In order to address our research questions, we employed a qualitative approach, combining observations, interviewing and documentary analysis in order to get closer to everyday CEO practices of mobilisation. A similar qualitative approach was taken successfully by previous NIHR HS&DR programme projects (e.g. Swan et al.,8 Schneider et al.,97 Pope et al.98) and is an accepted mode of study across the social sciences. In this section, we introduce our approach, reflecting on the different stages, the access and sampling strategy, data collection, analysis, and anonymity and confidentiality as a particularly critical aspect.
Research stages
The study consisted of five consecutive stages that informed each other, as depicted in . Notably, as is common with fieldwork-based studies, the stages themselves changed slightly from the original plan we envisaged at the start of the project. The stages also often overlapped in practice, which is again common in qualitative studies. For instance, because the study featured seven CEOs, who were studied consecutively, this meant that access and ethical approval were spread out over stages 0 and 1, given that access could be secured only approximately 2 months in advance, and that local research and development (R&D) approval had to be obtained. Analysis was also done throughout; not only in stages 2 and 3, but during periods of shadowing (stage 1).
Research stages for the KMobilis project.
Stage 0 was spent setting up the groundwork for study. This involved obtaining ethical approval from the local Research Ethics Committee, revisiting relevant literature (particularly on shadowing and executive work) and putting our access and sampling strategies in motion. In addition, we agreed that, as shadowing was not a method the research assistant (RA) and the principal investigator (PI) had previously employed, we would seek to do a week-long pilot study. Part of stage 0 was therefore spent seeking access to suitable local public sector executives, whom we might engage for this pilot. Finally, we worked to identify members of our scientific advisory panel, as a key support mechanism.
Having secured ethical clearance in March 2011 and broadly identified our next steps regarding access, we began stage 1 by firstly securing access to a university head of department and an NHS trust director, for the RA and the PI to shadow for several days, in order to gain familiarity with the method. These pilots were completed in early June 2011 and led to a revision and refinement of the methodology as discussed below. In addition, at the start of stage 1 we identified a member of the advisory panel whom the RA had previously engaged to secure access to public sector organisations, and asked him for assistance. This proved to be a very successful strategy, as we outline below, which meant we were able to commence our research stays with CEO1 (shadowed by PI) and CEO2 (shadowed by RA) in June 2011. Encouraged by our success in recruiting the first two CEO participants, during this stage we consulted with the members of the advisory panel with regard to whether to continue to focus on NHS CEOs, as opposed to ‘executive directors’ (a broader category) as our sample, which we had originally conceived in our project proposal. Given that this represented a rare opportunity to study in depth a very important group of NHS managers, the advisory panel agreed with this proposed change.
Following the completion of the first two case studies in autumn 2011, we began an initial round of in-depth analysis in preparation for our first advisory panel meeting on 22 September 2011. The aim was to share early empirical reflections and receive feedback, but also guidance on how we should progress with the next round of access and fieldwork. During the advisory panel meeting, we discussed at length the issue of sampling, and confirmed our decision to focus exclusively on NHS CEOs, as from our initial literature review it appeared that a clear gap in the literature existed on the topic. Feedback on the emerging findings was also received following a presentation to a group of CEOs at two Service Delivery and Organisation (SDO) Network CEO Forum meetings in January and February 2012. The remaining five case studies were conducted by the RA (four) and PI (one) in the following 12 months, following successful recruitment. This was an iterative process of access and fieldwork, whereby access and R&D clearance for one site were completed during fieldwork at another.
In summer 2012, we entered a dedicated stage of analysis (an early start to stage 2), in preparation for our second advisory panel meeting in November 2012. At this meeting, we presented emerging analytical categories across seven cases, as well as detailed examples. We received valuable feedback that helped us to refine the key analytical conclusions, and the structure for the upcoming individual CEO feedback sessions. Following the advisory panel meeting, we turned to setting up the feedback sessions, which because of the busy diaries of the CEOs and their coaches, whom we sought to have present where possible, took some months. In the meantime, we continued our analysis of the great amount of data gathered. The feedback from the advisory panel and the CEO sessions in turn subsequently informed our ‘focus group’ presentations, which were organised in conjunction with the Health Services Research Network CEO Forum, and were approached as a further feedback, engagement and practitioner validation exercise. The two very successful meetings took place on 9 April 2013 in London and 24 April 2013 in Manchester, with a number of NHS CEOs in attendance.
