3. RESEARCH DESIGN AND METHODOLOGIES
3.2 Research Methods
This section examines the specific methods adopted in this study. Table 1 illustrated the structure of the research undertaken including the contribution of each publication in the research question and research objectives, and the methods applied in doing so. A multi-method approach, covering literature review, document analysis, surveys, semi-structured interviews and cases, is adopted with qualitative data collection and qualitative data analysis undertaken.
The literature review informed the first research objective, while surveys and interviews undertaken using a MSA approach as well as document analysis helped in meeting research objectives two and three. The findings from this qualitative research undertaken as well as cases then formed the basis for the subsequent papers in addressing research objective four. These methods are described in the following subsections.
3.2.1 Literature Review
The narrative literature review undertaken (for instance, see Green, Johnson, & Adams, 2001) informed the first research objective in defining key concepts of the research and examining who is responsible for the delivery of healthy planning and active living initiatives. While generally limited to academic publications (Green et al., 2001), the narrative literature review undertaken as part of this study also extended to grey sources (for instance, see Pappas &
Williams, 2011), necessitated by the need to address the first research objective and the relatively limited academic investigation focused on the implementation of healthy planning and active living initiatives (see Section 2.5). This was also made possible through the adopted iterative research design. The inclusion of grey publications in the narrative literature review is considered appropriate given this had been the case in previous research. For example, scoping reviews (such as Pham et al., 2014) include grey literature in their scope and so do previous literature reviews on healthy built environments (such as Kent et al., 2011). A common criticism of narrative literature reviews is that they reflect the biases of the author, however this can largely be overcome by adopting appropriate literature review techniques (Green et al., 2001).
With this in mind, and as discussed in Chapter 2, the scope of the literature review undertaken as part of this research was limited to physical activity, identified by Jennifer Kent and Susan
Thompson (2014) as one of the three domains through which the built form influences health.
Studies from across the globe were included in the scope, however only materials written in English were considered. Where interview or survey respondents identified materials originating from outside Australia, they were also considered against the literature review scope (refer also Section 3.2.2, below). Secondly, materials identified in the literature review were screened, using the title of the article and abstract of the article. Where papers or grey sources were identified that related primarily to one of the other two domains (healthy eating and connected communities) they were excluded from the analysis. Lastly, materials were assessed in detail for their inclusion in the review. Although unsystematic (Green et al., 2001), narrative literature reviews allow for a summation of existing work in the field and for identification of gaps in previous research (Grant & Booth, 2009). The findings of the narrative literature review are primarily presented in Publication 1 and this exegesis (refer to Chapter 2).
3.2.2 Document Analysis
Iterative analyses of documents and relevant guidance and research relating to healthy planning and active living were undertaken and informed Publications 1-3 in identifying both barriers and enablers. Document analysis was conducted in addition to the narrative literature review (refer to Section 3.2.1, above), and included key documents such as policies that were identified by respondents (in either the surveys or interviews undertaken, refer to Sections 3.2.3 and 3.2.4, below). These documents were analysed iteratively, immediately upon receipt if returned with the surveys or simultaneous to the analysis of the interview transcripts. The scope of the document analysis was limited to Australian sources only. However, it is noted that no internationally sourced material was provided by respondents in surveys or referred to in interviews that had not already been identified through the literature review process (refer to Section 3.2.1, above). The Australian sources were analysed using the coding software NVivo 11, which provides an analytic aid and data management tool in the process of analysis (Zamawe, 2015). Document analysis is an effective way to triangulate data collected via other methods (including those below) (Bowen, 2009), and has been previously been applied in examining both local health policy and the policy setting of LG (Hoeijmakers et al., 2007).
3.2.3 Surveys
Qualitative, open-ended surveys (see Fink, 2003) comprising seventeen questions were distributed to LG practitioners in Australia and informed Publications 1-2 in identifying both barriers and enablers. Surveys present a suitable data collection tool (Neuman, 2014) that addresses the difficulties identified in reaching LG practitioners targeted in this research, including time constraints and no formal way for them to complete the surveys on behalf of their LG. Surveys have successfully been used in reaching Australian LG practitioners previously (Lawless et al., 2017).
Purposive sampling (see Neuman, 2014) was employed whereby Heart Foundation cases (National Heart Foundation of Australia, 2014a, 2015a) and the Healthy Spaces and Places case studies (Healthy Spaces and Places, 2012) were used to identify recent (within the last ten years) examples of successful healthy planning and active living initiatives across Australia. Cases noted in these resources were excluded where they were not recent (i.e. where they were not initiated within the last ten years), where they did not relate primarily to healthy planning and/or active living promotion (as defined for this research in Chapter 2) (e.g. healthy food initiatives) or where a stakeholder other than LG undertook the project. Refer to Appendix A1.1 for a summary of survey participant characteristics and Appendix A1.4 for a sample survey questionnaire.
Data in the form of survey responses were analysed in Nvivo 11 coding software, with sensitising concepts applied in initial analysis (Thornberg, 2012). This helped to avoid imposing the MSA lens on the data. This was important given MSA has not been widely used in the healthy planning and active living field to-date, particularly in Australia (see Section 3.1).
