Abstract
Introduction
The Journal of Research in Nursing (JRN) uniquely positions itself among nursing journals as the home for articles that seek to make connections between policy, practice and research. Many of the papers published are accompanied by a review or commentary that adds to the discourse around method, methodology and impact. It was unsurprising then, given JRN’s raison d’être, that there were numerous empirical studies describing mixed-methods designs as well as a number of thought-provoking and stimulating, theoretically driven discussion papers in this special collection. The collection has been designed to bring that combination of theory and real-world research together for JRN readers and revisits some of the excellent work published over two decades (see table below). We include one paper from the original journal NTResearch (which was subsequently rebranded as JRN, when the journal moved from EMAP to Sage Publications), the rest are drawn from the JRN.
Mixed-methods research (MMR) has been rather grandly described as the ‘third methodological movement’ (Hall, 2012:1) and could almost be cast in the role of peacekeeper in the paradigm wars that polarised some exponents of qualitative and quantitative approaches. However, there are examples of mixed-methods studies going back to the 1960s; indeed, grounded theory (Glaser and Strauss, 1965) could be classed as MMR. Arguably valuable for examining the complexity of healthcare, MMR is not without its critics and controversies, particularly from those who embrace very different philosophical positions. Some of these are debated in the papers in this collection, such as Giddings (2006), Gilbert, (2006) and Doyle et al. (2009, 2016). The endeavour of aligning an approach to the research question then determining ‘what works best’ remains an anathema for some, pragmatism gone rogue. Others suggest MMR merely acknowledges it is the problem or research question that drives (or should drive) methodology and method rather than the worldview of the researcher. Critics may consider what is eminently sensible and pragmatic is just too easy, or easier than other approaches, lacks methodological rigour and fails to address the problem that a ‘whole’ picture may be illusory and that research is by its nature full of gaps and always partial (Freshwater, 2007). At the same time, it could be contended that mixed-methods methodologists have a more demanding role ensuring the design, linkages between data, complementariness, skills of the researcher and/or team and ensuring integration are transparent and well executed.
Giddings, the author of one of the papers included in this collection, remarked ‘labels are for jars’ (2006:198) and, in research, many debates are riven by concerns about the language used to name endeavour. Words, like ideas, are never impartial. Language privileges, thereby according discourse status. In MMR that privilege can bring balance (O’Cathain, 2018) and transparency, but can also serve to reinforce the notions of knowledge hierarchy and otherness (Freshwater, 2007). This collection brings many of these contentions to the surface and we hope it stimulates you to engage in intentional critique of the bedrock of rigorous enquiry and good science.
For the reader unfamiliar with MMR, at its simplest it is concerned with research that combines the methodologies and methods of both quantitative (QUAN) and qualitative (QUAL) traditions. There are numerous typologies used in MMR and likewise by mixed-methods researchers, which can add confusion as they are not always consistently applied and have changed and adapted over time. Mixed-methods researchers purposively design and undertake research that links narrative (QUAL) and numerical (QUAN) findings. These data are often reported in the same paper to represent the connectedness of the findings. The researchers may even have intentionally integrated the data sets to demonstrate greater completeness. To achieve this, they may have used strategies such as triangulation. This term is used to describe intentional design or data capture to facilitate convergence, as if to a triangulation point. This convergence may enable the researcher to make inferences about the topic under investigation. There are numerous ways of achieving triangulation and, although that description is probably far too simplistic, it hopefully communicates the gist of the notion. If interested, and after reading papers such as Rose and Webb (1997) in this collection, we would encourage you to read seminal texts such as Denzin (1978) on the subject. We also adopt the shorthand of upper case (QUAN, QUAL) and lower case (quan, qual) to indicate the different weight placed on the approach adopted by the researchers in a study, or as reported in the paper, as the two may not be the same. Lastly, we also use the terminology convergent, concurrent and sequential in relation to study designs and data transformation to describe conversion of words to numbers. We have done this to guide your navigation of the collection but, again, if we have whetted your appetite, we direct you to texts such as Teddie and Tasakkori (2009) and Creswell and Plano Clark (2018) for a deeper dive into MMR.