As we continued our analysis, stage 2 and stage 3 effectively overlapped, although, following the two focus group events, we continued to refine our emerging findings and possible implications in preparation for our final event as the pinnacle of stage 4. The half-day research seminar took place on May 24 2013, with some 40 practitioners, academics and policy-makers in attendance. The feedback was very positive, with notable interest in the possible practical implications from organisations such as the NHS Leadership Academy. These different strands of engagement and feedback all in turn informed this final report.
How the methodology and research design evolved during the study
Like other qualitative studies, ours developed from the original proposal. This was because of the changes taking place in the NHS at the time, the results of the pilot study conducted early on in the process, and the opportunities that emerged while the project unfolded. In particular:
As the major reorganisation of the NHS was occurring at the time we were seeking access, which included most notably the abolition of primary care trusts (PCTs), it was decided (in conjunction with our scientific advisory panel) to replace PCTs originally identified as part of our sample with mental health trusts. It was also decided that the sample should aim (as much as possible) to reflect the ratio between the total number of acute and mental health trusts in England. In the end, our sample was five acute trusts and two mental health trusts.
Early in the study we decided, in conjunction with our scientific advisory panel, that a longer period of shadowing was necessary, as this would allow observation of knowledge mobilisation in practice over time and thus provide additional empirical nuance largely missing in the existing literature. In addition, the literature revealed that managers were almost never studied longitudinally (i.e. longer than 1 week per manager). Therefore 5 weeks was agreed as the target time period of fieldwork on each site. As this represented a more considerable empirical engagement, and following the practical preferences of the CEOs who preferred observations to be conducted as close together in time as possible, this meant there was no second cycle of shadowing following ‘initial feedback and validation’ (i.e. stage 3 of the original proposal). Instead, all the shadowing was completed prior to stage 3.
Following the results of our pilot study, and in view of the increased intensity of observations in situ (e.g. 5 weeks per CEO), we also decided against using some of the methods we had originally planned to. Specifically, the pilot study suggested that some of the methods mentioned in the proposal would not add sufficient value, and at the same time could also potentially jeopardise our recruitment and access efforts. Many of the executives we approached told us that they would give us access provided the study would not add to their existing heavy workload – or that of their top management team (TMT). In particular, the pilot study clearly demonstrated that diaries and logs would certainly represent an additional time strain on the subjects, and probably only duplicate information collected via shadowing, as well as provide an impoverished version of the CEOs’ work. Accordingly, we decide not to use these methods, and instead gained this type of information via informal consultation with personal assistants (PAs) and the CEOs’ informal reflections. Moreover, all CEOs gave us access to their weekly diaries for the months prior to our stays, so that we were able to collect the same data using less intrusive sources. We also decided not to use ‘interview to the double’. Interview to the double is a projective technique which requires interviewees to imagine that they have a double who will stand in for them at work on the next day. The interviewee-instructor is then asked to provide the necessary detailed instructions which will ensure that the interviewer-double is not unmasked. The approach helps practitioners to ‘observe’ and describe their own practice. This decision was made largely because, at the beginning of the project, we recorded a strongly negative reaction against it by one of the pilot study informants. The informant saw it as odd that she was being asked in the ‘interview to the double’ to describe the very practices we had just observed as if we had not. This method was therefore excluded. Finally, we did not formally and systematically interview the CEOs’ executive teams, again in order to avoid overburdening the already time-invested CEOs and their organisations. Based on the results of the study, we believe that our choices were justified, as any loss in data derived from not using these additional methods was largely offset by the unique nature of our sample, and the duration and depth of our shadowing-based observations.
As briefly noted above, after the first round of recruitment, and given our success in enrolling CEOs in the study, in consultation with the advisory panel we shifted the analytical focus from executive-level NHS managers to NHS trust CEOs. Firstly, our focus in the original proposal on ‘NHS directors’ rather than on CEOs was partially because of the perceived risk related to recruitment. This concern was echoed, for instance, by one of the reviewers at the outline stage of the proposal in February 2010, who noted:
my main concern is the potential difficulties likely to be encountered in recruiting the senior management participants given the investment in time and energy during a period when pressures within healthcare organisations are set to rise considerably in the economic context . . .
The early success in recruiting CEO participants suggested that this risk could be effectively addressed in practice. Moreover, our review confirmed that the existing literature was almost totally silent on the work of NHS CEOs and their practices of knowledge mobilisation. By refocusing the study, we could therefore provide valuable insights into a largely unexplored topic and social group, built on a data set as robust as, if not more so than, many of the existing studies which followed the semistructured observation-based work of Mintzberg.
