3.1 Research Methodology
Simultaneously contradicting results caused by different research orientations and methods are present in the field of scientific research (Tuomi & Sarajärvi 2018, 124). This research aims to provide a rich description of the phenomenon of knowledge sharing in multinational virtual teams. The research emphasizes how online collaboration tools and digital mediums are used in sharing knowledge in virtual teams and between virtual teams. Studying socially constructed phenomenon requires depth of investigation (Saunders, Lewis & Thornhill. 2009, 111). Wel-man and Kruger (1999, 189) describe phenomenology as follows: “the phenomenologists are concerned with understanding social and psychological phenomena from the perspectives of people involved”. Based on Welman and Kruger’s view, it could be claimed that this research follows the phenomenologist discipline as the focus is in the experiences of the subject group.
Philosophical disciplines are often linked to each other (Stanford Encyclopedia of Philosophy 2013) and it may not serve the purpose of this thesis to dive too deep into the philosophical debate.
Qualitative methods are applied due to the complexity of the subject and the interest towards thoughts and experiences of individuals working in virtual teams (Walle 2015, 9). Following the reflections of Valli (2018, 19), universal generalizations cannot be produced to the same degree with qualitative research data as with quantitative sampling, yet qualitative research findings do possess a certain level of transferability to similar contextual environments. In this research, the aim is to interpret and present findings from knowledge sharing in multinational corporation’s (MNC) virtual marketing teams that can be applied to other MNCs and organiza-tions operating virtually. To bring real-life context to a complex concept and contemporary phenomenon, case study methodology is applied (Flyvberg 2006; Dawes Farquhar 2012, 6; Yin 2014, 16).
Data is collected with individual semi-structured interviews, as they can provide depth in-sights when prior knowledge of the subject is limited (Hirsjärvi & Hurme 2008, 35). Partici-pants are knowledgeable and experienced in working in virtual teams and belong to one or more
virtual marketing teams at a MNC. Marketing experts are chosen due to marketing’s vital role in strategic knowledge sharing in MNCs (Kiessling et al. 2006). Interviews are digitally rec-orded, transcribed and transcripts are analyzed with thematic content analysis to reveal further findings. Data is analyzed with an abductive approach, in other words combining existing the-ories with novel insights drawn from the data.
3.2 Qualitative Research
Qualitative research often investigates and presents individual interpretations of the research subjects (Valli 2018, 19). Qualitative research methods are applicable in situations where con-clusions cannot be formed purely based on observations and where idiosyncratic thought of informants are of interest, such as thinking, motives, goals and feelings (Walle 2015, 9-10).
Valli (2018, 19) advises to choose qualitative approach when the research is a preliminary in-vestigation for future studies. This thesis uses a qualitative approach utilizing semi-structured theme interview as the data collection method and thematic content analysis as the data analysis method. The methods are chosen based on the explorative nature of this study, in which sub-jective individual thoughts and past experiences are of interest. (Walle 2015, 10).
Tuomi and Sarajärvi (2018) highlight that qualitative research can be viewed as an umbrella term for a vast amount of qualitative methods and research traditions that have had various characteristics from decade to decade and from geographical location and field of research to another. The authors also point out that different methodological guidebooks offer contradic-tory information about different methods, which makes interpretation challenging. They sug-gest especially junior researchers to keep in mind that different methodological guidebooks only give one point of view to qualitative research, which should not be generalized too heavily to cover all qualitative research and followed blindly. (Tuomi & Sarajärvi 2018, 10-13, 18.) What makes the discussion around qualitative research even more perplexing, is the confusion around terms method and methodology. Method can be viewed as a means to justify the infor-mation created in the study while methodology is a means to justify the applied method – whether it is valid for the purpose of the study. (Tuomi & Sarajärvi 2018, 13, 18.) To avoid
possible pitfalls, academic methodological literature is researched from several scholars before conducting the study.
Quantitative methods have dominated business research for a long time (Walle 2015, 3), but qualitative methods have increased due to emerging trends such as naturalistic movement in consumer research and the growing subfield of business anthropology (Walle 2015, ix). Walle (2015, 9) points out that although qualitative methods have had a rocky road towards respecta-bility among scholars, their reputation has grown in the recent decades.
Both quantitative and qualitative methods have their strengths and weaknesses, and Walle (2015, 8) argues that having a rich array of methods available provides flexibility for the re-searcher. When making methodical decisions, the specific research purpose, circumstances sur-rounding it and drawbacks of potential methods should be carefully evaluated. The choice of method should be justified, and limitations acknowledged. (Walle 2015, 8-11.)