Hence, the suitability of the lens was informed iteratively by the data collected. Using sensitising concepts, where data produce valuable ideas they can be further pursued by the researcher (Faulkner, 2009), especially valuable if an iterative approach is adopted.
Survey data were collected until data saturation occurred (for instance, see Hennink, Kaiser, &
Marconi, 2017), at which point additional information (subsequent rounds of survey distribution) was not considered necessary to provide added results or improve their accuracy
(Carter & Little, 2007). Data saturation was understood to have occurred as (1) the second round of survey results did not add significantly to the number of codes generated after the first round of survey distribution (indicating that ‘code saturation’ had occurred), and (2) the second round of survey results did not add significantly to the understanding of ideas that were coded after the first round of survey distribution (indicating that ‘meaning saturation’ has occurred) (Hennink et al., 2017).
3.2.4 Semi-structured Interviews
In-depth, semi-structured interviews were undertaken with healthy planning and active living advocates and informed Publications 1 and 3 in identifying both barriers and enablers. Purposive key informant sampling was employed initially whereby cases and participants are selected according to, and to meet, the aims or objectives of the research undertaken and to show different aspects of the same issue through the multiple case study approach (Carter & Little, 2007;
Creswell, Hanson, Clark Plano, & Morales, 2007). Participants then contributed to subsequent snowball sampling through suggestions of potential additional interviewees (Palinkas et al., 2015). Interviewees were selected based on their professional employment in Australia, and a demonstrable, recent contribution to both practice and the public discourse on healthy planning.
Semi-structured interviews were combined with MSA to examine local health policy development (Craig et al., 2010), as well as in Australian studies analysing healthy planning at the LG level (Allender, Gleeson, et al., 2009; Allender et al., 2011) and research into barriers and enablers to policy settings to reduce obesity (Dodson et al., 2009).
Semi-structured interviews allow for relevant points to be examined in greater detail as they arise during the conversation and the iterative approach to data collection allowed for early answers to inform subsequent questioning (Edwards, 2013). In all, twenty-eight interviews were undertaken. Refer to Appendix A1.2 for a summary of interview participant characteristics and Appendix A1.3 for a summary of interview questions. Following the initial transcription, transcripts were returned to the interviewee for examination of the data, to ensure accuracy, make any necessary amendments and importantly to allow any further information or knowledge to be added (Baxter & Jack, 2008). Similar to the analysis of the surveys outlined above, NVivo 11 was used to code the transcripts and MSA was applied as a sensitising concept
(Blumer, 1954; Bowen, 2006) and part of an abductive approach to analysis. An abductive approach lies between induction and deduction, and involves moving ‘back and forward between data and pre-existing knowledge or theories’, searching for the best explanations and interpretations (Thornberg, 2012, p. 247). Using an abductive approach to data analysis, theories do not test hypotheses, but instead act as a starting point for patterns and themes to emerge (Kennedy & Thornberg, 2017; Thornberg, 2012). Such an approach allows for and encourages data collection and analysis to occur simultaneously (Carter & Little, 2007) which aligns with the ‘linear but iterative process’ of information gathering and analysis adopted for this study (refer to Table 1) (Yin, 2014, p. xxii).
Similar to survey data as discussed above (see Section 3.2.3), interviews were undertaken until data saturation occurred, at which point additional data (subsequent interviews) were not considered necessary to provide added results or improve their accuracy (Carter & Little, 2007).
The American global health researchers Hennink et al. (2017, p. 605) note that data saturation can be broken down into ‘code saturation and meaning saturation’, whereby ‘code saturation may be reached with few interviews as it provides an outline of the main domains of inquiry, but further data are needed to provide depth, richness, and complexities in data that hold important meaning for understanding phenomena of interest’, or meaning saturation. Results from analysis were therefore considered against these two types of data saturation. Data saturation was understood to have occurred as (1) interviews ceased to add significantly to the number of codes generated (indicating that ‘code saturation’ had occurred), and (2) interviews ceased to add significantly to the understanding of ideas that were coded (indicating that
‘meaning saturation’ has occurred) (Hennink et al., 2017).
3.2.5 Cases
The use of cases allows for current phenomena to be examined (Yin, 1981, 2014) and is commonly applied in healthy planning and active living research (Allender, Gleeson, et al., 2009; Kent et al., 2011) and when a MSA framework is employed (Kingdon, 1984). The cases used in Publication 4 were purposively selected (Carter & Little, 2007) based in one instance on the definition of coastal community and in the other given noted success in implementing walkable and bikeable settings, and also on the author’s familiarity with each example (Palinkas
et al., 2015). Surveys (refer to Section 3.2.3) were based on pre-defined cases identified in the guidance used as sources for their distribution (Healthy Spaces and Places, 2012; National Heart Foundation of Australia, 2014a, 2015a). Examining cases using multiple evidence sources allows for the triangulation of data (Rossman & Wilson, 1985) as generalisable findings can emerge from examining a series of examples (Stake, 1978).
This chapter outlined the research design and theoretical underpinnings of the thesis, including MSA, and the methodologies employed. An overview of the research contribution of each paper with regard to the research objectives of the thesis, as well as the used methodologies was also provided. The next chapter presents a summary of each of the five publications and examines in greater detail their contribution to knowledge generated through this research.