88Finally, accessing NHS CEOs over a longer time period proved to be particularly challenging and in fact, as a result of circumstances beyond our control, our final CEO had to withdraw from the study only days before the start of fieldwork. This had a knock-on effect on the duration of the fieldwork (stage 1), which in turn required a 3-month extension of the project as a whole.
Access and sampling strategy
Sampling strategy
The choice of cases was based on a theoretical (rather than statistical) sampling strategy. Our aim was to maximise diversity in the sample for analytical purposes (statistical relevance was impossible given the low number of cases). Our original sampling strategy, following a shift in analytical focus, aimed at including six to eight CEOs across mental health, acute and primary care sectors. However, as noted above, early in the project we made a decision with the advisory panel that we would focus on acute and mental health trusts only, as PCTs were being disbanded and subsequent organisational structures were in flux. In order to ensure diversity in the sample, we aimed to engage trusts that varied in size, physical location, the relevant Strategic Health Authority (SHA), financial health (as determined via relevant regulator performance ratings) and organisational type [i.e. a foundation trust (FT) or a non-foundation trust]. We also considered the personal characteristics of the CEOs, for instance gender, professional background, time in post and if this was their first CEO role.
Access to participants
Access was obtained through a combination of personal contacts and direct requests to CEOs who were deemed potentially interested, for example because they had previously participated in developmental events, such as the SDO Network CEO meetings. The collaboration of the NHS Confederation and the support of one member of the advisory panel were critical, especially in the early stages of recruitment. A letter detailing the research questions, possible contribution and distinct ‘benefits’ for the CEOs from the study was initially sent to five acute CEOs, who the panel member felt might be interested. We subsequently set up meetings with four of them and were successful in gaining access to two, who became the first two CEOs in our sample. The possibility of comparative feedback, but also the opportunity to discuss the findings in a coaching session at the end of the project appeared a key factor in the CEOs’ favourable decisions. We used the same process, facilitated by the members of the advisory panel, to recruit the rest of the cases.
Study participants
Our final sample for the study was seven NHS trust CEOs. The sample size reflected time constraints of the study, and our strategy regarding diversity in the sample. A summary of key characteristics of our sample can be seen in .
Key characteristics of the study sample. MH, mental health.
The sample includes an almost even ratio of men to women as intended in our sampling strategy. Equally, we were successful in meeting other key aspects of it, such as diverse professional backgrounds: management (two from the NHS; one from the private sector), nursing (two) and medical (one). In addition, the sample included CEOs who managed both successful and less successful trusts (one financially struggling acute non-FT; two financially struggling acute FTs; two successful acute FTs; two successful mental health FT). These assessments were based on external validations of financial and operational success by relevant regulator bodies, such as Monitor performance indicators. They are also reported broadly enough so that anonymity is not compromised. Finally, the CEOs were at different junctures in their posts and careers. For instance, two CEOs were shadowed within a year of starting their current post, while three had been in their present post for longer than 5 years. Similarly, we observed those with previous experience at CEO level (n = 4), and those for whom the present post was their first CEO appointment (n = 3, though all had been in post more than 2 years when studied, including one CEO whose current post followed a merger).
Data collection
After securing access to the seven sites, and completing the necessary R&D approval for each, we employed three distinct methods in order to complete our investigations, which are all recognised parts of a naturalistic approach to research.
Observations and shadowing
The key method we used was shadowing, which involved closely following each CEO as she went about her working day, and observing as much of it as possible. This enabled us to understand the various nuances of CEOs’ practice as it actually happened. Practically, this meant we would normally speak to the PA a week before, to go over that week’s diary and highlight any meetings that were not appropriate to observe. Generally speaking, we had great success in terms of access, in that there were very few parts of the CEO’s day, particularly on site, that we could not observe. These were limited to some one-to-one supervisory meetings with more junior colleagues, human resources (HR)-related meetings concerning individuals, and private meetings (for instance with their pension advisors). We also mostly did not observe evening or day events of a particularly sensitive nature, or those that required extensive travel, as well as more informal chats the CEOs had with individuals, when it was made clear to us that this would not be appropriate. Finally, we could not observe the work the CEOs did at home, which was one of the notable limitations of shadowing as a method. This shortcoming was addressed by asking the CEOs to recount the work they did at home the following day, and accessing their e-mail correspondence wherever possible.
