CHAPTER 3 Research design
6. Qualitative content analysis
Content analysis of planning documentation (namely regional policy statements, district plans, consent applications, Environment Court applications/evidence/ decisions, project documentation and other Council reports) is relevant in this research, in order to record the existing land use policies that are in place for particular hazards in specific areas. According to Yin (2003) and Sarantakos (1998), the strengths of using
documentation in research is that documents are:
• Stable - they can be repeatedly reviewed;
• Unobtrusive – they are not created as a result of the case study;
• Exact – contains exact names, references, and details;
• Broad in coverage – long time span, many events, and many settings.
While documentation does have weaknesses (e.g. access, biased selectivity, and reporting bias (Sarantakos, 1998; Yin, 2003)), these limitations are not considered to be as prominent in this study because regional policies and plans are highly
formulation. Consent applications and Environment Court decisions are likewise accessible and have been through a democratic process.
Once the units of analysis have been obtained (in this case, coding of content within policies, plans, consent applications and associated documentation and decisions), qualitative content analysis involves identifying and evaluating the items that appear to be theoretically important and meaningful. The researcher then relates them back to the central question of the study (Sarantakos, 1998). The logic of using content analysis for published material is the same as the logic for other kinds of research — one must have a hypothesis or research objective; variables are selected; a way of measuring the variation in the variables must be achieved; then a way of reporting the findings must be presented (Bouma & Ling, 2004).
For the process of undertaking content analysis for the case studies, a number of existing methodologies were reviewed, both international and New Zealand-based (Berke, Crawford, Dixon, & Ericksen, 1999; Berke, Roenigk, Kaiser, & Burby, 1996; Ericksen, et al., 2003; IBHS, 2001). Within the New Zealand context, Ericksen et al. (2003) provide a methodology for evaluating plan quality under the Resource
Management Act 1991. In the light of an absence of international literature on robust methods for evaluating plan quality, Ericksen et al. (2003) developed their own. Each related question was coded (0, 1 or 2) and then used to create indices for each of the eight plan quality criteria or principles they used. While this methodology is specific to the New Zealand RMA context, it looks at the overall plan quality and is not natural hazard specific. However, the U.S. Institute for Business and Home Safety has produced a questionnaire for planners called the ‘Community land use evaluation for natural hazards’ (IBHS, 2001). Similar to the Ericksen et al. (2003) methodology, it has sections with specific questions, which are scored and rated.
For the purpose of this research, none of the evaluation criteria from IBHS (2001) or Ericksen et al. (2003) met the requirements of the research questions. Therefore new criteria were developed which would assist in answering the research questions in relation to plans that address natural hazards. A protocol was developed for coding the content to ensure that any plans being analysed were coded the same way.
Three plans (the Waikato CDEM Group Plan, WRC RPS, and the TCDC district plan) were analysed in detail to ascertain what natural hazards were included. This was then cross-tabulated against interview responses to a question asking which natural hazards were a concern to them for the Thames-Coromandel district (refer Chapter 6). A
relatively simple analysis was undertaken on the three plans to confirm what natural hazards were included in each plan. A table was constructed that highlights which hazards were included in each plan, and cross-tabulated with the responses from the interviews (refer Table 5.2, Chapter 5).
For the analysis of the interview transcriptions, Atlas.ti, a computer-based analysis tool, was used and is discussed in the following sub-section.
6.1 Computer aided analysis
Atlas.ti is a computer package specifically aimed at providing tools for qualitative data analysis. While there are other computer aided analysis packages available (e.g. NUD*IST, MAXqda and N-Vivo), Atlas.ti was selected for the analysis of the interview transcriptions as it was in use at my workplace at GNS Science. As such, using Atlas.ti kept other research analysis within the social science team in a consistent and
compatible form.
Computer aided analysis was a method I had not used before, in turn up-skilling myself in using a computer package that assists in qualitative analysis. The key focus of Atlas.ti for this research was the coding of qualitative interview transcripts. As there were 23 transcripts to analyse, Atlas.ti provided the opportunity to use a computer aided analysis tool. The following codes were used for key themes from the transcripts:
Avoidance Balance Balance: equal
Barrier Benefits Change required
Civil Society Communication Community responsibility
Consequences Cost Duty of care
Economic considerations Emergency management Finger point central govt
Finger point council Finger point individual Finger point insurance
Finger point local Finger point market Finger point national
Governance Increase risk Individual responsibility
Individual risk Innovative Insurance
Land use Leadership Legislation
Liability LIMS/PIMS Market
Market responsibility Mitigation Monitoring
National guidance No increase in risk Options
Personal responsibility Planning Pre-event planning
Relocation Residual risk Risk acceptance
Risk awareness Risk magnitude Risk management
Risk reduction Site specific State
State responsibility Sustainability Timeframes
Tipping point Uncertainty Urban vs rural
Worsen the risk
Once all the transcripts were coded, the analysis of the data could begin. The key themes from the interview questions were extrapolated into query reports. These were produced for the following coded themes: innovative, change required, leadership, barrier, risk reduction and mitigation. These themes were chosen as they either related directly to a question, or were a key result of a question.
From the query reports, further analysis was required without the aid of Atlas.ti to locate quotes that could be used to reinforce findings from the literature review and critique of the legislative provisions for hazard management (refer Chapter 5). While a useful tool, several limitations of computed aided analysis became apparent, as outlined below.
6.1.1 Limitations of approach
As a first-time user of computer aided analysis and Atlas.ti, limitations were
experienced in the use of this tool. Many of these issues faced are a reflection of my lack of experience in using a package such as this:
• It took time to learn the program, as there are no formal training courses provided
in New Zealand;
• Learning the program was via a 410 page handbook. It was often difficult to
reconcile the examples provided to my research needs;
• Some interviews were over two hours long, and these took a lot of time to code;
• From the codes presented above, it is acknowledged that I over-coded the data
due to the wealth of information that was available. While I coded all the themes being discussed, I only needed to focus on those specific to the questions answered.
• Once query reports were available following coding, they still needed further non-
In spite of these challenges, I would use Atlas.ti again to assist with analysing large volumes of qualitative data. With further use and understanding of the tools the package has available, I would become more efficient in its use.