The approach to grounded theory taken here is built upon a constructivist grounded theory (see section 3.3.1), as advocated by Charmaz (2000, 2002), which defines what is happening in the data, rather than discovers it. This calls for intimate familiarity with respondents and their world, thus seeking meaning rather than an objective truth (Charmaz, 2000). As a result, a world is made real, rather than the world, which is situated in the thoughts, the language, and the actions of the participants. It acknowledges that GTM is an emergent and iterative process that requires the researcher’s interaction within the field and the data. Instead of
‘truth’ and ‘validity’, Jamal & Hollinshead (2001) propose the alternative
Figure 3.2 Analytical framework for family holiday behaviours
Parental perspective on Parental (survey) and Whole family experiential dimensions (Figure 3.3) family holiday behaviour familial (interviews)
perspectives on the
(Survey) (Interviews)
- Perspectives - Travel patterns
- Family holiday definition - Temporality
- Motivations - Themes
Figure 3.3 Analytical framework for whole-family experiential dimensions
THEMES
TEMPORALITY Post-
PERSPECTIVES
On-
Pre-Family time Own time
Generation
Gender
Group dynamics
Table 3.11 Development of interview questions regarding themes through the research phases using the GTM
* √ theme present; X theme not present
measures of transparency and reflexivity for fifth moment qualitative research.
For this reason, details are provided here on how emerging themes necessitated a change in interview strategies in each phase of analysis while also mapping the influence of literature to the development of theoretical concepts (see Draucker et al., 2007). The temporal and conceptual development of the main theoretical themes to emerge through the GTM is explained here to provide transparency to the analytical process. Reflexivity will be elaborated on in the next section.
Central features of the GTM include the method of theoretical sensitivity, theoretical sampling, constant comparative analysis, and theoretical saturation (Glaser & Strauss, 1967). Theoretical sensitivity draws upon a wide range of literature also outside the field of study to inform the analytical encounter with the data (Dey, 2004; Glaser, 1987). This was acquired by conducting an extensive literature review that encompassed the wider family literature which led to the conceptual and methodological framing of this research. The analysis of data also happened alongside emerging tourism literature rather than just resulting from it (see section 2.6.3).
Theoretical sampling, the process of data collection directed by evolving theory, is preceded by selective sampling. Selective sampling is the identification of populations and settings prior to data collection, which in this study was the survey distributed through five schools and resulting in 110 parental responses.
These data was entered into SPSS and frequencies, percentages, cross-tabulations, and comparisons of means were conducted in line with the smaller sample size.
For this the quantitative data were broken into categories (e.g., accommodation type) which allowed comparisons between the different holidays (e.g., past and upcoming travel) and resulted in summary tables (see Table 4.1) and graphs (see section 4.2) which informed the family holiday definition.
The survey also contained open-ended data on motivations and family holiday characteristics which was manually coded before being entered on SPSS alongside the other quantitative data. For example, one respondent’s motivations:
“time with children; break from routine – everyone can relax” were coded for the three themes of ‘time together’, ‘break away’, and ‘relaxation’ which resulted in a
variety of 15 motivations emerging for family travel. Another example is for one respondent’s family holiday definition: “One where all members of the family spend time together doing things other than they would normally do – usually away from home” was coded for the three characteristics of ‘time together/whole family’, ‘different activities’, and ‘away/break from home’. This resulted in a variety of 10 different characteristics to a family holiday definition. Thus, coding according to the GTM occurred before SPSS was used as a quantification tool rather than an analytical tool. This allowed the codes emerging from the survey to be counted to create a definition of family holidays based on prevalence and to present them in graphs (see Figures 4.4 and 4.5). The resulting main themes then informed the first interview schedule and linked the open and closed questions in the survey to the interviews with the 10 families. These initial themes were then tracked over time and further refined in the three interview phases concurrent with the analysis phases (Table 3.11).
The data collected via selective sampling refers to a tentative categorisation from which to begin theory development (Draucker et al., 2007). Sampling was sequential beginning with selective sampling and moving into theoretical sampling when themes began to emerge, such as the theme of togetherness that emerged in the survey and then evolved in the interview phases. Theoretical sampling was used by modifying interview schedules to gain specific information regarding an emerging theme as the study progressed. For example, fun emerged as a theme in phase 2 analysis and was followed up in the subsequent interview phases. Table 3.11 illustrates how interview schedules were modified and demonstrates how additional questioning increased the complexity of the analysis.
