Phase Two: Data Collection And Provisional Analysis Of Qualitative Material.
We used the semi-structured in-depth interviews, non-participant observation, documentary analysis, Q methodological and conventional statistical modeling as a means of accessing the perceptions, information use and decisions of nurses.
Multiple Methods Of Data Collection.
We adopted a multi-method approach to data collection in light of the potential advantages of the approach:
triangulation – we considered that convergence (or explainable divergence) of results would add to the internal validity of the final analysis
complementarity – the methods complemented each other in that they examined overlapping but separate characteristics of information use and clinical decision making
initiation – we considered that fresh, unexpected, perspectives, contradictions or paradoxes may result from a combination of approaches
development – we felt that using the methods sequentially (so that each adds a new slant to the data and theory derived from the previously applied method) would inform the application of the methods which followed
expansion – we considered that breadth and scope to the project would be enhanced through the use of mixed methods (Turner, 1987 ).
The use of multiple methods had previously been used to good effect by McKeganey et al. (1988) to explore the micro decision processes and contexts of professionals. In their study, as in ours, interviews and observation each provided complementary slants on the problem of researching decisions and the information used in them. They summarise the strengths and weaknesses of each approach thus:
Table 2: characteristics of interview and observational data collection in decision making research
| | Interviews | Observation |
| Data level | Individuals | Processes |
| Decision points | Multiple | Few |
| Triangulation of accounts | Strong | Less strong |
| Components of decisions | Formal | Informal |
| Routine decisions | Less strong | Strong |
| Non-decisions | Weak | Weak |
| Rationality/non-rationality of decisions | Overstates rationality | Strong on non-rationality |
| Disclosure of private accounts | Less strong | Adequate |
(c.f. McKeganey et al., 1988)
Despite committing ourselves to multiple methods of data collection we were mindful of the fact that using multiple methods of data collection is rarely easy and that perfect data convergence is rare (Bishop & Scudder, 1997). Ultimately though we considered the complexity of having to incorporate possibly divergent results into an overall descriptive/theoretical framework was offset by the richness of description and gains in internal validity that the approach offered.
Depth Interviewing
Interviews are social events and contextual factors such as power, gender, preconceived notions of ‘other’ and ‘self’ play a role in interpretation (Baker, 1982; Tidd, 1994). Recognising this, we adopted an open ended approach based on the three or four key areas we wished to explore in detail (see the topic guide in Appendix B). The interviewer was left to explore and develop issues as they arose during the interviews. After Denzin (1970) we recognised that this approach had three advantages:
. it allowed nurses to use their unique ways of defining the world
. we assumed that no fixed sequence of questions was going to be suitable for all respondents
. respondents were able to raise important issues not contained in the topic guide.
The interviews undertaken in the pilot study were used as sensitising exercises for the interviewers and played no part in the final analysis of results. This exercise involved us exploring ways of asking questions and the level of probing required to isolate the phenomena of information use and clinical decision making in accounts.
108 Interviews were carried out by DM and CT. by arrangement with nurses. Most interviews took place in the Ward Manager’s office but DM had to a resort to a storage cupboard on one occasion! We offered interviewees a choice of location wherever possible in an effort to make them more comfortable and to foster a more neutral atmosphere.
Observation
The policy emphasis associated with the use of research information by nurses is a measure of the importance the profession and policy makers attach to this aspect of healthcare activity. Moreover, the volume of research papers on the topic of nurse decision making is testimony to the fact that nursing itself promotes the idea that nurses routinely make autonomous clinical decisions as a key characteristic of their professional status. It is fair to assume that many nurses are aware of these policy and professional expectations. It is also fair to assume that, having recognised this, nurses will adjust their accounts of research information use accordingly. For this reason, observational data was used to examine what nurses did as opposed to what they say they did. It was intended to capture the routine, easily forgotten, or that which they would rather not include in verbatim accounts of practice. The technique also gave us a qualitative sense of the frequency of use of information sources.
Jorgenson argues that observation is a suitable strategy for data collection in research where, 'there are important differences between the views of insiders as opposed to outsiders' (Jorgenson, 1988). Clearly there are elements of professional decision making, and the information-use that accompanies it, which have an insider-outsider dimension. We adopted the participant as observer5 model as described by Roper and Shapira (2000). This model suggests the participant as observer role increases the likelihood that the researcher will obtain key “insider” information about what it is like to be a member of the cultural group. This role also enabled us to validate observations with the participants while observing, interpreting and recording their decisions and information use.
Our status as observers, and the social and qualitative nature of the data collection, made it likely that our presence had an impact on the behaviour of the nurses themselves. We minimised these effects via a number of strategies:
. taking time to build a rapport with the nurses being observed
. wherever possible observing those nurses we had previously interviewed6
. spending enough time on a unit for the staff to get used to seeing us. This was aided by the case site approach generally which involved lengthy periods in the field – interviewing and auditing documents
. stressing our clinical (as opposed to academic) backgrounds.
Documentary/Resource Audit
Literature and text/electronic-based resources on wards give valuable insight into an organisation’s support for dissemination of information (Forster, 1994). We used documentary audit to describe the text based evidence available for nurses on wards. We also wanted to cross-reference sources referred to in interviews. The audit of the data took the form of a single researcher (DM) hand searching materials, recording titles and the type of resource involved (e.g. peer reviewed journal, text book, CD-ROM, educational resource pack) on each of the wards in the three sites. The results are presented in Chapter Six. It should be noted that the Coronary Care Unit in 36 Site Three was in the process of being refurbished and for this reason an audit of their documentary resources was not feasible: most of the material was packed away and sealed in preparation for their temporary relocation.
Provisional Analysis Of Qualitative Data This provisional stage in the analysis of the qualitative data extracted the initial descriptive themes relating to the perceived barriers to research information use by nurses, how nurses perceive (and actually) accessed research information, as well as those sources of information nurses discussed - or were observed using - in practice. At this stage the analysis focussed on the establishment of first-level coding categories which could be used to represent the main findings and as the basis for the further exploration using Q methodology (see below). The codes were first agreed by DM and CT and then the rest of the research team. Disagreements were settled by discussion within the wider team.