From a pragmatist-inspired approach, performing the analysis is equivalent to asking the question “What makes this phenomenon possible?” and seeking to produce an answer by offering a hypothesis. In this study this means seeking answers in how families cope in everyday life through the use of mobility. The process of analysis in this study is an adaptation of Kvale and Brinkmann’s (2009: 217-8) overall model for analysis. As such, the analysis actually begins during the interview, in which both the respondents and the interviewer engage in initial understanding and meaning condensation in situ. Hence the interview method applied in this study aims for a “self-correcting” approach in which the interviewer actively seeks to formulate and test initial interpretations of the respondents’ statements during the interview (Kvale and Brinkmann 2009: 217-8). This was followed by a more formal ex post analysis and interpretation of the interviews, which was eventually fed back and presented to the families for validation and further elaboration and comment (in the second round of interviews, P2). This section will elaborate on the ex post analysis that was performed after both interview sessions (P1 and P2).
As Kvale and Brinkmann (2009: 214) point out, there are no standard techniques for qualitative text analysis. Instead we should think of techniques for analysis as tools with different capabilities that may be useful and relevant in certain situations and contexts. Drawing inspiration from both the pragmatist abductive attitude, presented in the previous chapter, and grounded theory (Charmaz 2006, Strübing 2007), the analysis in this thesis is based upon the analysis techniques of coding and categorisation (Saldaña 2009, Kvale and Brinkmann 2009). With its use of systematic coding and categorisation for making sense of and creating meanings in the empirical interview material, the analysis can be seen as an iterative-cyclical process in which the analysis addresses the concrete empirical material descriptively and gradually moves towards the formation of more abstract concepts and understandings (Kvale and Brinkmann
2009: 260). The aim of this process, illustrated in figure 10, is to construct concepts that might eventually aid formation of new theory. This does, however, not mean, the process is or should be a-theoretical. Despite finding inspiration in grounded theory, the analysis performed in this thesis is pragmatically utilising existing theory as tools facilitating the coding and categorisation in the process of analysis.
Raw data
Code
Code
Code
Code
Code
Code
Particular General
Real Abstract
Category
Category
Theme/
Concept Theory
Sub – category Sub – category
Figure 10: Formal model of code-to-theory in qualitative inquiry, adopted from Saldaña (2009: 12)
CODING
As the first step in processing the raw material, the interviews were listened to, transcribed and re-read several times. In this way the researcher gained familiarity with the empirical material. Through this intensive process, initial ideas about and interpretations of the empirical material started to emerge. From this open explorative process, the analysis turned to the more systematic technique of coding. Coding is a de-contextualising analytical procedure that is a first step in organising and making sense
of data; or, as Amanda Coffey and Paul Atkinson put it, coding enables “us rigorously to review what our data are saying” (1996: 27). The material is coded by creating a series of codes or labels, often single words or short phrases, which is used to sort and organise the material. Coding, however, is more than instrumental labelling—coding is analysis. Johnny Saldaña writes that coding is “a heuristic (from the Greek, meaning
‘to discover’)—an exploratory problem-solving technique without specific formulas to follow” and that “[coding] leads you from the data to the idea, and from the idea to all the data pertaining to that idea” (2009: 8).
In this study the transcripts were systematically processed through several iterations of coding cycles in which multiple codes were created, refined and sometimes removed.
Codes were developed from the research ambition and scope of the study, inductively from the material itself and the respondents’ statements, and equally through theoretical readings in which the empirical material was approached through the use of various theoretical perspectives (see chapters 4 and 5 for elaboration of the theory as well as the analytical perspectives that have guided the analysis and coding process). Thus the coding process was performed using several “coding filters” (Saldaña 2009: 6), each one representing an analytical lens foregrounding certain factors of particular interest and guiding the interpretation of the empirical material towards them. Hence this systematic coding process allows for the same text to be multi-coded, embracing the complexity of the empirical material by facilitating multiple readings simultaneously.
In considering this process, it is important to stress that “since we bring ‘our subjectivities, our personalities, our predispositions, [and] our quirks’ to the process”, all coding is based on the researcher’s judgement (Saldaña 2009: 7). This illustrates the pervasiveness of normativity, pre-understandings and prejudices that were discussed in the previous chapter (see section 2-3 and 2-4) as a profound condition of the research practice that greatly impacts and shapes the analysis process. For instance, developing a code, choosing what qualifies as belonging to that code and establishing how codes are related to each other are all decided entirely by the researcher.
The majority of the coding process was performed in the computer software program NVIVO (see nvivo file in the appendix, Folder D). The use of NVIVO should be understood not as analysis but, as Coffey and Atkinson (1996: 172) state, as “analytical support”, a tool for preparing and organising data for analysis. Hence in the use of computer-assisted analysis it should be stressed that the computer is not auto-performing the analysis. It is still the researcher who makes decisions on what codes to establish, what to put in them, which leads to follow and so on, and NVIVO is merely a useful tool for doing this.
CATEGORISATION
Through the use of coding, the analysis seeks to find patterns or associations in the
empirical material. By juxtaposing, comparing and associating individual codes, the researcher searches for patterns or clusters of codes that help to make sense of the data. Saldaña offers this explanation:
I advocate that qualitative codes are essence-capturing and essential elements of the research story that, when clustered together according to similarity and regularity—a pattern—they actively facilitate the development of categories and thus analysis of their connections.