Theoretical sampling is closely tied to grounded theory coding which is based on at least a two-step process: (a) initial or open coding, and (b) selective or focused coding. For example, the initial coding of time with non-family became part of the main theme of own time. Through the process of constant comparison analysis, the concurrent collection and analysis of data informing the next phase, these initial codes were collapsed into core themes and integration of the theoretical framework was achieved. This means that coding progressively distils events and meanings without losing their essential properties (Charmaz, 2002). The third step
of coding as introduced by Strauss & Corbin (1990), axial coding, was deemed unnecessary because it adds complexity with little benefit to the analysis (see Charmaz, 2000; Glaser, 1992). Theoretical saturation is presumed to be achieved when new data fit into the themes already devised (Morse, 1995). In practice, saturation tends to be an elastic category that contracts and expands to suit the researcher’s definitions rather than any consensual standard (Charmaz, 2002).
After the comparative analysis of the four phases of research was completed, all the data fitted into the theoretical framework (see Figure 4.7) and theoretical saturation was deemed achieved.
Analysis was based on verbatim transcriptions by me of the 148 recorded interviews. The coding process was carried out manually in that data were initially coded by reading through the transcripts several times while making notes on a large piece of paper which was then sorted into themes and integrated into a theoretical framework. According to Lincoln (1998) in Charmaz (2000), part of interpretive work is gaining a sense of the whole body of data, all interviews and all stories. For this it was necessary to have the data organised initially by the different phases and then by family while simultaneously planning to assemble the parts in the form of themes. Only after the core themes were established was selective coding applied using the computer program NVivo 8. This program proved especially helpful with managing the volume of data. Other advantages of computer coding included the ability to do multiple searches using more than one code/theme simultaneously and according to the perspectives (e.g., generation).
The code and retrieve method supports the emergence of theory by searching the data for specific codes/themes or attributes and assembling ideas (see Charmaz, 2002). The management of the parts in NVivo was, thus, superseded by nuanced manual interpretive analysis which signifies a more holistic approach to the GTM.
The analysis can be illustrated using an example of one initial theme: VFR (Figure 3.4). In phase 1 survey VFR was included in the form of closed and open ended questions. Phase 1 analysis encompassed the quantitative data on VFR and the qualitative coding of time with friends and relatives for every participant mentioning friends or relatives (definition and motivation) including visiting or spending time with extended family (see sections 4.2 and 4.3). After entering the
codes in SPSS and comparing their frequencies with other codes it became a theme. This theme of importance of friends and relatives was included in phases 2–4 of interview questions after it continued to be relevant in the analysis phases (Table 3.11). Larsen et al.’s (2007) work helped the naming of this theme in phase 3 of analysis into time for (re)connecting and social obligation. The comparative analysis allowed these themes to be increasingly refined over the different phases and led to social support emerging as an additional theme from the data which was absent in the literature. These themes were then collapsed and integrated into the sub-theme of social connectedness/VFR which is part of the main theme of family time (see Figure 4.7). This also contributed to the development of the sub-theme of cooperation which is part of the internal dynamics. It became apparent that these themes transcend the temporal dimension of the phased approach.
Through the longitudinal approach different themes emerged that did not correspond with the phases of methodology but instead had an all encompassing presence like change of routine and relaxation. This meant that the temporal element was not as strong in the analysis because family holidays were part of a familiar routine or tradition (see section 6.7). Themes emerged in the analysis that were then taken up in the modified interview schedules (Table 3.11). While some themes were named after literature that was emerging at the time, e.g., generativity (Shaw et al., 2008), most themes were named by the participants (e.g., the two main themes) or directly emerging from the data (e.g., peer time).
The comparative analysis of all four phases of research resulted in collapsing of the initial themes into core themes and integration into the theoretical framework (see Figure 4.7). This meant a reorganisation of the themes into family time, own time, and internal dynamics, development and naming of new sub-themes such as cooperation as well as refinement of existing sub-themes such as change of routine. Thus, the presentation of the data in the findings chapters is now organised along the main themes and sub-themes (see section 4.5). The thematic or theoretical framework that resulted from the iterative grounded theory process of refining, extending, challenging or superseding the extant concepts is then used to illustrate key findings regarding the different generational, gender and group dynamic perspectives in chapter 7.
Figure 3.4 Illustration of the development of initial theme VFR through the GTM
Phase 1 VFR (Survey)
Phase 1 VFR time with friends & relatives