Pros and cons when using collaborative data analysis in nursing research: A discussion paper

Abstract

This discussion paper presents a discussion of pros and cons when using collaborative data analysis in nursing research, exemplified by our own attempt to include registered nurses from clinical practice in the analytical process of textual data. As the nursing profession covers diverse practices and roles, incorporating perspectives from these in nursing research is essential, to produce research that is recognizable and usable to both practice, education and academia. We therefore set out to try out a practical consultancy-style approach to collaborative data analysis in nursing research. The attempt proved to have several pros, few considerations and no cons.

Co-research

Co-research has the aim of bridging the gap between theory and practice and developing a more democratic and bottom-up approach to research. ¹ The benefits of co-research include diversity in knowledge and insight, which enhances the collection of meaningful data and strengthens the results as well as the sense-making process.^2–4 Co-research could amplify the voices of informants, research users, patients, and relatives, making them heard in the development of research concerning them and their lives. ³ Co-research can be used broadly and involve many different individuals. Examples could be researcher triangulation, inter-professional research teams or teams including patients and relatives. ⁴ A substantial amount of literature has been published on patient involvement and its benefits, risks, results and workability. ² Clinician and researcher co-research, however, has received less attention. Hartley and Bennington argue ‘… knowledge is not only transferred between academics and practitioners but is often jointly created through dialectical process of enquiry based on different interests and different perspectives’.^{1(p. 464)} Research then becomes beneficial for education, management, clinicians, academia and ultimately patients, relatives and communities. ⁴ Due to this, we would like to highlight co-research as a method that holds potential to strengthen and diversify nursing research. In this discussion paper, we would therefore like to put forward and discuss the pros and cons of co-research in nursing research, exemplified through our own endeavor of including two registered nurses from clinical practice in a collaborative data analysis of empirical textual data. Collaborative data analysis (CDA) is a way to engage non-academic research-users in the translational phase of the research cycle. ⁴ Although this is not a new approach, the use of CDA in research is still rather limited.^4,5 Cornish, Gillespie and Zittoun describe collaborative analysis as ‘A process in which there is joint focus and dialogue among two or more researchers regarding a shared body of data, to produce an agreed interpretation’.^4(p.79) Much of the emphasis and many of the benefits of CDA can be attributed to the multi and diverse perspectives brought to the table. ⁴ Studies of patients and relatives in CDA have concluded that, while the process was beneficial and rewarding, especially concerning lived experiences, it was very time consuming and presented several challenges.^3,6 Lack of skills in data analysis, magnitude of data and logistics were named as barriers to a widespread adoption of the method.^6,7 Locock et al., ⁶ who presented raw data transcripts to individuals with lived experiences, concluded that less data and more conversation was essential to increase user involvement in data analysis.

Process and experience of using collaborative data analyses

To develop a practical way of applying CDA we identified relevant literature that could assist us. We conducted literature searches in MEDLINE, CINAHL and SocINDEX as well as Google Scholar and books on qualitative research. Our searches did not yield any previous studies that have explored CDA as a method in nursing, nor studies that outline a practical way of applying CDA with nurses or other health professionals. Our literature search echoed the findings of Slattery, Saeri and Bragge, ⁸ who found that while studies often described clinicians such as nurses as end-users and consumers of research, there was no mention of how these groups could be involved in co-research and no recommendation for designs of co-research studies. We therefore set out to identify literature that used CDA as a method in general. This literature consisted mostly of studies with patients and relatives involved in CDA or in co-research in general. We used these studies, their methodical and practical findings and discussions, as inspiration to work out our own approach considering that we would be working with nurses as participants instead of patients and relatives. Inspired by Locock et al., ⁶ we decided that summarized findings would be our starting point for an analytical conversation with two registered nurses from clinical practice. We therefore conducted an initial analysis using the first of Kvale and Brinkmann's ⁹ three levels of understanding – self-understanding. At this level researchers summarize findings into categories with a minimum of interpretation. To do this, we combined the data from three group interviews we had conducted where the two registered nurses, among others, had participated as informants. We read it as a whole, identified codes, coded the material and, lastly, summarized it into five categories. These summarized findings were then presented to the two registered nurses at a CDA workshop with an aim of reaching the next level of Kvale and Brinkmann's ⁹ levels of understanding – critical common sense. A presentation of initial findings to workshop participants has previously been used in CDA when using a consultancy style approach,^3,5 but the actual details on how to go about it are not specified. We therefore set out to explore how an analytical conversation could be conducted. The Sage Handbook of Qualitative Data Analysis describes qualitative data analysis as ‘the classification and interpretation of linguistic material to make a statement about implicit and explicit dimensions and structures of meaning-making in the material and what is represented in it’.^10(p.5) We aimed to express and exchange our individual interpretations with the goal of a common sense-making process that could ignite a more comprehensive and profound understanding of the data material. The conversation started with reading the summarized findings aloud one category at a time, followed by the question ‘what can we make of this?’. This was followed by a conversation between the registered nurses and the first and second author of this article. The analytical conversation was not audiotaped, but notes were taken. After the workshop, we finalized the analysis using the last of Kvale and Brinkmann's ⁹ three levels of understanding – theoretical understanding.