(2009: 8)
Hence codifying, the systematic arrangement and ordering of elements in a hierarchy or a system, is a powerful and flexible heuristic sensitising tool in which elements can be brought together in different combinations and compositions. Andrew Abbott describes this meticulous process through the metaphor of decorating: “decorating a room; you try it, step back, move a few things, step back again, try a serious reorganization, and so on” (2004: 215). In the analysis multiple iterations of cutting and pasting codes, grouping and regrouping them into categories based upon shared characteristics or family resemblances (decided upon by the researcher), were done in the initial search for patterns in the data and the formation of tentative understandings and interpretations. (An overview of the initial state of codes and categories in the analysis can be seen in Wind (2012), appendix Folder A) As Saldaña notes, this categorisation process is facilitated through the researcher’s “tacit and intuitive sense”
(2009: 9) of what data make sense together in which the researcher is also drawing on theoretical resources and knowledge. This decision process can be approached in many ways, but was in this study performed by looking for patterns of:
• Similarities: Looking for things that happen in the same way., For instance, in the empirical material, one of the parents in each of the families in the study save one has reduced his or her working hours while the children are young as a way of making everyday life function.
• Differences: Looking for things that happen in different ways. For instance, the families use very different combinations of transport modes to get around even though they live in comparable neighbourhoods.
• Frequency: Looking for recurrences, things that happen often or seldom. For instance, the frequency of family activities, activities for which family members are together, varies across the weekdays, often peaking at the weekends.
• Sequence: Looking for specific orders or sequences of how things happen. For instance, many of the family members perform very particular and scripted mobility practices in which the sequence of actions often serves both a ritualistic as well as a practical purpose.
• Correspondence: Looking for how events are interrelated and affect each other.
For instance, in the empirical material the family’s everyday mobility is closely configured in relation to the events of bringing and picking up young children to and from day care and/or kindergarten.
These search heuristics were primarily deployed in the interview material, but other empirical material, such as observations of group dynamics and interactions during the interviews, the socio-economic characteristics of the families, ACTUM data and the experiences and observations from the mobile field studies, were used to support and qualify the presumed associations and patterns in the analysis. For instance, as will be argued in the analysis (chapters 7 and 8), the association between the family members’ mobility practices and the affective atmospheres of moving and emotional coping in everyday life was suggested in the interview material but further supported by the mobile field studies. Hence categorising is an abductive process in which the researcher creatively guesses at or suggests associations and patterns, which are then measured by their ability to make sense of the data. This guessing is of course not entirely unqualified as the researcher formulates hypothesis based upon other theories and an existing web of knowledge. In this study, through several iterations of re-coding and re-categorising, the two main categories of making everyday mobility and performing everyday mobility where established, each consisting of several sub-categories based on several codes (see figure 11 for an example from the category of making everyday mobility).
The de-contextualised codes are re-contextualised by condensing clusters of codes that are perceived as belonging together into sub-categories. In this study, as illustrated in the figure above, the sub-categories were formed from the clusters of codes. Themes and concepts are “outcome[s] of coding, categorization, and analytic reflection”
(Saldaña 2009: 13) and can be understood as tools aiding the analysis process in moving away from the concrete and particular of the empirical material and towards the more abstract and general. The three sub-categories illustrated in the figure above were used to formulate three key themes of instrumental mobility, mobility as social spaces and mobility as emotional management. Interrelated with these themes, higher-level concepts such as mobile care, mobile togetherness and mobile in-between were constructed (see chapters 7 and 8 for elaboration of these concepts). It should be noted that the generative analysis is not a linear but an iterative-cyclical process in which concepts are gradually developed, refined and sometimes dismissed through iterations of abductive generation, deductive testing and inductive verification with reference to the empirical data (Stjernfelt 2007: 333, Strübing 2007: 565). This process is illustrated in figure 12.
Evolving theory
Abduction
Data Data Data Data
Hypothesis Hypothesis Hypothesis Hypothesis
Abduction Abduction Abduction
Deduction/induction Deduction/induction Deduction/induction
The concepts, condensations of codes and categories produced in this analysis process are to be understood as tools which can be used for interpreting and understanding patterns or associations not only in the empirical data but also on a more generally.
By “inferring transfer” (Saldaña 2009: 13), by which it is meant that what is observed in a limited number of instances at particular times or sites might also be observed in comparable instances at other times or sites, however without implying determinism.
For instance, in the empirical material family members state that they often utilise everyday mobility for the recreational purposes of relaxation or listening to music.
It seems reasonable to believe that this is a general phenomenon occurring in much of everyday mobility, including outside of the 11 families in the sample. Therefore in the study the concept of mobile in-between is proposed as a way of understanding this phenomenon, and by inferring transfer this concept is lifted from the particular
Figure 12: Iterative-cyclical analysis process, adopted from Strübing (2007:567) )
occurrence towards a more general level. This does not in any way imply that the theoretical concepts produced in the analysis are universal laws applicable to all situations. On the contrary, we can easily dismiss the notion that all everyday mobility practices are enjoyable and have recreational elements; even the same mobility practice might one day be a recreational in-between and the next a tedious necessity.
Nonetheless, it is still reasonable to presume that other mobility practices may share this recreational element, even based on such a small sample as the one used in this study.
3-7: VALIDITY OF KNOWLEDGE AND ETHICAL