Data analysis

38

Table 1-9 Data volume by actor type and language composition Actor type Data volume

(pages)

Language composition (%) Chinese Japanese English

Event organizers 68 0 82 18

Event stakeholders 47 10 89 1

International participants 142 99 0 1

Total 257 56 38 6

Note: Fata volume is measured approximately by A4 pages (font size 11pt, single spaced).

the English information distributed by event organizers. As for event stakeholders, the 10% of Chinese data came from a travel agency operating in Taiwan, and the 1% English data came from English-speaking volunteers. Since almost all the international participants covered in the research sample were from Taiwan and Hong Kong, data gathered from them were mostly based in Chinese, with a small percentage of English data came from an Australian participant.

39 Table 1-10 Coding methods applied in data analysis

Methods Description Applications in this study

Process coding ⚫ A code connotes actions taken by the person.

⚫ Useful in capturing how the person acts or interacts to reach a goal or solve a problem.

⚫ To identify: (1) activities (actions and interactions) engaged by actors to cocreate value, (2) factors that enabled/disabled the activities.

⚫ Example: the value cocreation activity of “receive information” was captured by the phrase “the author told us about Tohoku Food

Marathon.”

Values coding ⚫ A code reflects the person’s values, attitudes, and beliefs.

⚫ Useful in capturing the person’s thoughts formed, perpetuated, and changed through social interactions and institutions.

⚫ To capture how the person values the experience.

⚫ Example: “establish destination and event image” was captured by the phrase “many gourmet foods” and

“really impressed with.”

In vivo coding ⚫ A code corresponds to a word or short phrase used by the person in his or her actual language.

⚫ Useful in capturing how the person evaluates an experience.

⚫ As a supplement to the above two methods to identify a word or phrase that was indicative of the person’s own perception on value cocreation.

⚫ Example: the phrase “let our

colleagues become aware of the new values being created” was coded to show value formation.

Note: Summarized from Saldana (2016) and adjusted to fit with the purpose of this study.

methods to identify a word or phrase that was indicative of the person’s own perception on value cocreation (Saldana, 2016).

Figure 1-1 illustrates how open-coding was done, based on a statement from an interview with an international participant. For illustrative purpose, the original Chinese language is shown here in English. As the example shows, the phrase “was at a book (about running marathons overseas) launch party” illustrates that the participant searched for information about running marathon overseas. Similarly, the phrase “the author told us about Tohoku Food Marathon” illustrates the participant received information. Therefore, following the open-coding technique, “search information” and “receive information” were marked as codes.

Likewise, the phrases “took the idea from Medoc Marathon”, “many gourmet foods”, and

“revitalize the local economy”explained why the participant was “really impressed with Tohoku Food Marathon”. Together they illustrated the value being cocreated and were therefore marked as the code “establish destination and event image”.

40

Figure 1-1 Example of open-coding

Codes were iteratively refined during the coding process as new codes were created and other codes modified or merged with existing codes. As shown in Figure 1-1, after several rounds of coding and comparison of the data, “search information” and “receive information”

were found to coincide with each other and therefore the two codes were combined as “search and receive information”. As the coding process went on, a codebook was maintained to specify the definition and properties for each code.

The second step involved a process of pattern-coding (Miles et al., 2014). Large amounts of data summarized in the previous step of open-coding were grouped into a smaller number of categories. The process “pulls together a lot of material from first cycle coding into more meaningful and parsimonious units of analysis” (Miles et al., 2014, p. 86). In this stage, concepts that share a certain property are organized into a tree structure. Categorization of codes makes it possible to reduce the number of units a researcher is working with and to clarify the main themes emerging from the data (Strauss & Corbin, 1998).

Figure 1-2 illustrates how pattern-coding was done; i.e. similar codes were grouped into categories. The example in Figure 1-1 has shown how the code “search and receive information” was identified. Likewise, another code “process and share information” was identified in the open-coding process. Both “search and receive information” and “process and share information” were carried out by international participants, and were observed in their interactions between and among other actors. By comparing the properties and characteristics of both codes, their commonalities were identified in how international participants circulated

41

Figure 1-2 Example of pattern-coding

the information. Moreover, such information was guided by the value propositions communicated by the event organizers. Following pattern-coding technique, the codes

“search and receive information” and “process and share information” were grouped under the category “acquiring and propagating value propositions”. Similar procedures were carried out to group remaining codes into the corresponding categories. Each category represented a different value cocreation practice.

