CHAPTER THREE: METHODOLOGY – DOOR TO THE RESEARCH PROCESS
3.3 Plan for the research
3.3.5 Case study approach
3.3.5.6 Vignette method
The vignette method involves ‘presenting respondents with one or more scenarios and then asking them how they would respond when confronted with the circumstances of that scenario’ (Bryman 2016: 259). Vignettes focus upon the simulation of real-life events, through the creation of fictional situations which are inspired by real events which have occurred.
Vignettes can be presented in different forms, for example a live event, or a case study presented in written form (Robson and McCartan, 2016). Hughes and Huby (2002: 385) state that ‘vignettes, used alone or in conjunction with other research techniques, can be valuable research tools in the study of people’s lives, their attitudes, perceptions and beliefs’. The vignette method can be used as a means to elicit views from people providing a service (Robson and McCartan, 2016). Rahman (1996) identified that the responses to vignettes by female carers of older people were similar to how they would respond in real life. Vignettes were also useful for McKeganey et al (1996), who found that participants were more likely to respond to vignettes about sharing needles among intravenous drug users and provide truthful information, rather than giving socially appropriate responses. Employing the vignette method needs to acknowledge that it does not completely capture the realities of what it studies, but the interpretation of vignettes by participants is valuable to research (Hughes and Huby, 2004). In a study by Hughes (1998), difficulties arose in the data analysis from the vignette method when some participants responded in the first person, becoming the character in the vignettes, with others responding in the third person. Distinguishing what participants really think compared to what they think is socially important is a factor to consider in the vignette method (Finch, 1987; Whittaker, 2002). This research needed to ensure that vignettes were kept consistent and the same ones were presented to all participants, to allow comparison between what might be seen as important and what actually is important, based on the responses given.
105 3.3.5.7 Coding and NVivo: Data analysis
Qualitative data can be analysed by coding. Coding is useful to: identify relevant phenomena;
find examples of these phenomena; and to establish differences, commonalities, patterns and structures amongst these phenomena by analysing the data (Seidel and Kelle, 1995). Codes are ‘tags or labels for allocating units of meaning to the descriptive or inferential information compiled during a study’ (Basit, 2003: 144). Codes can enable concepts and ideas to be developed through being linked to specific parts of the data. Thus, the researcher can take a heuristic approach, and learn more about the particular topic being researched by going beyond the data itself to form a deeper level of analysis (Coffey and Atkinson, 1996). Ideally, data such as interview transcripts should be examined several times, as ‘by reading and rereading the corpus, you gain intimate familiarity with its contents and begin to notice significant details as well as make new insights about their meaning’ (Saldaña, 2014:584).
Qualitative data analysis can be done through a specialist software, called NVivo 11. NVivo is useful for various purposes. The software can help to manage the data, which is often messy in qualitative research, including not only interview transcripts, photographs and questionnaires, but also audio, video, published research and notes. NVivo helps to manage ideas generated from the research, to search the data by running queries to answer certain questions, to visualise the data and the relationships between different items, and to report from the data using their contents, ideas and knowledge produced from the data and the analysis process to reach the outcomes (Bazeley and Jackson, 2013). Using NVivo as an analytical tool has limitations to consider. NVivo has been construed as distancing the researcher from the data, and not allowing the researcher to manually get close to it. Concerns are raised about using a ‘code and retrieve’ function, which may exclude other analytical processes. Using the computer to run NVivo may lead to perceptions that the analysis is positivist rather than constructivist. Finally, computer use through NVivo is critiqued for only supporting grounded theory approaches, and is misconstrued and critiqued for creating its own approach to analysis (Bazeley and Jackson, 2013). The data analysis process needed to be transparent in showing that these limitations were overcome through good practices. The next section discusses the overall outcomes of the research methods and how the research was eventually conducted.
106 3.4 Outcomes of research methods
The next sections of the methodology discuss the outcomes of what was planned for the research, as outlined in section 3.3. In practice, the nature of the WASH needs of PM women being hidden knowledge meant that the methodology evolved over time, and had to adapt to different ways of finding the data compared to conventional methods (literature review, design, data collection and analysis). The original planned research process incorporated these four planned stages. In reality, the outcome of the research methods ultimately led to six stages in the research process: literature review, phenomenological review, research design, case study selection, data collection and data analysis (Figure 3.3).
107
Figure 3.3: Outcomes of the research process
108 The six stages of the process reflect how finding hidden knowledge about WASH needs for PM women requires alternative approaches. Firstly, the planned use of a literature review to identify the objectives of the research did not materialise, and the wide literature gap made it challenging to identify a case study based on a gap in locations that had been studied.
The literature review did not meet its purpose to identify the objectives and led to stage two, the phenomenological review. In the phenomenological review, women from the UK and the USA who were either PM or menopausal participated in feminist oral history interviews about their WASH needs during the perimenopause. This identified their experiences of PM symptoms, and the hygiene needs that were universal to all PM women, irrespective of their WASH context, such as bathing or laundry. The WASH themes that emerged from the phenomenological review were the basis for formulating the research objectives and research questions for the study, which could not be established from the literature review.
Stage three was research design. The results of the phenomenological review were used to design the research and select the methods to be used to research the WASH needs of PM women. The research was designed around a case study method, and incorporated participative methodologies, based on the principles of considering ‘whose reality counts’ and ensuring that PM women are placed at the centre of the study for their voices to be heard.
The research design process had several iterations. The results of the interviews from the phenomenological review were analysed in stages by reflecting on emerging themes as interviews were conducted using a research diary. As these themes emerged, new literature of relevance to the research could be found relating to these data. Thereafter, this literature was added to the literature review. This process was repeated several times until the data had reached saturation point, and an appropriate research design could be established.
Stage four considered case study selection. Attendance at the 37th International WEDC Conference in Vietnam brought opportunities to network and establish links with potential partners for the research from low income countries where WASH services were lacking. This conference led to links being made with a research partner in Accra, and the data was collected in Ghana.
Stage five of the research was data collection in Ghana, with a pilot study followed by two stages of data collection with two cohorts. The first cohort was PM and menopausal women,
109 and the second cohort was environmental health professionals. Data was collected in two low income urban communities in Accra and Kumasi with PM and menopausal women. The tools selected in the research design were used to gather data about the hidden WASH needs of PM women. These data were then fed back to local environmental health professionals in the municipalities, to establish the solutions to meeting the WASH needs of PM women.
Stage six was data analysis. The original plan to use NVivo as a qualitative data analysis tool was implemented, and the data was thematically analysed. Attempts were made to create models using NVivo, and to look at the data in the context of existing theoretical models.
However, this was ineffective in establishing links between the data and identifying relationships to discuss. The data coded in NVivo was then used to establish themes for the analysis using nodes produced from the software, but models in NVivo were not used. The outcomes of the six stages of the research are now discussed in greater depth in the next sections of the chapter.