Triangulation
The earliest paper we included was published by Karen Rose and Christine Webb in 1997 and offers a thought-provoking meander through the use of triangulation. For the uninitiated, this is where different strategies, for example data, researchers, theories, or methods, are used with the intention of increasing the credibility, or validity, of the study. This reflexive account describes Rose’s doctoral journey. It navigates the anxieties engendered by concerns that her work would be open to the criticism of being ‘poorly constructed’ (Rose and Webb, 1997:109) if solely dependent on unstructured interviews and observation. With the benefit of hindsight, given their methodological hermeneutic phenomenological position, that worry was possibly unfounded. It nevertheless is a not uncommonly voiced disquiet among novice researchers. They describe selecting the Mood Adjective Check List, offering a well-reasoned rationale for the choice. Yet they found the administration of the instrument, in the context of fieldwork, counterproductive. First, they argued it might have affected the participants. This caused the authors to question the appropriateness of the instrument in an already emotionally charged situation, interviewing informal caregivers of terminally ill cancer patients. Secondly, they raise concerns about dissonance between perspectives. This is an issue that will be expressed repeatedly by authors in the context of this MMR special collection. Lastly, they question, or as Rose and Webb (1997) pen, rethink, triangulation. This reframing related to questioning triangulation as an activity or device for enhancing, possibly amplifying, and therefore arriving at a better understanding of a phenomenon.
Howarth et al. (2018) also allude to triangulation in their report of a mixed-methods evaluation of a nature-based intervention – therapeutic horticulture – designed to reduce the social isolation of people experiencing mental health problems. The evaluation formed part of a mental health recovery programme. The methodological design is described as grounded theory (Corbin and Strauss, 2008) coupled with a mixed-methods design and incorporating action research. Three different data-collection methods were used for methodological triangulation including the Recovery Star™ tool (Triangle, 2008) and focus-group and semi-structured exit interviews with volunteers. This QUAN → QUAL design reported both data sets but not any integration or triangulation process. The authors suggest the results indicated the use of this social prescribing approach supports recovery. Asset-based approaches, the team posit, help people to become empowered, gain independence with less reliance on services. Although the study is a little sparse on detail of how the different design features were executed, the results indicate real benefits for people recovering from mental health problems. As Dilworth (2018: 491) says in his commentary that accompanied the original publication ‘this paper makes a strong argument for creative attempts to help people integrate into their communities’ and ‘they [participants] felt valued for what they could do and their contributions’.
The political imperative and research
This leads neatly to Giddings’ (2006) paper intriguingly subtitled ‘Positivism dressed in drag?’, where she argues that MMR has the potential to become just an easy solution. Although described as the best of both paradigm worlds and a solution for addressing complexity, acceptable to funders, it is outlined that the challenge for health and social care researchers is not to accept naively the post-positivist imperative. Rather, the task is to use approaches that are sufficiently creative to capture the intricacies of context, interaction and difference, and that engage co-operatively with those that are the focus of our research. Given that clarion call for engagement, both Ivankova (2017) and Howarth et al. (2018) pick up the thread of emancipatory participatory methods. Ivankova (2017), for example, describes an MMR participatory action research framework as the basis for engagement of stakeholders in creating research and driving translation of outcomes into delivery of care. She places particular emphasis on the importance of integration of data sets generated in MMR; readers of this collection may wish to consider how effectively authors achieve this as they work through the papers.