An average day of shadowing began with our arrival at the start of that CEO’s working day. We usually had a designated seat at each site while the CEO was in her office, normally behind or near the CEO’s desk, so that we could observe as she worked on her computer. We used an iPad, which allowed us to take immediate electronic field notes on the go, but also to look up any information online to understand better what or who was being discussed, which improved their quality. As the CEO went to her meetings, we followed along, chatting informally or staying quiet as appropriate. In all instances, the executive teams and closest collaborators were aware of the study prior to the start of fieldwork. As we entered each meeting, the CEOs would briefly introduce us and ask us to give a short overview, and then carry on with her task. We normally found a seat out of the line of sight, away from the table, and observed quietly. Normally, CEOs and their teams got used to our presence after a couple of days, so we were generally not introduced during later meetings. Similarly, if a patient or staff member stopped to chat with the CEO in the corridor, we either stayed while they finished, or left the CEO and returned to the office, if deemed more appropriate.
For practical reasons, we often spent 1–2 weeks with one CEO, then moved to another, then switched back. This allowed us to track issues over a longer period, and indeed observe how work changed over time. It was also a practical solution, as we found that CEOs found a stay longer than 2 weeks very demanding. This also avoided the researchers becoming too accustomed to the site to the extent that they no longer had a ‘fresh pair of eyes’ on it.
In total, we completed the following periods of intense shadowing (in consecutive order from when we first entered the organisation):
CEO1: completed in December 2011 (total 6.5 weeks)
CEO2: completed in November 2011 (total 5 weeks)
CEO3: completed in February 2012 (total 5 weeks)
CEO4: completed in June 2012 (total 5 weeks)
CEO5: completed in May 2012 (total 5 weeks)
CEO6: completed in June 2012 (total 4.5 weeks)
CEO7: completed in September 2012 (total 3.5 weeks).
Notably, we mostly met our very challenging target of 5 weeks of shadowing per CEO, making this one of the most longitudinal, in-depth studies of senior executives ever completed. The exception was the final CEO, where because of pressing time restrictions of our project only 3.5 weeks of shadowing could be done. This was nevertheless an empirically rich period, and sufficiently long that we could observe some issues over time, thus meeting our analytical requirements.
Interviews
In addition to shadowing, we used formal interviews in some cases to supplement our observations. They allowed the CEOs to reflect in a structured way on their knowledge mobilisation and working practices, and to offer reflections on why we might have observed what we did. These were very useful to us in terms of clarifying or refining existing insights.
In particular, we conducted five formal, semistructured interviews during the study (see Appendix 1 for the interview schedule). These were conducted in the final week of observations, or some time following shadowing, depending on the CEO’s diary. They lasted between 38 and 65 minutes, and were recorded and transcribed verbatim. The two remaining CEOs were interviewed informally, as part of daily observations. In their case, the decision was made that formal interviews would not be appropriate, given that both CEOs shared their reflections extensively during on-site visits, and their diaries made scheduling a formal interview exceedingly difficult. These ethnographic interviews were also transcribed verbatim and used in the analysis. In addition, we conducted two formal interviews with two different PAs, which were recorded and lasted approximately half an hour each. Such interviews were not conducted in other settings because it was decided that, owing to pressing work commitments, these would be very difficult to arrange. Instead, as was the case also with the two PAs formally interviewed, the other PAs were frequently engaged informally, via ethnographic interviewing in situ.
Documents
A final key source of data were documents obtained during periods of observation, including meeting papers, articles referenced by the CEOs, and copies of publications consulted. Given the nature of each CEO’s work, particularly the number of large formal meetings attended, but also the level of comfort by each CEO with regard to sharing internal documents, the number of documents gathered for each site varied greatly, from thousands of pages to approximately 100. Such internal documents were supplemented by externally available information. This included, for instance, annual trust reports, news articles and regulator documents. We accessed these prior to our stay and throughout the study. This allowed us to gain a greater familiarity with the trust and the CEO before starting, but also to keep abreast of any major changes since our departure.
Analysis
The analysis was carried out as a reiterative continuous process which proceeded in parallel to the study from the start of stage 1 onwards. Each of the two empirical researchers regularly read and reflected on the individual sets of field notes, including before each return to the field, and wrote analytical notes as part of the research diary for each CEO. These were eventually prepared into extended analytical memos, which synthesised the analysis of emerging insights from each case, and enabled cross-case comparative analysis. In addition, the researchers spoke regularly while in the field, sharing notable impressions as they happened. The process was facilitated formally by regular team meetings between all three team members, presentations in academic and practitioner fora (see Appendix 2), and regular meetings with the advisory panel. Given the sheer amount of field data collected, the feedback and suggestions from the advisory panel were particularly important in helping us to identify what analytical insights appeared most compelling or should be explored further. Following the discussion with the advisory panel, we continued to refine our list of emerging first- and second-order analytical categories, by continually rereading the field diaries and comparing emerging insights as we perceived them. These were similarly informed by our continued reading of the broader knowledge mobilisation and information work literatures, as well as that on managerial work.
