Data coding and analysis

Org 1 Org 2 Org 3 Org 4 Org 5 In total Events in the homes:

’Start-up meeting’ 2 2 2 2 3 11

’Training’ 1 1 2

’Evaluation meeting’ 1 1

Events at the office:

Occupational team/office meetings

2 3 2 1 2 10

‘Other types of events’

3 3

Total number of observed ’events’

26

Table 4. Overview over the observations (reduced to ‘events’).

grouped the notes from the five organizations together as well as sub-headlines that indicated the type of event observed (e.g., team meeting, start-up meeting, evaluation meeting and informal talk). This phase of the analytical process served two purposes. First, it gave me an opportunity to re-familiarize myself with the data I had collected as well as the data collected by the other ReKoHver project members. For example, I found it helpful to listen to the recordings of those focus groups I had not conducted myself because doing so allowed me to hear the tones and dynamics of the voices, which are not always easily captured in written form. Second, this process allowed me to develop a tool I could use to manage my data. The database gave me an overview of the data and made it easier to analyse it without losing track of the organizational origins or the potentially important aspects of inter-organizational differences.

In the second phase, I systematically performed what grounded theory labels “open coding” (Strauss and Corbin, 1990, 1998). I started out by re-reading the transcripts of the interviews and focus-group discussions, and coding them line by line in the Nvivo database. Initially, I expected to code the observations in a similar way in the Nvivo database. However, I decided not to do so because the dynamics of this data seemed to get lost when I started to code the data in the Nvivo system. Therefore, in order to maintain the dynamic and processual insights, I read through all of the field notes and coded them using the comment tool in Word. While undertaking the open coding of the transcripts and field notes, I focused on what was being said about rehabilitative home care work in terms of for instance practices, tasks, outcomes and actors, as well as their functions and the relations among them. I also listened for details on what was being done by whom, how, where and with what implications (e.g., tension, happiness). Overall, this initial coding gave rise to 45 empirically grounded codes, such as

“success stories”, “nurses”, “manager”, “maid”, “women”, “experts”, “expertise”, “optimize professionalism”, “a good life”, “resources”, “self-reliance”, “decline”, “passive”, “sadness”,

“relatives”, “hands on the back”, “a care gene”, “nurturing”, “training”, “withdraw”, “distance”,

“traditional”, “new”, “tension”, barriers”, “labour division”, “pride”, “cost-effectiveness” and

“improvements”. Although I found this empirically grounded open coding intriguing and encouraging, it was also frustrating in the sense that the material and the high number of codes seemed overwhelming.

In the third phase of the analytical process, I began to group codes for similar content into aggregated categories (Strauss and Corbin, 1990, 1998). Some codes could be grouped under aggregated empirical categories such as “recipients and their (aging) bodies” (e.g., “resource”, “decline”), “professional roles and their (embodied) characteristics” (e.g., “nurse”, “manager”, “maid”, “women”, “expertise”,

“professionalism”), “professionals’ practices and tasks” (e.g., “training”, “hands on the back”,

“nurture”, “tools”, “hands on”), “rehabilitation initiatives” (e.g., “teamwork”), “outcomes of rehabilitation” (e.g., “success stories”, “cost effectiveness”, “improvements”, “barriers”, “failure”),

“relational dynamics” (e.g., “tension in the home”, “tension among colleagues”, “division of labour”,

“recognition/rewards among colleagues”), “emotional expressions” (e.g., “pride”, “anger”) , “new versus traditional”, and “nurturing versus rehabilitation”. Some of these categories overlapped. For instance, “new versus traditional” and “nurturing versus rehabilitation” cut across most of the categories. Therefore, some pieces of text were assigned multiple codes.²⁸.

In the fourth phase, I used Nvivo to gather all transcript-based pieces of text with similar codes together under the affiliated aggregated categories. I did so in lengthy Word documents in which I also manually added the extracts from the observation transcripts with similar codes. At this point in the process, I started to become aware of the multiple, ambiguous ways that rehabilitation influenced the workers and their work. It was clear that there was some tension that was masked in a way that made it difficult for me to understand and conceptualize. For example, in the “professional roles and their (embodied) characteristics” category, multiple signifiers of the workers’ roles coexisted (e.g., “maid”,

“women”, “professionalism” and “expert”). However, these signifiers did not coexist peacefully – rather, they emerged in conflicting ways. For instance, “maids” and “women” were contrasted with

“professionals” and “experts”. At the same time, these tensions signifiers seemed to be ordered around some kind of boundary or distinction. For example, it was clear from the text that “maids” and

“women” were continually associated with each other and shared the same illegitimate connotations, while they were contrasted with “professionals” and “experts”, which were two signifiers that seemed to share legitimate connotations. Similarly, the codes in the “professionals’ practices and tasks”

category revealed that “training” and “hands on the back” were contrasted with “nurture” and “hands on”. This (dis)order appeared to emerge around the “new versus traditional” and “nurturing versus

28 For example, if a piece of text described professional roles and their practices, it would be assigned multiple codes.

rehabilitation” categories. Nevertheless, this insight taught me little about what was at stake in this (dis)ordering. At this point, I turned to critical litterature, and started to raise new questions.

