國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Practical application of GT methodology. In practice, the methodology involved gathering, coding and analyzing data in an iterative manner. In this way, initial analysis can guide subsequent data gathering and analysis in a constant comparative process as conceptual categories are developed (see Figure 3).
Figure 3. Coding & analysis process. Adapted from Charmaz, 2006; Corbin & Strauss, 2008; Hahn, 2008; Strauss & Corbin, 1998.
The steps in this process for the present study are summarized in more detail below.
1. Preliminary research: literature review for sensitizing concepts, developing research problem and exploratory questions
2. Phase 1 Data Collection and Analysis:
a. Data collection: (participant observation and in-depth interviews) b. Initial (open) coding of transcripts and notes in CAQDAS
c. Memoing, refining of codes and creation of categories
Theoretical+
categories++
categories*sorted,*
related*
Intermediate,+axial+
coding+
codes,*categories*
related*by*incidents,*
interactions+
Re5ining+codes+&+categories:+
memoing,*sorting*and*combining*
Initial,+open+coding:+
raw*data*(transcripts,*notes)*"broken*up"*and*
labeled++
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3. Phase 2 Data Collection and Analysis
a. Continued data collection (interview guide adjusted based on emerging categories from Phase 1)
b. Memoing, refining of codes, categories
4. Intermediate (axial) coding: using context and comparison to develop conceptual categories around particular incidents and interactions
5. Refining categories: sorting, diagramming, and relating of codes and categories in order to develop primary, theoretical categories
6. Writing out the results and findings: using the completed analysis of codes,
categories, and relationships to address the exploratory research questions and present any additional findings
The goal of this study was exploratory investigation, and therefore in this case elements of GT methodology were used to suggest key processes for interpretation and further study. Some GT studies aim to develop a full, substantive theory explaining the phenomenon that may include a unifying “core” category, however this was beyond the requirements and the scope of the current study.
Epistemological and ontological orientation. In philosophical orientation this study is sympathetic to the critical realism proposed by Maxwell (2012). This is not necessarily the critical realism of one of its progenitors, Roy Bhasker (1975/2008), though he is one of the proponents that Maxwell (2012) considered in describing how critical realism preserves the distinction between ontology and epistemology, utilizing an epistemological constructivism while maintaining a “critical” ontological realism. This position is related to what Schwandt (2000) described as “‘weak’ constructionism” and what others have described as the ontology of post-positivism (though not its epistemology) (Guba & Lincoln, 1994; Lindlof & Taylor, 2011).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The implications following from this position are that there is a world that exists independently of our perceptions and theories, however all of our descriptions of it are constructed via language, meaning-making, and social context (Oliver, 2012). This study similarly takes a constructivist stance in acknowledging reflexivity—the effects of researcher-participant interactions on the construction of data—as well as relationality—
aspects of power and trust in researcher-participant relationships—in the interviewing and data-gathering process (Hall & Callery, 2001). However, critical realism holds that social constructions are another part of the reality we experience, which is complex, multi-layered and multi-causal (Oliver, 2012). Causality in critical realism is not a linear cause and effect relation, rather it is contextual and emergent, operating as generative mechanisms and tendencies with effects on multiple levels (Alvesson & Sköldberg, 2009). Though it is impossible to obtain a certain, “objective” understanding of the world and all knowledge is incomplete and fallible, some empirically-supported theories and descriptions may be found to work better at making phenomena intelligible than others, in a pragmatist manner
(Maxwell, 2012; Oliver, 2012).
The notion of a world existing independently but against which human perceptions are constantly tested and revised is also similar to the sense in which Blumer (1980) wrote of his (and Mead’s) stance regarding realism and idealism (or constructivism). Blumer’s
symbolic interactionism (SI) has in fact been interpreted as an implicitly critical realist stance (Maxwell, 2012). Symbolic interactionism holds that people act based on the meanings of things, these meanings are derived from social interaction, and they are modified through an interpretive process (Blumer, 1986). The structure or organization of human groups exists as a process of fitting together of actions (Blumer, 1986).