Quantitative methods can be characterized by standardized gathering of the data with predeter-mined categories – such as yes vs. no or Likert’s scale – which enables using a variety of math-ematic and statistical methods for analysis. Although controlled quantitative analysis minimizes the risk of subjective bias, rigorous manipulation of the data also increases the risk of distorting the findings, especially if the predetermined categories are artificial or inappropriate for the specific research purpose. Qualitative methods on the other hand are more prone to subjective bias. They are more flexible, and appropriate when the aim is to preserve the complexity of the data, enabling richer, more realistic and sometimes unexpected findings. (Walle 2015, 8-11.)
3.3 Case Study
This research explores the complex phenomenon of knowledge sharing in virtual teams, and the aim is to collect empirical data from a human action perspective: knowledge sharing is viewed as an action that humans perform. According to Flyvberg (2006) in the human affairs research there is only context-dependent knowledge. To bring consistency and context into the study of complex concepts and varying interpretations, a case study approach is applied. Yin (2013, 16) defines case study as an empirical in-depth investigation of a contemporary phenom-enon within its real-world context. Case study is often appropriate when the research aims to
answer questions of how, why or when (Farquhar 2012, 6). As this research investigates the virtual knowledge sharing phenomenon, i.e. how knowledge is shared in virtual teams using available online collaboration tools, case study approach is justified.
Knowledgeable members with considerable experience in virtual teamwork are chosen from different multinational virtual marketing teams in one MNC. Marketers as the case sample rep-resent a professional group universal to most MNCs, enhancing the transferability of the find-ings. Although the participants do not necessarily co-operate directly, the same overall infra-structure of virtual work applies. MNCs have substantially different marketing infra-structures (Pires, Rocha, Borini, & Rossetto 2015). As the main interest and research problem of this study is knowledge sharing in multinational virtual teamwork, marketing strategies and activities are not addressed specifically. The specific teams or the organization itself are not of interest, the focus in on the professional knowledge and experiences the participants have gathered from many different virtual teams and multinational working environments over the time. The con-textual decision to choose marketing experts from a single MNC is done from practical reason-ing as this enables cost-effective and convenient snowball samplreason-ing (Tuomi & Sarajärvi 2018, 73), and ensures accountability by adequate contextual narrowing (Valli 2018, 162).
Qualitative research does not aim to produce statistical generalizations (Tuomi & Sarajärvi 2018, 73), but analytical transferable generalizations are pursued, as is typical for a case study approach (Valli 2018, 68). Like in any research, the focus of the case study is crucial for its success (Farquhar 2012, 7). Selection of the case and concepts is central when evaluating the research quality and transferability of a case study, as they represent the research problem and topic. Tight binding between theory and empirical findings is vital to any scientific research project. The case’s representability of the studied phenomenon enhances transferability of the findings. Transferability of a case study can also be increased with thorough theoretical discus-sion, existential explaining and careful describing of the investigated phenomenon. (Valli 2018, 168-169.)
3.4 Semi-structured Interview
As the research subject is not well known, qualitative interviews provide necessary in-depth insights (Hirsjärvi & Hurme 2008, 35) more efficiently than a quantitative approach possibly could. This study explores the complex and abstract concept of knowledge of which multiple definitions have been developed ever since the debates of ancient Greek philosophers (Nonaka 1991; Nonaka 1994). Approaching such a subject with qualitative interview where the re-searcher is present decreases the risk of different interpretations among respondents. Usage behavior of different online collaboration tools is examined from an individual user experience point of view rather than user statistics which also validates qualitative interview as the data collection method (see e.g. Herzog et al. 2013).
Theme interview is probably the most commonly used research interview method (Vilkka 2005, 101), and many qualitative interviews within business research fall into the semi-structured category (Eriksson & Kovalainen 2008, 82). Semi-structured interview can be viewed as an intermediate form of structured and unstructured interview (Hirsjärvi & Hurme 2008, 47). Peo-ple’s interpretations and meanings of concepts are crucial, as well as the meanings that arise from interacting (Hirsjärvi & Hurme 2008, 48). Interviewees respond to their own and varying interpretations of the same questions and due to this it may be difficult to compare the empirical data (Eriksson and Kovalainen 2008, 82).
Semi-structured methods are characterized by the fact that some aspect (e.g. topic, issue or theme) of the interview has been set – but not necessarily all. In semi-structured interview the questions are same for everyone, answers are not tied to the pre-defined options and thus the interviewees can answer in their own words. (Hirsjärvi & Hurme 2008, 47.) In a theme inter-view, the interview is built around specific themes and the questionnaire is developed in ad-vance, but the interviewer may change the wording (Hirsjärvi & Hurme 2008, 47) or the order of questions (Eriksson & Kovalainen 2008, 82).
