6
Categorisation by Method
Communication, Depression and Primary Care
As I was assigned the task of conducting a mixed methods review of the literature on communication, depression and primary care, my next concern was to address the sorts of methods researchers in this field were using. I had been aware as I was scoping the literature that some of the articles explicitly stated their methodological stance in their titles e.g. ‘Patients perceptions of entitlement to time in general practice consultations for depression: a qualitative study’ (Pollock and Grime, 2002) or ‘Unwritten rules of talking to doctors about depression: integrating qualitative and quantitative methods’ (Wittink et al., 2006). Whilst the methodological framework was made explicit in these cases however, most studies did not provide such indicators, and following a team meeting with my colleagues, it was agreed that dividing my papers up into categories according to the methods used would be helpful to get a picture of what should be included in a mixed methods review. I had some degree of confusion at this point as to what, precisely, my mixed methods review would include. Would it include qualitative, quantitative and mixed methods papers, or would I focus only on those papers which were mixed methods? Taking a stock of the methods used in the papers that I had, I felt, would go some way in answering these questions by giving me an idea of the number of papers in each category. Whilst there have been different systems advocated in the literature for classifying mixed methods studies (Bryman, 2007; Morgan, 1998), I decided that developing my own taxonomy for my papers and then linking this back to the literature would be the most appropriate way of proceeding as I had papers that were mono method as well as mixed methods studies.
To begin categorising my papers, I firstly removed the two papers that were editorials on the field of primary care and depression. These papers would be useful later in writing a review, however were not empirical studies and thus were excluded, leaving me with 41 papers and abstracts for categorisation. The next stage of my categorisation involved me reading through firstly the abstract summary of the methods for each paper, and then the more detailed ‘methods’ section within the paper, including the analysis phase. For some papers, I discovered that it was necessary to read the entire paper including the results in order to categorise the paper, as the methods section had not been explicit enough about the degree of integration of the qualitative and quantitative findings. Whilst I began by jotting down categories for the methods used e.g. ‘single method qualitative study’ in a notebook with some of the names of the articles below it, I quickly came to realise that this system was a rather tedious method by which to categorise the papers and instead opted for a rather large table and the use of post-it notes! On the post-it notes I wrote the methods the articles appeared to be using, and began creating piles underneath them. The first two categories I created were ‘single method qualitative studies’ and ‘single method quantitative studies’ under these headings, I categorised all those papers which had used a single method of enquiry, whether this be semi-structured interviews or a questionnaire. I was careful to exclude at this stage those studies which had used qualitative or quantitative data by which to sample for participants, as I was unclear at this stage whether these could be considered single method studies. After going through the first 8 papers or so, I decided that I would exclude from my system the summaries of abstracts that I had written for the papers I had not yet got access to in full. This was because, having read both the abstract and methods section of 8 other papers, it was not always possible to classify a paper’s methodology on the basis of its abstract alone. Indeed, some papers described as qualitative studies in the abstract and title later described relying on a survey to sample individuals into the study, which muddied the waters of my single method categorisation and forced me to address the question of how, precisely, I was going to define a mixed methods study for the purposes of this review. Within the literature, different definitions of mixed methods study exist. For writers such as Darlington and Scott (2007), a mixed methods study is one that uses different methods within one study, even if they produce the same ‘kind’ of data, i.e. qualitative or quantitative. Other definitions emphasise the usage of the two different kinds of data, even if one is used for sampling and is not included in the results (Tashakkori and Teddlie, 1998), whereas others argue that a mixed methods study is one in which both qualitative and quantitative data are generated and integrated at the stage of analysis (Creswell, 1995). Whilst to those first approaching mixed methodology the range of definitions presented in the literature may appear daunting or confusing, it is in the process of writing a literature review that the need for a robust and clearly defined definition becomes central.
For the purposes of my own categorisation system, I decided to draw a distinction between those studies that produced either qualitative or quantitative data, and those studies that handled both. I was struck when reading through the literature by the number of studies that used a plurality of methods, but produced one form of data. When studying depression, primary care and communication, these studies utilised such methods as focus groups which combined the use of both discussion and vignettes (Prior et al, 2003), or for those studies using quantitative data, the combination of such methods as self-report questionnaires with the analysis of video recordings of consultations (Tylee et al, 1995). These studies differed from the studies that I referred to as ‘single method qualitative’ or ‘single method quantitative’ studies as they used a range of techniques in which to generate their data. I therefore referred to these groups as ‘multi method qualitative’ and ‘multi method quantitative’ studies. These two groups ended up constituting over a third of the papers I classified, with 8 multi method quantitative papers, and 4 multi method qualitative.