Importantly, we also continued to be in touch with the CEOs studied following each research stay, sharing regular updates to keep them abreast of the study’s development. These conversations also allowed us to gain insights into how specific issues had progressed or the CEO’s focus of work changed since our departure. For some CEOs, it was also an opportunity to share their thoughts on the emerging findings, how we could best engage their colleagues, and what would be the most useful research outputs.
Feedback
A critical component of our research design was the individual feedback session with each CEO. These were envisaged as an opportunity to share in-depth comparative insights, and receive CEO feedback on its accuracy and helpfulness as part of a targeted validation exercise. The sessions were a great success, in that the CEOs overwhelmingly found it useful, and an accurate depiction of their work. Their detailed feedback in terms of clarifying certain points, or reflecting on how they and the trust had progressed since, provided further richness and accuracy to our material. In addition, the sessions were designed to work as a two-part development opportunity, where we would present emerging findings, and then a personal coach would work with the CEO to draw out relevant implications for their practice. Though we could not accomplish this in every instance, as some CEOs preferred to receive feedback without the coach, in three cases it was a very valuable and positive process, with seemingly great potential as a mode of facilitated learning and personal development for executives. We took careful notes of the three cases and will be reflecting on them with a view to developing a model for research engagement, for testing in future studies.
Anonymity and confidentiality
The highly sensitive work of the CEOs, the practicalities of shadowing as a method, and our assurances as part of the access agreement all meant that issues of anonymity and confidentiality were particularly acute. We met these challenges via several research strategies. For example, we took great care to protect the anonymity of the CEOs themselves, by never discussing our observations outside the project team, by writing fully anonymised notes, and by making a professional judgement regarding what was private and what was public. In particular, although we occasionally observed the CEOs during acutely sensitive conversations, we made sure not to take written notes, unless this was seen as absolutely necessary in giving an accurate depiction of the CEO’s working realities.
While in the field we employed standard principles of ethnographic (observation-based) practice to strike a practical compromise between retaining the anonymity of our CEOs as the chief participants and the ethical requirements which necessitated making every individual that we encountered informed. This was addressed chiefly by obtaining written consent prior to an observed encounter. When this was not possible or not appropriate, we made sure not to take any notes relating to that individual directly, but recorded afterwards that an encounter with an anonymised individual took place, and the CEO’s role in this engagement if appropriate. Importantly, the CEOs themselves occasionally asked us not to introduce ourselves before a meeting, as this might affect the tone of the meeting in question. Where such a request was made, we abided by it, as our method depended to a large extent on being as unobtrusive to the CEOs as possible. In general, explanations about our study were kept to a minimum and employed only when necessary. Again, this was essential in the light of our small sample, which meant that, if an individual knew we were studying the CEO on site, they could more easily deduce which of the CEOs we described in this report was the one they knew, thus effectively breaking confidentiality.
To retain anonymity in our public feedback (e.g. final event) and in this report, we decided to refer to all CEOs by the feminine pronoun. This was done for easier reading, instead of the more cumbersome ‘s/he’, but also as a seemingly appropriate compromise given our promises of anonymity and our analytical focus on the practices of NHS CEOs, not their genders. Equally, our promises of confidentiality, but also analytical interest in the practices (as opposed to the organisations) of the CEOs, meant that we had to strike a compromise with regard to the contextual information we could publicly share regarding the trusts that the CEOs ran and their immediate institutional environments. In particular, though there are over 200 acute and mental health trusts in the NHS in England, this is nevertheless a relatively small community of peers, each with considerable insight into other trusts, as we occasionally found during our research stays as well. However, as the particular context in which the CEO worked emerged as a critical component to understanding their knowledgeability in practice, we had to provide sufficient insight into organisational structures, wider institutional drivers and particular objectives each CEO was working towards for the reader to understand at least some of the differences between the research settings, and how these may have contributed to the emergence of a particular knowledge infrastructure. A summary of pertinent contextual information relating to all seven trusts can be found in .