Gilbert (2006) offered a different typology for considering mixed-methods approaches: the practical, the technical and the political. This emerged from consideration of the melee surrounding philosophy, theory, discipline, techniques and the everyday mechanics associated with doing research. This categorisation described practical as designs where the methods are employed independently from each other with ‘no methodological commitment’ (2006:210). Whereas technical, as the name implies, was outlined as demanding in design terms and sought to ensure methods to capture data are employed with appropriate care and transparency but are not necessarily methodologically obliged. Lastly, those studies that are designed within an overarching theoretical framework he described as political. The utility of the typology was illuminated with various examples described as technical (Spilsbury and Meyer, 2004, 2005), an example of which is included in this collection. We have included Spilsbury and Meyer’s paper (2005) not merely because of the ‘technical’ label but because of the account of a single-site case study of the impact of one group of workers (healthcare assistants) on another (registered nurses). A highly topical, possibly politically charged, issue in the contemporary healthcare context where role blurring, fracturing, creep, proliferation and assimilation are all part of the complexity of delivering effective care. Another feature of this study that is worthy of note is their creative use of different data collection methods to capture the nuanced and negotiated nature of healthcare work. More recently, Freshwater (2017) described the various paradigms within MMR as dialectical, pragmatic and relating to critical realism. At the crux of these debates are different notions of realities and how to combine and integrate them or, as Freshwater (2017:329) described, ‘to produce a whole… that is greater than the sum of the individual qualitative and quantitative parts’.
Typology and design in MMR
In the first of two papers, Doyle, Brady and Byrne (2009) present an overview of MMR. They take the reader through the paradigm debates, justification for using MMR, the different typologies and summarise the limitations of the approach. They argue that MMR has a valuable contribution to make to complex health and social care research. In their second paper, they revisit the landscape of MMR but place greater attention on the issues, or ‘hot topics’ as they describe them (Doyle et al., 2016:623). They take the reader on a ramble through different designs prior to examining the challenges associated with integration or bringing together the different elements of data capture to make (meta) inferences. This they discuss using a typology of different levels to illustrate how the process can be undertaken. They identify the not uncommon challenge for mixed-method researchers where the qualitative and quantitative elements are reported as if unconnected; indeed, they are sometimes reported separately as if two studies.
Luk’s (2009) paper describes an evaluation of the effects of introducing a protocol designed to enhance the caring behaviours of nursing staff in hospitals providing long-term and rehabilitative care in Hong Kong. A pre- and post-test mixed-methods design, or in MMR vernacular, a sequential QUAN → QUAL or possibly QUAN → qual study. Patients and carers were asked to complete a questionnaire before and after introduction of the protocol. Then focus group interviews were undertaken with staff on two wards, sampled on the basis of scores on the patient questionnaires (one lower, the other high).
Rath et al. (2017) present a mixed-methods pilot study of a structured exercise intervention on older women with cancer. They describe how the impetus for the qualitative study came from a chance remark of a relative about the benefits of the programme. This highlighted the need to capture the rich insights that participation brought. This study is described as a sequential explanatory design and could be notated as QUAN → qual (Creswell, 2003) or might better be described as a post-hoc qual design. The salient findings are illustrated in Table 3 (Rath et al., 2017:276) where the results from the two phases are compared. The absence of statistical significance is unsurprising as the study was a pilot designed to test acceptability and the possible impact of exercise, but the authors claim that all concepts or variables are clinically significant as a consequence of the findings of the qualitative analysis. As Downing (2017) suggests in her review of this study, after the addition of the qualitative phase, this foregrounded the unintentional impact of an exercise programme on women that might otherwise have been lost.
Wye et al. (2016) set out to investigate the relationship between the use of an electronic record system for coordinating palliative care and the likelihood of a patient experiencing a home death. Described as a service evaluation, the study avoided the need for NHS ethical approval, but the team obtained the necessary university approvals and study site governance and data security permission. The research team is described in full and, interestingly, largely made up of qualitative researchers. There is reference to a larger study but it is not clear what that was, making it difficult to judge what sort of MMR design was used; was it convergent, sequential or concurrent and did the quantitative and qualitative data carry equal weight? The team conducted a large number of interviews, captured in a variety of ways, to maximise spread and ‘representativeness’ of stakeholders. Framework analysis (Ritchie and Lewis, 2003) was used for the analysis but few verbatim quotes are included in the text of the report and most data were quantified. That said, the approach highly effectively interrogated the assumption that a relationship existed between active health records and place of death.