The selected thematic areas and interview questions serve as the interviewer's checklist and necessary point of reference for guiding the discussion during the interview (Hirsjärvi & Hurme 2008, 66). The interviewer must make sure that all topics are covered during the interview, and at the same time be prepared to inquire for more in-depth responses. If the interviewer keeps too close to pre-planned questions, the important topics raised by the interviewee could be
prevented. (Eriksson & Kovalainen 2008, 8.) In a theme interview, either in an individual or a group interview, the interviewer must ensure that the discussion remains in the selected themes (Hirsjärvi & Hurme 2008, 61). According to Eriksson and Kovalainen (2008, 82) an advantage of a theme interview is that the contents are mostly comprehensive and systematic, while the tone of the interview can be fairly informal and conversational.
Familiarization with the target group helps framing the questions of a theme interview (Vilkka 2005, 105.) In addition to knowing the target group, it may be beneficial to find out about the interviewee’s background such as work history, to understand what guides his/her thinking around the research question. (Vilkka 2005, 110-111.) If the knowledge of the target group, culture or situation is lacking, gaps in interpretation may come up and different interpretations may distort the subject. The interviewer should familiarize oneself with the research, the topic, the research problem and the objectives of the research as well as the methods. (Vilkka 2005, 110-111.)
3.5 Qualitative Content Analysis
Content analysis is a flexible and versatile method: it can be purely quantitative or qualitative or something in between (Klenke, Martin & Wallace 2015). Content analysis can be deductive, inductive or abductive (Tuomi & Sarajärvi 2018, 79). In this research content analysis is con-ducted with an abductive approach. Deductive approach can be described as theory-based, in-ductive as data-based and abin-ductive as theory-guided. (Tuomi & Sarajärvi 2018, 79.) Dein-ductive approach proceeds from general towards specific and can be used to test a theory in a new context. Inductive approach moves from specific towards general and can be used when there is little previous research about a particular phenomenon. (Klenke et al. 2015.) Abductive con-tent analysis can be described as a form of analysis directed by theory yet not based on it alone.
In abductive content analysis the logic guiding the analysis combines pre-known theoretical models with insights drawn from the research data. It is up to the researcher to decide how much existing theories are guiding the reasoning. (Tuomi and Sarajärvi 2018, 79-80.) Timmer-mans and Tavory (2012) describe abductive analysis as a form of reasoning where a situational fit is sought between different observations and theories, and phenomenon of interest is per-ceived in relation to other phenomena. One use case of abductive analysis is in the process of
hypothesis formulation, whereas the formulated hypothesis can be tested with deductive anal-ysis. Examining existing theories far and wide without prejudice inspires formulating novel hypotheses and findings. (Timmermans & Tavory 2012.)
Content analysis can be viewed as a systematic coding process in which data is broken down to pieces, conceptualized and finally put back together in new ways (Klenke et al. 2015, 98).
Content analysis begins with reading the text numerous times to familiarize with it and to de-velop an understanding of the context of the text. The researcher has to analyse own previous knowledge on the subject that is brought into the project. (Klenke et al. 2015, 96.) Before be-ginning the analysis, the researcher determines units of analysis, if they are words, sentences, paragraphs or larger expressions of thought (Klenke et al. 2015, 96; Tuomi & Sarajärvi 2018, 91).
Next the original data is reduced: relevant content that captures the studied concepts is high-lighted and organized under fewer content categories (Klenke et al. 2015, 103; Tuomi & Sa-rajärvi 2018, 92). The researcher adds notes about thoughts, impressions and associations that arise (Klenke et al. 2015, 99). This reduced data is broken down into pieces and clustered under sub-categories representing key concepts and forming a basis for further analysis. Higher level categories are formulated by connecting related sub-categories around key concepts or phe-nomena of the study. The abstraction process i.e. conceptualization of the original data is con-tinued until a point of theoretical saturation is achieved, where further conceptualizing no longer provides valuable knowledge. (Klenke et al. 2015, 99-100; Tuomi & Sarajärvi 2018, 92-93.) Bias in content analysis often stems from the selection of content to be analysed and assump-tions of the researcher, resulting in insufficient coding framework. Other potential factors in-fluencing the success of the analysis are coding fatigue and lack of training. (Klenke et al. 2015, 103, 107.)