Contextual information on the trusts examined in the study
Patient and public involvement
The particularities of this study meant that involvement of patients and the public remained limited to the final phase of communicating the results. This aspect of the research was discussed and agreed with members of the University/User Teaching and Research Action Partnership (UNTRAP), based at the University of Warwick. In particular, as the study’s practical relevance was for the leadership community of the NHS, as opposed to the broader public or patients, to ensure that the study was appropriately informed by views of NHS practitioners, we included notable figures from the NHS community in our advisory panel. In addition, confidentiality concerns prevented us from involving the public directly, as it was important to minimise the number of individuals aware of the identities of our participants. It is our intention to meet the obligations of publicly funded research by post-project dissemination via other media than academic publishing, allowing us to share central insights from our study with a broader audience who may be interested.
Limitations of the study
The nature of the study and the methodological choices made prior and during the project determined not only the strength and validity of the results, but also necessarily some of its limitations. These notably include the following:
While the use of shadowing allowed us to gain in-depth insight into the information and knowledge work of NHS CEOs, our focus on CEOs meant that knowledge mobilisation practices were observed only from their individual perspectives. Accordingly, although we often observed and noted the interplay between CEOs and their senior teams as part of CEOs’ knowledge mobilisation as an inherently social and interactive process, and documented the unique type of knowledge and information work that CEOs carry out, we could not observe how individual directors actually accessed ‘evidence’, gained informal insights and prepared the documents that we saw deployed in their encounters with the CEOs. Accordingly, and also reflecting our original project objectives, this study is necessarily silent on the knowledge mobilisation dynamics of a TMT as a complex whole. We suggest that this topic would strongly benefit from further empirical investigation.
The use of shadowing as a method, in combination with the theoretical orientation towards knowledge as a form of practical mastery, tends to produce accounts that are necessarily non-judgemental, thus appearing as potentially ‘empathetic’ in character. Namely, the non-judgemental nature of the accounts derives from an analytical focus on describing and documenting the practical methods through which individuals accomplish work; in this case, through which CEOs make themselves knowledgeable. In this sense, the method necessarily produces agnostic accounts that focus on practical situated capabilities, rather than shortcomings in relation to a particular normative ideal. While the methodology allowed us to identify context-specific difficulties, challenges, and internal or external contradictions (e.g. practices that did not seem to reflect the demands of the environment as presented in the daily work of that CEO), it did not allow us to correlate such activities with any general criteria of success or failure. In other words, the methodology predisposed us to capture what it made sense for informants to apparently do, without asking whether this was right or wrong, or could be done better. Although some of these aspects emerged from comparison between cases, and in fact were noted in feedback sessions with the subjects, the project was also not designed to pass judgement on the conduct of individuals. The absence of normative assessment based on certain external criteria was in fact a crucial factor in the negotiation of access. It should be added that the nature of the study also limited in part the possibility of deriving and sharing specific ‘lessons learnt’ emerging from the systematic comparison of cases. This is because a detailed description of the methods (and conditions of work) used by the CEOs would breach the guarantees of anonymity given to all participants. Finally, one may suggest that shadowing generates a natural empathy with the subject under study (and at times a sincere marvel at the competences of the informants), which may lead to a lack of critique. This reflects the method’s tendency to cast the researcher in the role of a quasi-apprentice, a position that requires an acceptance of the validity of the practice being learned. In such a situation, an apprentice who disputes the practice of the master cannot in fact learn, as critical capacity emerges later as an effect of the comparison of different forms of practice and personal experience. Following standard ethnographic practice, both researchers employed reflexivity as part of accounting for one’s role and perceptions on the observations made and conclusions being drawn. Though we believe we were largely successful in mitigating the effects of a ‘shadow’s proximity’, it is likely that residual effects of a ‘beginner’s marvel’ still transpire in some of our accounts. That being said, such familiarity would, we believe, be more problematic if our project explicitly set out to offer a normative critique or assessment of the practices under study. As we recounted in detail above, this was neither the intention nor the possibility of our study as originally conceived.
Finally, the lack of statistical generalisability of our results is a natural, though nevertheless important, limitation. Namely, while we have generated a number of new and, we believe, useful conceptual categories that can be explored beyond the seven cases reported here, we can say little about the nature of information work by the larger population of NHS CEOs in England. This reflects the nature of our study, aimed as previously noted at theoretical rather than statistical generalisation.
99 The limitation does not weaken our conclusions, however. On the contrary, the careful sampling strategy, sustained period of observation and comprehensive process of member validation make our results robust. This limitation, however, prevents us from construing our results in terms of generalised statements such as ‘NHS CEOs in England do this’. This is also one of the reasons why we cannot establish correlations between observed practices and performance outcomes.