The third step involved comparing and making inference of interrelationships among categories of value cocreation practices engaged by different group of actors. In the first and second step, open-coding and pattern-coding were carried out respectively on data sets that correspond to event organizers, event stakeholders, and international participants. In the third step, findings from the three groups of actors were integrated to present a holistic picture of value cocreation.

Coding was done by using a qualitative data analysis software NVivo 12 (released in 2018).

Applying this CAQDAS (Computer Assisted Qualitative Data Analysis Software) facilitated the analysis process in several ways. In particular, it was used for storing and maintaining the data, coding, linking codes and text segments, editing and refining codes, creating analytic memos, and for visual presentation of the data and findings (Miles et al., 2014).

Continuing from the example of open-coding and pattern-coding shown in Figure 1-1 and 1-2, Figure 1-3 illustrates how coding was done in NVivo 12. Open-coding was done by selecting the texts to be coded, and then specify the code name. In the example shown in

42

Figure 1-3 Example of the coding process in NVivo 12

Figure 1-3, the highlighted texts were coded as “search and receive information”. Next, pattern-coding was done by pulling the related codes under a tree structure. In the example shown in Figure 1-3, “search and receive information” and “process and share information”

were grouped under the category “acquiring and propagating value propositions”. For simplicity and clarity, the screenshot of NVivo in Figure 1-3 was edited to show step 1 and 2 in one place. In practice, the two steps were done separatedly.

(2) Data analysis quality assurance

The quality of data analysis was considered from the following four criteria: construct validity, internal validity, external validity, and reliability.

Construct validity refers to the accuracy with which a study’s measures truly reflect the concepts being studied (Yin, 2014). Triangulation of data collected from multiple sources (interviews, participant observations, SNS and blog posts, and archival data) allowed for an in-depth study of the case and was expected to increase the construct validity of the research findings.

Internal validity refers to the extent to which the cause-effect inferences made by a study are trustworthy (Yin, 2014). Triangulation from different perspectives (event organizers, event

43

stakeholders, and international participants) was expected to increase the internal validity of the research findings.

External validity refers to the extent to which the conclusions of a study can be generalized to other contexts (Yin, 2014). Since the findings and discussions of this research were based on a single case study, its generalizability cannot be fully discussed as with researches based on multiple case studies. Building on the results of the current study, the author expects to extend the current research topic in future projects. With respect to the generalizability of the current research, rich descriptions of value cocreation activities are presented for the readers to judge if the findings are transferrable to the circumstances they wish to investigate. Such transferability is one of the four measures proposed by Smith (2018) to judge generalizability in qualitative research.

Reliability refers to the consistency and repeatability of a study (Yin, 2014). In this research a number of tactics were used to ensure reliability in the data analysis procedure. First, the entire set of data was kept in a case study database for future reference and examination.

Secondly, concurrent with the data analysis process, an analytic memo was kept to document the author’s deliberations on code selection, how the inference was taking shape, as well as emerging patterns and categories (Saldana, 2016). Thirdly, a codebook was maintained to document the definition and coded texts for each code. The codebook was reviewed by experienced researchers. Specifically, coded data in Japanese (mainly applied to chapter 2 and 3) were reviewed by a Japanese native who holds a PhD degree in tourism; coded data in English and Chinese (traditional Chinese used in Taiwan and Hong Kong; mainly applied to chapter 4) were reviewed by a Taiwanese native who holds a PhD degree in human resource management. Several rounds of commenting, revising, and discussing were carried out between the author and the two reviewers to reach a final set of reconciled codes. This peer reviewing process functioned as a way to validate the reliability of analysis (Saldana, 2016).

This chapter has provided an overview of the research project presented in this thesis.

Section 1.1 set forth the background and relevance of the research topic. Section 1.2 and 1.3 presented an overview of the existing literature on sports tourism and value cocreation, as well as identified gaps in the literature that motivated the research objectives. Section 1.4 elaborated on the research methods applied to investigate the interactions among event organizers, event stakeholders, and international participants involved in the value cocreation phenomenon as occurred in the research context: Tohoku Food Marathon. The following chapters will present empirical evidence of value cocreation from the perspectives of event organizers, event stakeholders, and international participants.

45

2. Value Cocreation from the Perspective of