Now that I had distinguished between those papers generating entirely qualitative or entirely quantitative data (whether using a singular or plurality of methods), I was left to categorise those studies that generated and used both forms of data. One of the issues I particularly struggled with was how to classify those papers which made use of two kinds of data, but one was used merely for sampling. I felt that these papers were markedly different than those where the two forms of data were integrated in the analysis and presentation of the results. I therefore decided to classify those where there was integration at data analysis as the ‘pure mixed methods’ studies. I decided to refer to them in this way as these were the studies that had often been specifically designed with this integration in mind, and presented themselves as such (e.g. Bogner et al, 2009; Barg, 2006).
The final two classifications I used to order the papers I had accumulated were ‘quantitative mixed methods’ and ‘qualitative mixed methods studies’. I used these classifications to classify those papers that had used a combination of qualitative and quantitative data during the study, but nevertheless presented as either primarily a qualitative or quantitative study. For example, the study by Burroughs et al (2006) used quantitative data from a feasibility study with which to recruit elderly participants into their study, whereas the study by Sleath and Rubin (2002) began with qualitative analysis of audio recordings of consultations in primary care using a coding instrument which could then be operationalised into quantitative variables for statistical analysis.
These seven categories made up my final taxonomy for classifying the studies I had identified. Whilst the means of classification that I used was far less sophisticated, than, for example, Bryman’s 48 category system (Bryman, 2007), it enabled me to get a much clearer picture of the types of methods being used in this area, and thus what sort of review would be suitable to reflect on the literature. It was only through systematically analysing the methods of each study that this system was developed, as I discovered that some studies referring to themselves as a ‘qualitative’ or ‘quantitative’ study, came to be classified differently by my own system. Table 1 shows these categories, and the number of studies in each category, after the abstracts and editorials (10 in total) were removed from my sample.
Table 1: Categorisations and number of Studies
Category / Number of StudiesSingle Method Qualitative / 6
Single Method Quantitative / 3
Multi-Method Qualitative / 4
Multi-Method Quantitative / 8
Qualitative Mixed Methods / 3
Quantitative Mixed Methods / 2
Pure Mixed Methods / 7
Total / 33
Having completed the categorisation of the studies, I could then begin to think about which to include for my review.
References
Barg, F., Huss-Ashmore, R., Wittink, M., Murray, G., Bogner, H. and J. Gallo. 2006.‘A Mixed-Methods Approach to Understanding Loneliness and Depression in Older Adults’ in Journal of Gerontology 61 (6) pp.s329-s339.
Bogner, H., Cahill, E., Frauenhoffer, C. and F. Barg. 2009. ‘Older Primary Care Patients Views Regarding Antidepressants: A Mixed Methods Approach’ in Journal of Mental Health 18 (1) pp. 57-64.
Bryman, A. 2007. ‘Barriers to Integrating Quantitative and Qualitative Research’ in Journal of Mixed Methods Research1 (1) pp 8-22.
Burroughs, H., Lovell, K., Morley, M., Baldwin, R., Burns, A. and C. Chew-Graham. 2006. ‘Justifiable Depression’: How Primary Care Professionals and Patients View Late-Life Depression? A Qualitative Study’ in Family Practice 23 pp.369-377.
Creswell, J.1995. Research design: Qualitative and quantitative approaches Thousand Oaks, CA: Sage.
Darlington, Y. and D. Scott. 2002. Qualitative Research in Practice: Stories from the Field Buckingham: Open University Press.
Morgan, D. L. 1998. ‘Practical Strategies for Combining Qualitative and Quantitative Methods: Applications to Health Research’ in Qualitative Health Research 8 pp.362-376.
Pollock, K. and J. Grime. 2002. ‘Patients’ Perceptions of Entitlement to Time in General Practice Consultations for Depression: Qualitative Study’ in BMJ 325 pp. 687-693.
Prior, L., Wood, F., Lewis, G. and R. Pill. 2003. ‘Stigma Revisited, Disclosure of Emotional Problems in Primary Care Consultations in Wales’ in Social Science and Medicine 56 pp.2191-2200.
Tashakkori, A. and C. Teddlie. 1998. Mixed Methodology: Combining qualitative and quantitative approaches Thousands Oaks, CA: Sage
Tylee, A., Freeling, P., Kerry, S. and T. Burns. 1995. ‘How does the Content of Consultations Affect the Recognition by General Practitioners of Major Depression in Women?’ in British Journal of General Practice 45 pp. 575-578.
Sleath, B. and R. Rubin. 2002. ‘Gender, Ethnicity and Physician-Patient Communication about Depression and Anxiety in Primary Care’ in Patient Education and Counseling 48 pp. 243-252.
Wittink, M., Barg, F. and J. Gallo. 2006. ‘Unwritten Rules of Talking to Doctors About Depression: Integrating Qualitative and Quantitative Methods’ in Annals of Family Medicine 4 (4) pp. 302-309.