ready.

Interviews: selection, recruitment and data collection
Qualitative interviewing in (supranational) political institutions requires reflecting on additional components in preparing the research process. First, we can assume MEPs and partly EP staff are used to speaking in public, being interviewed and defending their positions. Accordingly, our prospective interviewees had probably often encountered questions comparable to those in the questionnaire and were prepared to answer inconvenient or open questions. Nevertheless, as experts (although not necessarily GM experts), interviewees perform a double role as professionals and as individual persons (Abels and Behrens 2005, 175-176). Interviewees (re)produce common narratives, so to say 'shared stories' or 'dominant interpretations' (Yanow and Schwartz-Shea 2013), which are coherently interwoven with certain institutional logics, norms and rules.
Second, interviews are characterized by specific interactions and potential perceptions of the interviewer as either a layperson or an expert, as an accomplice or critic; all attributions may negatively impact the interview flow (Bogner and Menz 2005, 62-66;Brown 2018;Soedirgo and Glas 2020). The research team aimed at pre-empting potential negative effects such as distrust by formulating neutral interview invitations, providing long information sheets and using informed consent forms to assure anonymity and retroactive deletion of interviews if requested.
Third, interviews may (re)produce verbally and non-verbally socio-structural and situational factors such as age, gender, professional status or opinions (Abels and Behrens 2005, 177). Brown (2018, 20) stressed '(…) both race and gender (…) have real-world consequences for the study of politics' and perceptions by interviewees as either insider or outsider shape access, information sharing and also data interpretation. Preparing the interviews and throughout data collection and analysis, the project team continuously reflected about the EP research setting and how the all-female, all-white, relatively young, research team might influence interview situations, including each researchers own positionality (Littig 2005;Sarikakis 2003;Soedirgo and Glas 2020). Although quite seldom, we experienced gender-related interactions, for instance, interviewees lecturing us about the functioning of the decision-making process or content of specific policies, asking senior researchers whether they were doing their PhD's, or being relatively unresponsive towards (young) female researchers. The incidences occurred more often for right-wing and sometimes for 'mainstream' actors, illustrating the necessity to actively reflect and prepare in advance (Soedirgo and Glas 2020).
Starting from the assumption that GM practices should be detectable in any committee and any MEP and staff member should be able to talk about it, requires a broad target population. We aimed for interviews with 'mainstream' MEPs and not solely with MEPs publicly promoting gender equality and GM, gathering data in several committees and not only the FEMM Committee, interviewing across all political groups, and engaging with both political and administrative actors. This strategy was enabled by the EUGen-Dem-project, which focused on political groups and allowed us to cover important EP bodies so far understudied from a gender perspective and often neglected in GM research. Yet, covering all EP bodies formally in charge of GM was impossible, due to the multiple different EP organizational layers, particularly the number of committees, and the different approach needed for interviewing EP administration and staff. 6 Despite our intention to broaden the field of interviewees, the invited 'mainstream' persons nevertheless often forwarded us to 'the gender people' of their political groups, or recommended us to contact members of the FEMM committee. This reaction shows how the long and path-dependent history of GM implementation impacts research strategies. With gender equality actors and FEMM taking the GM lead, other actors misunderstand their own role in the promotion of gender equality. Common understandings of gender equality (or GM) as something attached to gender equality actors make it more difficult to collect data on other actors and bodies. Some invited persons even ridiculed gender equality when rejecting invitations, illustrating the presence of direct opposition to gender equality in the EP. All responses also showed us how potential interviewees read our positionality as researchers working on gender equality and we reflected on how to (re)position ourselves (cf. Soedirgo and Glas 2020).
Some ideas to counteract such problems seemed unethical to us such as withholding information about the research topic (Creswell 2013, 56-60;Gläser and Laudel 2009, 51-56). For instance, drafting interview invitations without mentioning gender equality would probably have worked, but would potentially destroy any trust needed for interviews the moment interviewees realize the importance of gender issues in our research. One strategy we selectively used for known oppositional actors was to refer in invitations to 'equalities' instead of to 'gender equality'. Economic and social policy interviewees were recruited with invitations emphasizing their engagement on the specific policy initiatives we aimed to analyse, downplaying but still mentioning the gender perspective. We also made use of the multi-national research team, addressing potential interviewees in their own language. Yet, recruiting interviewees 'beyond the gender crowd' was tedious and time-consuming. We sent hundreds of emails and reminders and called MEPs offices (calling sometimes helped), but most invitations were nevertheless rejected. Having a researcher on a longer field work stay in Brussels, able to address potential interviewees directly and in person, significantly facilitated recruiting.
Nevertheless, the final dataset contained only a few interviewees directly in charge of gender equality or GM; most were simply members of different committees and political groups or held administrative positions. Thus, we were able to implement our intention of flipping the idea of expert interviews upside down: instead of interviewing people overseeing GM, we included anyone in the EP population. 'Flipped expert interviews' enabled us to assess how diffused GM in the EP is: whether it is institutionalized only in formal rules and practiced by gender equality actors, or whether it is integrated in the daily processes of all actors. Moreover, it allowed us to differentiate between the EP's key power players, its committees and political groups, and their role in advancing or hindering GM implementation (Elomäki and Ahrens 2022).
