Chapter 4 Methodology
4.4 Analytical methods
This section begins by outlining the results of content analysis of the interviews. Various statistical techniques used for questionnaire analysis are then outlined.
4.4.1 Analysis of interviews
There are different ways of sorting and organising qualitative data. For analysing the interviews, content analysis - an exploratory and inductive coding method - was adopted. The first step was to produce a transcript of each interview. This provides a preliminary form of analysis and the opportunity to engage the data (Dunn 2005). The transcripts were then broken into relevant, distinct words, phrases, and sentences. The goal was for each component to represent a singular reason for why a person valued a place. Phrases were not broken up if doing so changed the meaning.
The next step was to interpret the data and to provide explanations. This requires an analysis and search for themes and categories. Categories were then inductively generated on the basis of their capture of the diverse meanings of responses. A coding guide was developed by which each category was defined with several examples taken from the sample. Manifest or descriptive codes were employed to structure and reduce the data. All data were provisionally coded using this guide. This provided a systematic overview of the data and also enabled me to locate and retrieve issues, topics, information, and themes which did not appear in a sequential manner (Mason 2002). Then a coding structure was developed, whereby codes were grouped together depending on their similarities,
substantive relationships, variations, and conceptual links (Cope 2005). The codes can reflect themes or patterns that are visible on the surface and stated directly (Cope 2005; Dunn 2005; Richards 2005). Patterns and themes were identified that cut across
individual experiences. Statements that have meanings in response to my research
objectives were identified. The whole procedure was interactive; with categories proposed and tested by attempting to code the data, modified in response to noted ambiguities, and retested. Approximately half the data were used to develop the coding scheme. All the themes will be outlined in Chapters 6 and 7.
A major challenge with analysing interview content is to ensure that I as the interpreter am doing it in meaningful and sensitive ways, rather than imposing my own
Ch4 - Methodology
interpretation inappropriately or without justification (Mason 2002). This is particularly difficult because qualitative inquiry is all about how the data are interpreted and making sense of the themes and meanings in the context. To minimise bias and
misinterpretation, the coding categories and allocation of text blocks to these codes were checked by my supervisory team. All the themes that emerged from my interview analysis were presented in Chapters 6 and 7.
4.4.2 Analysis of the questionnaires
Questionnaire responses were coded and initially entered into a spreadsheet. Before commencing, the data were scrutinised for outliers. Respondents who chose more than one answer to a single-choice question or did not answer at all were treated as missing values. The negatively worded item in the place attachment scale was reversed.
The data were then analysed using the Statistical Package for Social Sciences (SPSS). Various methods were deployed: descriptive statistics, factor analysis, and correlation analysis. First, the responses were analysed using descriptive statistics such as
frequencies, percentages, and means to provide an overview of the variables. Factor analysis was then used to reduce the variables in the place attachment scale to a smaller number of underlying latent variables (factors). Factor analysis is a data reduction technique where a large set of variables is reduced to a smaller set without much loss of information (Dawis 1987). Then principle component analysis (PCA) was adopted: variables that are correlated with one another but largely independent of other subsets of variables were combined into factors. The number of factors and their interpretation were then decided. To interpret a factor, one attempts to understand the underlying dimension that unifies the group of variables loading on it (Tabachnick & Fidell 2007). These defined factors underlie answers to individual questions and identify the pattern in the responses to a set of questions (DeVaus 2002; Tabachnick & Fidell 2007). For subsequent correlation analysis, factor scores were calculated by using SPSS. Factor scores are estimates of the scores subjects would have received on the factors had they been assessed directly (Tabachnick & Fidell 2007).
Lastly, correlation analysis was implemented to explore the relations among the
variables and factors. Various methods were deployed. According to the nature of each variable, chi-square test, independent-samples t-test, and one-way between-groups
Ch4 - Methodology
analysis of variance (ANOVA) with post-hoc tests were used to explore variables correlated with senses of place, attitudes to tourism developments, and perceptions of tourism impacts. Some variables were reclassified before applying these analyses. Five continuous variables were reclassified to three or four groups. Total frequency of visitation was reclassified in four groups (once; two to nineteen times; twenty to ninety- nine times; more than ninety-nine times). Total length of visitation was reclassified in three groups (ten years and less than ten years; more than ten years; less than twenty- five years; more than twenty-five years). Number of companions was reclassified in four groups (none; one person; two to four people; more than four people). Length of
property ownership was reclassified in three groups (less than nine years; nine to twenty years; more than twenty years). Age was reclassified in five groups (18~30, 31~40, 41~50, 51~60, >60 years old). Five variables were then reclassified in order to have a greater number of subjects in each category and enable the analysis with other variables.
Frequency of visitation in the past one year was reclassified in four groups (none; once; a few times; more than a few times). Length of each visitation was reclassified in three groups (one day or less; two to seven days; more than seven days). The level of
education completed was reclassified in three groups (Secondary school and under; university; TAFE/Technical college). Employment was reclassified in two groups (recreation and tourism related, others). For Tasman National Park, appropriate potential new tourism operation on private land near to the Park was reclassified in four groups: nature-based lodge; camping (campground with designated campsites and dispersed camping with no or very limited facilities); no development; and other developments (major hotel, small hotel/motel, serviced apartment, bed and breakfast accommodation and caravan park). Appropriate potential new tourism operationin the Bay was reclassified in three groups: camping (as for the National Park); no
development; and other development (major hotel, small hotel/motel, nature-based lodge, serviced apartment, bed and breakfast and caravan park). For some cases, stakeholder cohort type was reclassified in four groups (local community people; non- local visitors; members of environmental groups; and Tasmanian Government staff).
Ch5 - The case study sites