The emphasis on social interaction for meaning generation and on action in the process of organizing, as well as Blumer’s emphasis on closeness to the social phenomena
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
studied, all complement the topic and methodology of the present study. In fact, the symbolic interactionism of Blumer and the pragmatism of Mead that is one of its main influences are both foundational to the development of GT (Corbin & Strauss, 2008; Heath & Cowley, 2004; Puddephatt & Charmaz, 2006). Important beliefs implied by these influences are: 1) that phenomena are partly determinable using naturalistic analysis, 2) that experience is located within a larger framework and context, meaning is created through social interaction, 3) there is no individual/collective duality in the generation of knowledge, and 4) there is no need for a separation of knowledge and everyday action (Corbin & Strauss, 2008; Heath &
Cowley, 2004). Critical realism and constructivism likewise have found a suitable methodology in the utility and flexibility of GT (Charmaz, 2006; Lee, 2012; Mills et al., 2008; Oliver, 2012; Urquhart, 2001). In the case of this study, a critical realist position, though not essential to the mechanics of the GT research methodology, informed its intent, possibilities, and spirit. In short, critical realism and GT methodology—with its symbolic interactionist context—form a cohesive ontological and epistemological paradigm for this qualitative, exploratory study.
Data Collection
In order to investigate the social reality of a hackerspace community in a non-Western context, a nascent hackerspace in Taiwan was selected as the primary research site for the reasons stated above. I began researching the emerging Taipei Hackerspace community in January 2013, before a physical location had been secured, and continued to do so after a space was chosen and community activity increased.
In order to gain a detailed account of the Taipei Hackerspace community network, data from multiple sources was collected during the course of one year. The primary data sources were: 1) participant observation and 2) in-depth interviews with Taipei Hackerspace participants. Participant observation notes (n=10) were written up after returning from “the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
field”. Interviews (n=8) were first recorded with a digital audio recorder and then
transcribed. Interviews in Mandarin were then translated into English, as well. The unit of analysis for the study was the Taipei Hackerspace community network. The units of observation were the topics or interactions represented in participant observation notes and interviews.
Data collection was conducted over the course of 11 months and was divided into two phases (see Table 1). Phase 1 included six instances of participant observation and two semi-structured, in-depth interviews. Phase 2 data collection was comprised of four additional visits to the hackerspace and in-depth interviews with six additional participants.
Table 1
Data Collection
Data$Source$/$Type$ Phase$1$
Feb.%May)2013) Phase$2$
Sept.%Dec.)2013) Total$
Feb.%Dec.)2013)
Participant)observation)(#)written)accounts)) 6) 4) 10)
In%depth)interviews)(#)transcripts)) 2) 6) 8)
In many cases during participant observation specific dialogue was noted. Direct quotations of participant dialogue from these notes were likewise coded and analyzed.
Auxiliary information sources included observation that was not explicitly documented, as well as observation of material culture and online communication. These observations were not explicitly coded but nonetheless provided context for topics raised in participant
observation notes and interviews, guided subsequent data gathering, and also entered analysis through researcher memoing.
Participant observation. Participant observation, as a type of fieldwork, entails social involvement with research subjects and can “supply empirical findings about little known or stereotyped populations, particularly those outside the mainstream” (Gans, 1999, p.
‧
540). As it requires immersion in the community being studied, participant observation
“enables researchers to discern how a social world appears to its participants” (Lindlof &
Taylor, 2011, p. 136).
Participant observation field notes on the Taipei Hackerspace community were recorded during the period February-December, 2013 (see Table 2). Aside from an initial organizing meeting that took place in the Taipei offices of one co-founder’s employer—a leading international Internet and technology company—and a second “Founder’s meeting”
that began at a café, all other notes were taken in the Taipei Hackerspace itself. During the data-gathering period, there were some scheduled meetings, events and workshops, however the majority of time was unscheduled, “drop-in” time. As long as a key-holder was present, the space was open to the public. In choosing times to gather participant observation field notes, I sought to get a balanced picture of hackerspace activities; my notetaking covered meetings, a class, open houses, as well as unscheduled free time at the hackerspace. The data generated through participant observation offered an ongoing counterpoint to data from in-depth interviews.