We suggest 'flipped expert interviews' are a beneficial method and a fruitful angle for studying GM implementation. Had we only interviewed gender equality actors, we would have missed important nuances, such as the perceptions of those who should (but often do not) consider gender perspectives in their daily work. As explained further below, the interview analysis provided us with in-depth insights on how far GM as a strategy and a formal EP obligation has trickled down into the daily work of MEPs and staff. It also allowed us to better assess where formal rules and informal rules clash and what kind of (indirect) opposition occurs. For instance, the analysis of opposition led to distinguishing between opposition to (a) promoting gender equality (for instance, on the side of radical right groups) and (b) changing parliamentary processes and practices (for instance, when ECON rejects responsibility for gender equality).

Tracing GM in data: measuring implementation with interpretative methods
Grounded theory provided the basis for coding and analysing our data (Corbin and Strauss 2008). This is an originally micro-sociological approach enabling to generate metatheoretical explanations grounded in data (Creswell 2013), a 'unified theoretical explanation' (Corbin and Strauss 2008, 107) shaped by the narratives of the interviewees.
When conducting the interviews, we realized our interviewees often tended to focus on policy content from a gender perspective and were less inclined to describe policy processes either impeding or facilitating the integration of gender perspectives. Thus, it was easier to acquire information on which gender issues were discussed in a committee or political group and how, but less on what factors induce these discussions taking place (or not) or taking a particular shapefor instance focusing on descriptive representation instead of topical issues.
Moreover, interviewees tended to provide a positive self-representation and simultaneously criticized other political groups or committees for being non-committal (or for asking for 'too much', if the interviewee belonged to the TAN spectrum or was not interested in gender equality). Sometimes, the general questions often sparked spontaneous rejections of gender equality as a relevant topic for specific policy fields and reflections of the difficulties to integrate gender perspectives in the political groups' work. The various reactions allowed us to estimate whether the interviewees were familiar with the formal EP rules on GM or saw it as part of their daily work by mentioning it themselves. Furthermore, our interview analysis was informed by three theoretical approaches: Social Constructivism, Feminist Institutionalism and Micro-Politics. Simultaneously capturing actions (micropolitical approach), their institutional context (FI) and the how actors construct meanings (social constructivism), allowed us to generate a 'thick description', analyse a single case and study policy processes from a feminist, interpretivist perspective (Vromen 2010, 249).
Social constructivism scholars engage with the way gender is constructed in discourses and practices leading to prioritizing certain political solutions over others (Kantola and Lombardo 2017), in the EP, for instance, by barring critical gender perspectives (Ahrens, Gaweda, and Kantola 2021). FI provides analytical insights for explaining the gendered formal and informal foundations of political institutions, gendered mechanisms of continuity and change, the impact of gendered actors and for revealing the 'hidden' challenges to institutional change towards gender equality (Kenny 2007;Krook and Mackay 2011;Waylen 2017). Here, we analysed gaps between formal rules for GM, how they were taken up in practice by different EP bodies, and additional informal rules either supporting or countering the formal rules. Micropolitical approaches, finally, account for the multiple, day-to-day steps taken to shift the power balance between institutional actors (cf. Wiesner 2019). We used these to theorize how (gendered) actors pushed the limits of formal rules in favour or rejection of GM and whether gender equality actors created new 'rules in use' which successively were affirmed as 'rules in form'.
All interviews were recorded, transcribed, anonymized, and then coded in an iterative process by the team members using the qualitative analysis tool AtlasTi while interviewing continued; an approach labelled 'dynamic research process' (Ackerly and True 2010, 42). In several meetings, the team developed deductive and inductive codes which covered both topical and theoretical issues. The team coded all interviews on a rolling, circulating basis with each team member participating equally, which also helped to sensitize for different previous situated knowledge and positionality of team members (Ackerly and True 2018, 261;Lareau 2022, 27-29, 151). Generally, codes were used as extensively as possible to secure full coverage of important empirical material. They were saturated and validated by adding further details from every interview until no new information was found.
Core to analyse GM implementation was the code 'GM'. We coded text segments when interviewees (1) used the term explicitly for either policies or practices or (2) discussed GM implicitly. Such implicit references included, for instance, descriptions of making amendments related to gender equality on committee draft reports, raising gender perspectives in committee debates or opposition to the integration of gender perspectives in a specific policy field, and comments on the relevance of gender to a specific policy issue. Deciphering the specific GM-related differences in the narratives, often meant including additional material from the content analysis (statutes, rules, websites), and extracting the information of actual practices, was particularly timeconsuming.