Through reflection on our experience of using CDA as well as comparing transcripts, analyses and notes, we were able to identify four specific outcomes of using CDA. The first outcome was validation and feedback. The registered nurses attending the CDA workshop felt the summarized findings captured what had been said and discussed during the group interviews as well as mirroring their daily lives and practice as nurses well. In other studies, involving patients and relatives, CDA has been found to be very beneficial with regard to validation and feedback. ⁷ Second, our experience was that the participants brought nuance and depth to the initial summarized findings we had produced. After the CDA workshop, the analysis was rewritten and finalized by revisiting the transcripts and using the notes from the CDA workshop. When comparing the summarized findings with the final findings produced after the CDA workshop it became clear that the formulation and wording had slightly changed. Some categories had also been naturally highlighted while others had received less attention. The analytical conversation resulted in reducing the initial categories identified as it became apparent that some categories actually reflected the same phenomenon. This is similar to the findings of other studies using CDA with patients.^3,11 These studies found that patients and relatives participating as co-researchers interpreted data through an overlapping but different lens than the academic researchers and were thus able to nuance results as well as to rename or highlight certain aspects. Other studies have found that the depth and nuance added by lay-researchers enabled academic researchers to draw connections between phenomena by confirming initial ideas and adding specificity. ¹² Third, an outcome of using CDA was in regard to clinical implication of findings. The registered nurses from clinical practice spontaneously moved the focus of the conversation in the direction of clinical implications. This is not uncommon in co-research, especially if the team consists of a combination of insiders and outsiders.^1,2 Academics who are outsiders often aim to produce general and nomothetic theories, whereas clinicians who are insiders look for practical solutions to the challenges they experience. ¹ Our intention was not to produce general and nomothetic theories but to theorize nursing using clinical practice as the starting point. However, the different perspectives that allow for rich interpretation and collaborative knowledge also allow for different foci. Retrospectively, we have concluded that during the CDA process we alternated between focus on the overall aim and focus on our own practice. This made us consider that this might be the general situation for participants in CDA. Naturally, participants might have their own practice and implications for their own practice in mind. In this instance, the registered nurses reflected on the findings with their own clinical practice and we reflected on what the findings would mean for nursing education and research. It is our experience that this alternation between the common collaborative aim and one's own practice leads to both collaborative knowledge and personal reflection and knowledge generation. Perhaps the most prominent outcome of the CDA process was that we as researchers grew in our understanding. For us, the CDA workshop was very fruitful and rewarding. The spontaneous validation of the summarized findings voiced by the registered nurses and the alignment they expressed between the summarized findings and their lived experiences as registered nurses from clinical practice added additional robustness to the findings and confirmed their relevance. The analytical conversation further provided a sense of confidence and faith in the summarized findings and supported the further direction of the analysis. This is in line with previous studies describing the major benefit to academic researchers of involving co-researchers who are also end-users.^2,3

Considerations and conclusion

While we were able to identify four pros of using CDA in our study, we are not able to identify any specific cons, but would like to highlight a few considerations. First, in our study we decided upon letting summarized findings produced by us be the starting point of the analytical conversations as opposed to transcripts, based on the recommendation of ‘more conversation less data’ by Locock et al. ⁶ We are, however, not blind to the fact that producing summarized findings is an analysis in itself – all be it an initial one – from which the registered nurses in our study are excluded. Despite the validation of the summarized findings we received from the registered nurses who were also participants in the group interviews that were being analyzed, we are not able to conclude that the results would not have been different had the registered nurses been a part of the entire analytical process – in fact they most likely would.

A possible con when using CDA is that of the lay-researchers’ ability to engage. We have reflected on the ability of the registered nurses to engage in the CDA workshop. While our experience was that they were proficient in engaging, the registered nurses themselves expressed a slight hesitation about their own ability to take part in the CDA. Registered nurses in the clinical setting have knowledge of research methods from their bachelor studies, but this knowledge may be partly forgotten in busy clinical everyday life. Researchers are usually so used to analyzing and reflecting that they easily forget how these qualifications are learnt and that non-academics may not be as trained or experienced in these areas. ⁶ Studies involving patients and relatives as co-researchers often describe the lack of familiarity with analyses as an obstacle. Many academic researchers include or reflect on the possibility of including training in data analysis for lay-researchers. ¹³ This could be a good idea in order to ensure basic skills and competences to participate, even in studies like ours involving health professionals as co-researchers. It is, however, important to reflect upon how much training lay-researchers can receive before they transform into actual researchers, and the uniqueness of their involvement is lost. The fine line between lay and professional researchers is even thinner when working with health professionals. In our study, we concluded that the registered nurses did not need training before the CDA workshop. A decision we feel was right given the registered nurses’ ability to participate and add valuable input.

An important consideration when using CDA is team constellation and how it is reflected in the findings. Experience, age, specialty and practice, clinicians, academics or managers are all aspects of team constellations that affect the collaborative knowledge and thereby ultimately the findings. Collaborative data analysis offers a unique way to subsidize the research team with whatever perspective or lens might be missing or deemed beneficial. Ultimately, as everyone brings his or her own perspective to the team, the constellation of the CDA team matters and should therefore be considered, described and reflected upon in studies using CDA.

In conclusion, we would like to emphasize that CDA has many pros. The approach has potential in nursing research as it offers a way to involve nurses from different nursing contexts and across different roles, who bring their unique perspectives and knowledge of nursing into the analytical process. It is therefore our hope that more research conducted by nurses in the future will use co-research and/or CDA and thus potentially add to a more collective and comprehensive knowledge base in nursing. There is a substantial amount of research on involving patients and relatives in co-research. However, more research is needed on how to specifically engage health professionals such as nurses in co-research and CDA as well as the concrete steps of the CDA process.