Table 2
Participant Observation Instances (Coded)
Date Location Identifying Event(s)
February 2, 2013 Int’l tech co. offices, Taipei 101 Taipei Hackerspace “Mini Faire”
March 23, 2013 Frog Café/Taipei Hackerspace (Da-tong District, Taipei)
Founder’s meeting and touring the space
April 16, 2013 Taipei Hackerspace (Da-tong District, Taipei)
New furnishings, RC copter, chatting
April 28, 2013 Taipei Hackerspace Open House
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
May 17, 2013 Taipei Hackerspace Socializing, aquaponics, DIY speakers
May 30, 2013 Taipei Hackerspace Key-holder meeting
October 19, 2013 Taipei Hackerspace E-book publishing class
November 5, 2013 Taipei Hackerspace Key-holder meeting
November 20, 2013 Taipei Hackerspace Drupal (content management software) meet-up
December 10, 2013 Taipei Hackerspace Open House
In-depth interviews. Based on initial exploratory research combined with participant observation of the group, interview subjects were selected in an evolving manner from the Taipei Hackerspace population using non-random, theoretical sampling. Theoretical
sampling, where the sampling criteria emerge along with the study itself, is a basic principle of grounded theory (Coyne, 1997). With theoretical sampling, when emerging trends are seen in the data, subsequent elements may be chosen in an effort to confirm or question them (Koerber & McMichael, 2008). Selection of participants followed from criteria emerging from the study itself and interviews took place in the hackerspace with the exception of one conducted in a café to accommodate the interviewee. Interview participants were categorized as one of three subject types: 1) key informants, 2) core participants, and 3) peripheral participants. These subject types corresponded to participants’ relative involvement in the Taipei Hackerspace. Two principal co-founders of the Taipei Hackerspace were considered key informants, five core participants were regularly involved in the hackerspace, and one peripheral participant was chosen because of a unique perspective, having shown initial interest in regular involvement but later displaying reluctance to participate. An effort was also made to select participants that were roughly representative of the larger group
‧
demographically, and included a mix of male and female, as well as Taiwanese and non-Taiwanese. It is interesting to note that fully five interviewees were technology sector entrepreneurs. (Details regarding interviews and participants can be seen in Table 3.)
The interviews with participants were semi-structured, but with a great degree of flexibility, allowing the researcher and participants to pursue unanticipated directions in the conversation. An interview guide (see Appendix B) was used. These questions were designed to yield answers in practice that would contribute to answering the research questions, and were iteratively adjusted to actual participants and situations (Maxwell, 2009). Interviews averaged slightly over 52 minutes in length and the total length of interview audio was seven hours. All interviews were recorded using a digital audio recorder then transcribed—and translated into English, as necessary—at a later date for coding and analysis.
Table 3
Marketing/Entrepreneur English May 30,
2013 44
Engineering/Freelance English May 30,
2013 37
MU38 38 M U.S. Key
Informant
N.A./Data center &
network engineer English Septembe r 7, 2013 57
design entrepreneur Mandarin Novembe r 20, 2013 38
‧
Marketing coordinator English December 2, 2013 45
* Subject IDs represent the participant’s gender-nationality-age; e.g. the subject ID for a 26 year old Taiwanese female would be “FT26”.
Observation, material culture, and online communication. Additional sources of information lent context and counterpoint to the primary data from participant observation and in-depth interviews. Based on the literature review, participant observation field notes, and interviews, certain interaction patterns, online communications, and components of material culture emerged as salient to the developing Taipei Hackerspace community. To provide one representative example, a number of interviews referenced a particular project prototype: the Facebook “Like” counter (see Appendix C). It was clear that this project also involved sensitizing concepts from the literature: 1) the use of open source soft- and
hardware, 2) the hacking or repurposing of a consumer product, and 3) collaborative learning.
This project emerged as a salient example in the research and my observations of its aspect and use entered the analysis through the memoing and higher level coding processes (described further in the following section on analysis). Similarly, participants’ use and organization of the physical space was an emergent topic, so my observations regarding interactions with the physical space informed the analysis. Regarding language preference in online communications, an emergent topic in interviews, my ongoing monitoring of Taipei Hackerspace online communications provided context for analysis. The selection of these sources for analytical purposes can therefore also be described as theoretical sampling, as it was guided by emerging themes.
These additional information sources contributed an important counterpoint to participants’ accounts, as documents, artifacts, and material culture can reveal aspects of social reality that human participants do not acknowledge or of which they are not even conscious (Hodder, 2003). In other words, social actors themselves may overlook material
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
culture because they are surrounded by it. As Lindlof & Taylor (2011) wrote,
“overfamiliarity with the concrete elements of a scene tends to breed undersensitivity to their presence and rhetorical effects” (p. 222). Though sometimes unnoticed even by participants, considering material culture and even the spaces in which people interact can aid in
understanding social and structural relationships in qualitative research (O'Toole & Were, 2008). Artifacts like the Facebook “Like” counter, material culture present in the physical space, and online representations are important for a number of reasons: as reified objects that represent certain values or power structures, as projections of symbolic identities, and simply as productive resources (Hodder, 2003; O'Toole & Were, 2008). Hardware, tools, furnishings and decorations are all fair game for critical engagement. The habits and character of a group or community engaged in an activity together can be reflected by their surroundings and material culture, in other words, by their “stuff” (O'Toole & Were, 2008).