Phases three and four were, to some extent, influenced by the review of the literature that I undertook during the analytical process. In order to pursue my curiosity and focus on (dis)ordering, in the fifth phase I started to move systematically between the aggregated empirical categories and the literature that resonated with those categories. More specifically, I started to experiment with applying different theoretical building blocks that might explain the (dis)order within each aggregated category. For instance, when trying to analyse the recurring category of “professional roles and their (embodied) characteristics”, I found it increasingly helpful to engage with concepts and literature that described professionals’ identities, the regulation of such identities through discourses of differences and the outcomes of such regulation attempts (e.g., Du Gay, 2008, Davies and Thomas, 2002, Alvesson and Willmott, 2002). In addition, as descriptions of workers’ practices and their relations with recipients were plentiful, I found it natural to engage with concepts and literature that focused on care workers’

relational and embodied work practices in order to understand why, for example, “hands-on” practices were contrasted with training practices in this context (e.g., Twigg et al., 2011; Sullivan, 2007).

Moreover, given the fact that actors continually talked about “success stories” in relation to their rehabilitative work in ways that seemed to contradict other stories, I found it increasingly informative to engage with literature that could help me theorize about the implications of professionals’ sharing such stories and norms in each other’s company (e.g., at team meetings) (Barker, 1993, Orr, 1998).

Through this process, I realized that my data called for different theoretical building blocks and toolkits and, as such, a multi-paradigmatic framework that could cover such core concepts as identity, regulation, power dynamics, gender, distancing techniques, narratives/war stories, group norms, embodiment and performativity. It was also in this phase that my critical, feminist-inspired stance was formed. The fifth phase was an abductive analytical phase (Bryman and Bell, 2011) in the sense that while I used the literature that resonated with the empirical categories, I used it and literature on the history of home care organizations in order to reconsider and specify the empirical categories, the core problem and the overall research question(s).

Thus, my intensified readings in the fifth phase helped me understand that the reoccurring contrasts, tensions, complexities and ambiguities revealed by my categories could not be reduced to “traditional

versus new” or “rehabilitation versus nurturing”. Rather, they concerned the coexistence of multiple overlapping and conflicting ways to both discursively make up and more materially mark up what professional home care work involved. For instance, the hands-on practices were rhetorically made illegitimate not only by discursively associating the practices with (female) over-nurturing but also because this work was usually carried out by low-status care aids. As such, I began to understand that the contrasts, differences and divisions, which I labelled the “boundaries of professionalisation”, were important in this context because they were not neutral, even if they seemed to be. Instead, they represented a professionalization mechanism that constituted professionalism, and that had ambiguous implications for the workers and their risk of (re-)marginalization. However, despite the profound dominance and effects of this mechanism, it was still contested. In addition, I started to understand that these boundaries of professionalisation could not be reduced to home care work, as they expressed more generic but historically and socially constructed contrasts that, in the name of professionalisation, seemed to differentiate and draw boundaries between the included and the excluded, the legitimate and the illegitimate, the valorised and the non-valorised, and the professional and the non-professional in organizational settings in the west, especially in female-dominated workplaces on the margins.

Through this new problem identification and understanding, I was able to begin qualifying my key analytical focus. In the final analytical process, I re-coded the data using a core, guiding analytical approach in which I systematically focused on how matters of professionalization and marginalization intersected at work, how boundaries were established to legitimize and privilege certain elements (e.g., actors, technologies and structures) and marginalize other(s), and the implications of this boundary drawing for the work itself and for workers’ risk of marginalization. In other words, I began to re-analyse the empirical categories by raising new questions about how the boundaries of professionalisation were constituted and how they functioned in rehabilitative home care workplaces – by whom, how and with what implications for the work and the workers. As mentioned in Chapter 1, I did not explore the boundaries of professionalisation in a pre-determined, fixed, binary sense. Rather, I sought to explore the polyvocal and ambiguous aspects of the boundaries that were drawn between what was (un)acceptable to say and do as home care professionals in multiple sites and multiple situations, as well as the implications of this saying and doing, and of the locations where these activities took place for the work and workers. Consequently, I focused on the boundaries that the

actors voiced and interpreted on their own, and on the less obvious boundaries that were implicitly embedded in the practices, face-to-face interactions and tools at different sites and situations.

Ultimately, my focus was the co-emergence of both the discursive-regulative and the more material-practice-oriented aspects of the boundary drawing. As I discuss in the next section, the development of this re-coding system was closely affiliated with the three articles that comprise the analytical body of the dissertation.