case analysis (Eisenhardt, 1989). For the analysis of the longitudinal study, I adopted a process theorizing approach, with the aim of tracking the changes of events in descriptive narratives over time (Langley, 1999). Although the focus of the coding varies from paper to paper, the coding was based on the data sets described above. I coded the transcripts from the interviews, field notes from participant observation, and the collected documents using the software NVivo version 11.
In general, I coded my data following three steps. For the first step, I conducted open coding on the collected data, which was organized according to different research foci. At this stage the codes are constructed inductively from raw data. In paper 1 and 3, I have also captured an event-time series to document the changes over time (Pettigrew, 1985; Poole et al., 2000). These event-time series helped me to build a connection between events and the emergent analytical themes at a later stage of coding. The initial open coding results in a wide range of unstructured first-order codes that are used as the basis for future coding and analysis.
For the second step, I zoomed out of the data and turned to different sets of relevant literature. During this stage, I used analytical concepts (e.g., institutional logics, temporary organization, technological frames of references, knowledge sharing, and adaptive governance) as my interpretive device (Bowen, 2006), or sensitizing device (Klein and Myers, 1999) to abstract patterns from the first-order codes. The process of second-order cording was reconfigured throughout the data collection, iterating between empirical materials and the research literature. By repeating the cycle of coding, I was able to generate theoretically informed and empirically driven insights of the focus of each study, improving the transferability of these theoretical insights (Lincoln and Guba, 1985).
For the third step, I revisited the second-order codes and aggregated them into analytical dimensions. I then linked these dimensions to build the final theoretical insights. Examples of the code tables and detailed descriptions of coding procedures can be found in each paper.
research process, and each of them also include other issues such as theorization and literature review, I will split the evaluation in two for the clarity of writing.
4.7.1 Data Collection Evaluation
During data collection, I paid particular attention to the interaction between my informants and me as a researcher (Klein and Myers, 1999, p. 74) to improve the trustworthiness of the collected data.
More specifically, in case 1 and case 2 where I conducted a single case study and had limited time to engage with the informants, I did what Guba and Lincoln called “prolonged engagement” (1985, p. 301), that is to invest time in building trust with the informants, and learn about the surroundings where the collaboration takes place. I asked my informants to give me a tour of their work settings prior to the interviews, to expose myself to the multiple contextual influences (i.e., the space where servers are located, the way of communication between government and their stakeholders) that impinge on the informants’
daily organization. On occasion, where possible, I also asked for the opportunity to visit the site again and to talk with some of the informants.
In case 3, where I conducted longitudinal studies, and case 4 where I had more time to engage with the informants, I further combine “prolonged engagement” with “persistent observation” (Lincoln and Guba, 1985, p. 32) to identify among the contextual characteristic and elements that are most relevant to the problem and focusing on them in detail. It is among these observations at the beginning of my fieldwork where I uncovered the significance of WeChat groups in holding together the collaboration among stakeholders.
Nonetheless, approaches such as “prolonged engagement” and “persistent observation” may also suffer from potential systematic “distortions” in the informants’ narratives (Klein and Myers, 1999, p. 77).
Especially considering I was introduced as an “evaluation researcher” in case 1, 2 and the first year of fieldwork in case 3. And there was a noticeably positive tone in the initial narratives among our initial contacts. To reduce these possible “biases”, I have triangulated different data methods and sources (Lincoln and Guba, 1985, p. 306) by, for instance, including accounts of the same events from different project stakeholders, or comparing some of the accounts in interviews and in official documents.
4.7.2 Data Analysis Evaluation
In analyzing the collected data, I carefully related theoretical abstractions and generalizations to the details of the cases as I experienced them during data collection. For instance, after the initial coding of the data collected from the preliminary interviews, I observed group differences in terms of perceptions, the use of WeChat, as well as opinions about ownership among the stakeholders, which had to do with the sectorial
These iterative processes between theoretical abstractions and empirical findings also lead to the detection of “possible contradictions between pre-conceptions and actual findings” (Klein and Myers, 1999, p. 72), resulting in the revision of theoretical preconceptions of this dissertation, and increasing the generalizability of the research findings. For instance, derived from the literature review, my initial focus was to understand how congruence on social media use contributes to the governance of the collaboration, which was later on proved otherwise in the empirical data.
In addition, I also paid attention to the critical reflection of the social and historical backgrounds of the research setting (Klein and Myers, 1999, p. 73). For instance, I investigated the historical setting of the administrative reform and e-government development in China (e.g., in section 4.4.1.2 and 4.4.1.3). The results showed that although China has undergone a different path of public administration reform compared to Western countries, the Chinese context has showcased a compatible level of institutional complexities, which provides a suitable context to understand the government transformation in a fast-changing e-government environment.
I also contextualized my understanding of each case during data analysis by digging into the historical setting of each case, in order to understand how the current situation under investigation emerged. For example, in case 1, the story on the turn of the project – Lu’an governments feeling threatened by the Longsun company request for fees for every change they make - did not surface until I looked into the historical development of the case. This particular incident informed me about the dynamic nature of the governance practice of these collaborations.
During data analysis, I also tried to engage with these social and historical details by providing thick description of the case context in each paper. The descriptions present to the readers a more balanced picture about how the governance and organizational form of e-government collaboration changed as it developed.
In general, the development of the research design follows a hermeneutic circle (Klein and Myers, 1999, p. 71) by iterating between the phenomenon of e-government collaboration, empirical data within the local context, as well as the theoretical perspectives of IT and organizational change. As showed above, during data collaboration and analysis, I engaged in research efforts to contextualize the cases, iterating between empirical details and existing theoretical insights of meta-phenomenon, as well as including multiple accounts of my inquiries from a variety of informants and over a span of time. These efforts helped me produce empirically rich data (Lincoln and Guba, 1985), and generalize the empirical details to theoretical insights (Klein and Myers, 1999; Lee and Baskerville, 2003; Walsham, 1995) on the governance and organizational form of e-government collaboration through the mediation of social media.
In sum, in section 4, I have explained the ontological, epistemological and methodological choices of this dissertation to help the readers to understand the research paradigm that guides the research design of this dissertation. I then presented the administrative reform and e-government development in China, as well
have evaluated the development of the research design to show the limitations and implications of these developments to the generalization of the findings. In the next section, I present the findings based on the empirical cases and explain how they answer to the research questions.
5 FINDINGS
In this dissertation, I used four papers to address the overall research question of this dissertation:
• How do the governance and organizational form of e-government collaboration occur through the mediation of social media?
In this section, I will address the overall research question by answering to the two sub research questions in the following.
• How does the governance of e-government collaboration occur through the mediation of social media?
• How does the organizational form of e-government collaboration occur through the mediation of social media?
In section 5.1, I address the first sub research question by unfolding the relation between governance of e-government collaboration and enactments of social media. In section 5.2, I address the second sub research question by unfolding the relation between the organizational form of e-government collaboration and enactments of social media.