Rigor & Credibility
A number of strategies have been proposed to enhance the “validity” or internal consistency of qualitative studies, thereby increasing credibility. The terms and criteria for qualitative studies may not be the same as in the case of qualitative research (Gasson, 2004).
The strategies applicable to this study included intensive, long term involvement, collecting
“rich” data, making comparisons, looking for negative cases, researcher reflexivity, and triangulation (Creswell & Miller, 2000; Maxwell, 2009). The definition of “intensive, long term involvement” is somewhat subjective, however it is best represented by the use of participant observation (Becker & Geer, 1957). Though the participant observation in this study may not qualify as intensive or long term relative to studies of larger scale, it
nonetheless allowed reasonable familiarity with the Taipei Hackerspace environment and its participants, thereby providing valuable perspective on individual interview responses. In-depth interviews and their transcripts, along with participant observation notes, provided
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
thick, rich data replete with detail that worked to balance out any respondent
disingenuousness or observer bias, as Becker (1970) proposed (as cited in Maxwell, 2009).
The strategies of making comparisons and looking for negative cases are explicitly
incorporated into the analytic process of GT methodology used in this study (Gasson, 2004;
Strauss & Corbin, 1998). Comparisons may also be made with cases from secondary research as well as prior researcher experience (Maxwell, 2009). Accounting for reflexivity, or the role of researcher-participant interaction in construction of data, can also enhance the rigor of a qualitative study (Gasson, 2004; Hall & Callery, 2001). In this study, reflexivity was also incorporated throughout the research process, by being mindful of my own potential for biases and acknowledging the co-constructed nature of the data. Finally, triangulation, through the use of multiple data sources—multiple interview subjects and instances of participant observation at different times, for example—as well as through the use of more than one data-gathering technique, contributes to increased consistency and rigor (Lindlof &
Taylor, 2011; Patton, 1999). These strategies described were employed to increase the internal consistency and overall credibility of the research findings.
Ethical considerations. After I had spent some time in the community and was therefore a “known quantity” to the most active members, I was open about the fact that the Taipei Hackerspace was my thesis research site. Members were aware that I was conducting interviews with other members, and, aside from one, these interviews all took place semi-publicly in the hackerspace itself. (“Semi-semi-publicly” here means that while other members in the space were aware that an interview was being conducted, they could not overhear what was said; interviewees could speak without worrying about being overheard.) Members of the community therefore were aware that interviews was part of my research, however they may not have been aware of the field notes component of the research. I was in the habit of observing onsite and then writing up notes afterward in an offsite location. In order for
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
participant observation to serve the purpose of triangulating with interview responses, it seemed necessary to refrain from broadcasting the fact that my observations were also data, for fear of affecting participant behaviors. Interview participants gave informed consent to be interviewed for the purposes of this research. Although interviewees generally seemed unconcerned about remaining anonymous, I nonetheless used an identification scheme (seen in the chart above) that allowed me indicate specific individuals while not identifying them by name.
Data Management
Participant observations notes, transcripts of in-depth interviews, and researcher memos produced a trove of rich, detailed data for analysis. One partial solution to the problem of how to manage and organize this data was the use of computer aided qualitative data analysis software (CAQDAS).
Use of CAQDAS. A large amount of data was generated during the course of this study. In order to aid in various data management and analysis tasks, computer aided qualitative data analysis software (CAQDAS) was used. As identified by Lewins & Silver (2006), a CAQDAS program can offer the qualitative researcher several types of
functionality, including: 1) a project based management structure for importing and
organizing data, 2) “closeness to data” via near instantaneous accessibility of transcripts, etc., 3) the ability to flexibly code, re-code, and thereby link chunks of data, 4) search and
retrieval based on text excerpts, codes, and other descriptors, 5) writing tools for
conveniently creating memos, and 6) the ability to view and output different subsets of data.
Due to the large amount of data generated during a grounded theory study involving
interviews and field notes, as well as to the central role of coding for the GT methodology, a CAQDAS program offered these distinct benefits. For this study, I chose to use a CAQDAS program called “Dedoose”, from SocioCultural Research Consultants, LLC, due to its
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
functionality, relative ease of use, and cost-efficiency. Though it is a relatively new offering, a number of academic studies in different fields have used Dedoose, and it offers all of the functionality of a standard CAQDAS (Brown III, Yen, Rojas, & Schnall, 2013; Marion Suiseeya & Caplow, 2013; Ruf & Back, 2013).
In the present study, the field notes generated through participant observation, transcriptions of digital audio recordings from in-depth interviews, as well as memos and
In the present study, the field notes generated through participant observation, transcriptions of digital audio recordings from in-depth interviews, as well as memos and