Chapter 4: Research questions and methods
4.3 Methodology
4.3.3 Embedding a qualitative element
As discussed in section 4.5, many participants provided detailed accounts about the acceptability of the Communicative Participation Item Bank (CPIB) and wider issues of speech and communication, which exceeded the level of detail I had expected.
Research with human beings always involves an element of unpredictability (Cziko, 1989) and it is important to respond to participant wishes. It would be unethical to waste these data, which could firstly provide a valuable insight into participants’ experiences of speech and communicative impairment and secondly provide a participant’s eye view of the acceptability of the CPIB. It was agreed that secondary analysis of anonymised already collected data did not require an ethics amendment.
Once it had been decided that analysis of these comments would be performed, I had to determine the most appropriate analysis framework for these data. Evidently, it is more challenging to design a suitable analysis once the data have already been collected since data collection cannot be modified to suit the chosen analysis framework. Whereas quantitative analysis primarily addresses questions of fact (see section 4.3.1), qualitative analysis methods primarily address questions of process and reason (‘how’ and ‘why’ questions) and offer greater insight into participant experiences (Richardson et al., 2011, Sullivan, 2010).
Qualitative research is built on different philosophical foundations from
quantitative research. Adopting a positivistic research philosophy leads a researcher to be sceptical of participant experiences, use objective methods, favour quantitative data, seek strict experimental control and emphasise the importance of replicability (Robson, 2002). Not all philosophers and researchers share the positivist view. Brewer (2000) emphasises the importance of studies approximating real-life situations to have ecological validity. Social constructionists, for example, emphasise the formulation and maintenance of knowledge through social processes (Burr, 2003, Berger and Luckmann, 1966, Barnes, 1974). Adopting a social constructionist research philosophy leads a researcher to
examine evidence in terms of whether it is plausible and compelling rather than seek truth, investigate why people hold certain views, often use language-based research methods and value participant accounts (Sullivan, 2010). Moreover, judgements of the
transferability of findings outside their original context are left to the reader’s judgement (Richardson et al., 2011). It has been argued that constructionism (the related term ‘constructivism’ is used by some authors) is the only appropriate framework for qualitative
research (Guba and Lincoln, 1989). The compatibility of qualitative and quantitative methods is discussed in section 4.3.4.
I decided that qualitative methods would be the most appropriate to analyse my comment data. They are uniquely suitable to the analysis of textual data, provide an insight into participant experience and emphasise the social perspective. Regardless of whether the data were collected orally or in written form, the data could be considered textual in nature. There were several potential analysis frameworks for data of this nature. These included discourse analysis, conversation analysis, thematic analysis and
qualitative content analysis. I shall now outline these in turn and provide a rationale for my decision.
Discourse analysis (Wiggins and Riley, 2010) is a means of assessing discourse, which sees it as representing a particular construction of reality, which in turn has
consequences for the speaker’s social interactions and self-concept. It seeks to
understand how the combination of words into a text or other discourse form projects a view of reality. Discourse can include written, oral and pictorial information. Conversation analysis is a means of assessing a variety of structures within a conversation (Forrester, 2010). These include turn-taking, sequence and emphasis. It seeks to understand how people interact during a conversation.
The above approaches focus on construction of meaning from discourse and interaction during conversations respectively. I decided that they were not suitable for my analysis. I required an analysis framework that instead focuses on extracting key themes from a text or transcript. Therefore, I investigated thematic analysis and qualitative content analysis further, in order to assess their suitability.
Although thematic analysis and qualitative content analysis are both common methods in qualitative healthcare investigations, there has been a lack of definitional clarity regarding the distinction between the methods (Vaismoradi et al., 2013,
Sandelowski and Leeman, 2012, Braun and Clarke, 2006). These methods both employ a relatively low degree of interpretative transformation (Sandelowski and Barroso, 2003). The methods have a lot of shared ground, although I shall outline some key differences. For an exhaustive discussion of these two methods, see Vaismoradi et al (2013).
Thematic analysis is a purely qualitative analysis method (Braun and Clarke, 2006). On the other hand, although qualitative content analysis is now widely used, early content analysis was primarily quantitative (Graneheim and Lundman, 2004). Content analysis allows simultaneous quantitative and qualitative analysis (Gbrich, 2007). Content analysis allows analysis of patterns of word use and communication strategies in addition
to extracting themes (Powers and Knapp, 2006). In this regard, it is a more versatile approach.
I decided to use qualitative content analysis (QCA). In addition to performing qualitative analysis of the key themes in the data, it allowed me to quantify how many participants contributed to each theme. This in turn permitted me to assess the most common themes across the sample. I performed an inductive rather than deductive content analysis since there was limited extant knowledge about the phenomenon of study, with the result that it was more appropriate to work from the specific to the general. Analysis procedures were based on a published framework (Elo and Kyngäs, 2008) and are described in detail in section 6.3.1.