After contrasting with additional material, we re-coded 7 the coded text segments with new codes developed based on our theoretical approach. These covered, among other things, understandings of GM (e.g. descriptive representation), perceived difficulties (e.g. lack of knowledge, patriarchal norms), forms of resistance (e.g. denying the relevance, seeing GM as 'gender ideology'), strategies to push GM forward, and the different actors involved (e.g. FEMM members, other committees, political groups). We then contrasted the narratives to find similarities, differences and patterns thereby opening the possibility to carve out meta-theoretical arguments used by interviewees illuminating underlying formal and institutional constraints.
Through the iterative process, we also tied the coded material back to our three-fold theoretical approach. From a social constructivist perspective, we established that interview quotations cannot be treated as descriptions of an 'objective' reality but as perceptions by and narratives of the interviewees. Thus, we treated them as reality constructions that allowed us to examine which kinds of institutionalized settings, understandings of GM and power relations were (re)produced and by which actors. Some narratives were more dominant than others, but all shaped the way GM was understood and implemented within the EP (Lombardo, Meier, and Verloo 2009). For instance, we discovered common conceptual misunderstandings of GM among the interviewees. These included seeing it solely as descriptive representation (number of women/men represented in a committee etc.), instead of as a tool for substantive representation (promoting gender equality in policy fields), or seeing it as a responsibility of those in charge of gender equality, instead of the responsibility of everyone (Elomäki and Ahrens 2022).
For FI, we specifically compared formal rules as laid down official EP documents with informal rules and practices of GM implementationor often rather lack thereof. Our findings showed considerable differences along committees, political groups andunsurprisingly -'closeness to gender equality policy'. Importantly, many interviewees did not directly describe certain practices, we had to deduct these from the narratives by piercing different bits and pieces together for a comprehensive overview. Traces and difficulties of GM implementation occurred throughout the interviews. GM faced problems of 'nested newness' (Mackay 2014), where actors perceive the new rules as contradicting their previously learned usual (informal) rules and practices, which then leads to 'forgetting' the new GM rules. We found committees and political groups with strongly institutionalized patriarchal gender norms were more likely to oppose GM because 'they have 'learnt the script' of informal unequal gender rules of behavior' (Mergaert and Lombardo 2014, 15).
Finally, we applied the micropolitical approach to examine the power relations among committees, political groups and individuals, and how the interviewees and EP bodies strategically used, reinterpreted or changed formal GM rules in daily practices. The FEMM committee appeared quite creative: it proposed own-initiative reports on gender aspects in different policy fields, thereby overstepping formal and informal committee boundaries and forcing other committees to get involved. For instance, FEMM proposed a report on gender and taxation and thus intervened into the portfolio of the powerful ECON Committee, which first tried to block the report altogether but finally opted to co-author it. This was the first time ECON engaged with topical gender equality issues. Simultaneously, FEMM's possibilities to integrate gender perspectives are constrained by its less prestigious status, which invites other committees to the micropolitical strategy to undermine the formal GM rules by disregarding FEMM's opinions and amendments.

Conclusion
In this article, we discussed interpretative methods as particularly well-suited approaches to researching norm-and value-driven topics, such as GM implementation in the EP. Building on the broader literature on interpretative methods and previous research on GM implementation, we interrogated 'flipped expert interviews' regarding their advantages and limitations in capturing formal and informal practices and processes in the EP's committees and political groups and the power relations affecting actors.
Previous research on GM focused mainly on actors in charge of the strategy, on the content of gender equality policy or on anti-gender mobilizations. Using the EP as a case study, we highlighted that interviewing 'mainstream' MEPs, political staff and EP administration, in other words, actors not formally tasked with gender equality or even opposing it, allows for nuancing our knowledge about GM implementation and its challenges. Going beyond the 'usual suspects' like FEMM committee members, provided us with in-depth insights about trickle-down effects of GM (or the lack thereof) on the daily work of MEPs and staff. Comparing the contrasting interview narratives furthermore allowed us to better assess where formal rules and informal rules clash and what kind of (indirect) opposition can occur.
Unquestionably, our approach has limitations and is arduous and time-consuming, because the newly focused potential interviewees were often difficult to recruit. Yet, overall, we find that employing interpretative methods to extend the potential interview population to 'non-experts' helps painting a broader picture of contextual factors and micro-dynamics that illuminate processes of politicization around contentious topics. Notes 1. Given our case, we use the EP GM definition although many other definitions exist. 2. The Ethics Committee of the University of Tamperegranted ethics approval for the project. 3. For parliamentary ethnography and the EP see Miller (2022). 4. We suspend discussions on how the pandemic impacted data collection, because most interviews were conducted before the EP closed in spring 2020. 5. Currently: European People's Party (EPP); Progressive Alliance of Socialists and Democrats (S&D); Renew Europe; Identity and Democracy (ID), Greens/European Free Alliance (Greens/ EFA); European Conservatives and Reformists (ECR); and the Left group (GUE/NGL). 6. For instance, we did not interview EP staff in charge of human resource management. 7. This was also necessary, becausedespite the team's reflective and sensitizing research processnot all members were fully knowledgeable about GM.