Language and MMR
Given the increasing number of older people, especially frail elderly, living in residential and home-care settings, Fleming and Kydd’s (2018) study to explore what makes nursing homes homely seems timely. This paper reports the final stage using Q methodology of a three-part study. Interestingly, they found residents identified ‘belonging’ as a dimension of homeliness, whereas staff and family members placed importance on characteristics already articulated in national standards. Q methodology uses qualitative methods to capture subjective accounts, then seeks to categorise and group them (Brown, 1996). This sorting can be by hand or using a statistical package, but analysis collects the person-by-person loading on a particular factor or construct. The authors make no effort to define the study in terms of an MMR typology. It might be described as convergent or sequential exploratory, but it is difficult to estimate the weight of each phase as only the final Q methodology stage is reported. This paper underscores the challenges facing researchers when trying to accurately describe what they have done but not finding fit with existing discourses. For authors the focus should be on transparent communication hence alignment of purpose, sequence and integration are more important (O’Cathain et al., 2008; Hong et al., 2018).
Saint Arnault and O’Halloran’s (2016) study sought to understand the ‘healing trajectories’ of women who had experienced and left abusive domestic relationships in a rural region of Ireland. Similar to the Rath et al. (2017) study, it is theoretically driven but in this case draws on cultural determinants of help seeking and concepts from neurobiological theories. They describe the study as a sequential explanatory MMR design with a nested quantitative sub-sample. It is not totally apparent if this report relates to just the initial sequence (baseline data) or reports the nested sub-sample; we leave that to the reader to untangle. This challenge of adequate description is one raised recently by O’Cathain (2018), who makes the observation that researchers often change their descriptions over the lifetime of a study. Sometimes a result of how terminology is refined over time, possibly in the case of MMR as a relatively newish star on the firmament, researchers are developing the discourse. Hence, how a study is reported may differ between the protocol and any publication of the results. As no registration of study is mentioned, it is only possible to conject if this was the case here. Other features of this study worthy of note are the transformation through quantification of qualitative data, convergence, and the role of theory in MMR.
Simons (2007) offers a fascinating and thought-provoking account of the ‘personal tensions’ (p. 74) encountered while undertaking an embedded qualitative case study as part of a randomised controlled trial. Embedded is an interesting trope that readers may want to consider (see Plano Clark et al., 2013; O’Cathain, 2018), particularly when presented together with the intriguing statement that ‘any research taking place alongside a trial cannot interfere with the protocol’(2007:77). This could be considered a curious position when the term embedded is used to describe the overall design. Embedded suggests nested within, but in this study more accurately appears to be parallel to, or observant of, the investigation rather than being integrated. Indeed, as the rest of the very eloquent paper illuminates, the author’s experience of the subordinate nature of qualitative research in the trial context, possibly due to what she describes as ‘conflating method with paradigm’ (2007:79) and the delivery imperatives associated with keeping output targets on time, are stark. This paper is particularly significant given the upsurge in funded qualitative sub-studies as part of randomised controlled trials but still far from operating where the qualitative phase is as equally valued (O’Cathain, 2018).
Conclusion
Any reader of research has to make a judgement about the worth of the investigation. In MMR that decision is more complex as the consumer has to adjudicate the separate components to assess the quality of the science, then the contribution it may make to the body of knowledge in the field. This is particularly important in nursing where the impact of our work is often poorly articulated, understood and so intertwined with others contributing to care delivery that it can become lost. In this collection we have attempted to show the continuum and advancement of MMR as seen through the lens of one nursing journal and, importantly, the contribution nursing scholars have made. We have included papers that have sought to examine the many debates surrounding the use and value of MMR. We have highlighted the challenge that language precision presents for authors when describing paradigm, methodology and methods in MMR. We have also sought to bring to the fore the potential of MMR to integrate the voices of those that have traditionally been subjects in quantitative research, participants in qualitative studies and latterly more commonly as co-producers of research.
