Previous Chapter - Thesis Index - Next Chapter
To reference this chapter: Berglund, H. (2005). Toward a Theory of Entrepreneurial Action, Chapter 3 - Methodology, PhD, Chalmers University of Technology.
Methodology can be said to comprise two things: the underlying assumptions and justification guiding the choice of methods, and the technical aspects of the methods themselves including procedures and methods for analyzing empirical material, dealing with ambiguities and so on (Alvesson and Deetz 2002). Management research is most often concerned with the technical aspects of methods, but as argued in the introduction and previous sections, entrepreneurship research may stand to gain from a deepened reflection over issues such as the nature of human intentions and actions, and the relationship between scientific ‘third person’ and entrepreneurial ‘first person’ knowledge. This chapter will therefore discuss some methodological assumptions before describing the technical aspects of the methods used.
Ontology is about the nature of being, or in this case the nature of human action. As discussed in the introduction, action can be understood from two opposing perspectives: one where intentions somehow cause action, and a more holistic version of situated action. The first perspective reflects an ontological dualism in that action is divided into a mental state that exists independently of, and gives rise to, a specific action that takes place in the physical world (e.g. Ajzen 1991). The second perspective views action as a form of situated coping in which intentions and actions are inseparable. Here action exists in its own right and is not merely an epiphenomenon that is made up of more ‘real’ mental states, social/structural pressures, physical behaviors and so on (e.g. Dreyfus 1991).
This second position, which is adopted here, implies that actions do not really exist outside of both the concrete circumstances and the greater social and historical situation in which they appear. This emphatically does not mean that humans do not have intentions or that entrepreneurs do not explicitly plan ahead. It only means that to comprehend and study action, especially creative action (Joas 1996), it is important to acknowledge it as more than isolated intentions and performed behaviors.
This ontological position is reflected in an epistemology that prioritizes entrepreneurs’ life worlds or lived experiences, i.e. the immediate experience of a situation or phenomenon as it is lived through, over abstract knowledge or reflection (cf. van Manen 1990). The focus on lived experiences follows a hermeneutic and phenomenological tradition where human knowledge is not seen as an objective mirror of some world ‘out there’. Instead all human knowledge is seen as relative to situations, social contexts, purposes, previous experiences etc. This means that an entrepreneur’s knowledge of something like risk (in the context of entrepreneurial action) consists in the meaning that risk has for him or her and in the ways that risks are perceived and enacted in specific situations.
The hermeneutic stance also affects researchers’ attempts to understand subjects’ experiences. Traditionally investigators have tried to neutrally observe how things appear ‘in themselves’ (e.g. the entrepreneur’s experience of risk) by recourse to objective rules that validate knowledge in such a way that personal opinions or preexisting theoretical frameworks do not influence results. Such neutrality builds on a dualistic ontology and a realist epistemology where the world can be known independently of any human interpretation. Such neutrality is seen as impossible. Access to the world, e.g. subjects’ experiences, is always complicated by the researcher’s culturally and socially conditioned preconceptions and interests. Such preconceptions constitute the basis for understanding in the first place. Preconceptions and general knowledge interests are what provide the initial and often ill-structured understanding of a phenomenon that triggers interest, leads to the formation of tentative hypotheses, and also implies certain modes of inquiry rather than others (Packer and Addison 1989).
The unavoidable influence of preconceptions tends to give people a sense of the circularity of understanding. Unfortunately the circularity of human knowledge is often misunderstood as being vicious instead of essential (Dreyfus 1991: 200). The fact that there can be no procedures that guarantee objective knowledge of reality is no reason to give in to full-blown relativism. On the contrary, the very fact that an inquiry starts with certain preconceptions and interests indicates that research is not mere speculation that shoots beyond available evidence in a random way. The results therefore rest on a dynamic process involving researchers’ preconceptions, observations and reflections. This points to the importance of reflecting on why the research is undertaken and in what way results are produced and justified.
Methodology concerns criteria for choosing methods of investigation in a way that both fulfills the original research purpose and harmonizes with ontological and epistemological positions. The stated purpose is to explore the subjects’ meaningful experiences of risks, opportunities and the role of self in order to better understand the venture creation and development process. Methods should emphasize the situatedness of experiences, but retain focus on the individual rather than the situation as such. This suggests that methods should be qualitative and idiographic, i.e. seeking to appreciate the unique and concrete. The explorative ambition also demands an open mind to unexpected insights that the investigations may reveal.
The methods should consequently pay close attention to specific personal and contextualized experiences, while at the same time using techniques that ensure the trustworthiness of results. To this end researchers should thus strive to be explicit about the basis of their interpretations, employ systematic procedures to monitor and check their influence, and also try to display the results in lucid and transparent ways that allow for confident conclusions. However, too much formal rigor will introduce risks such as limited scope of findings, overconfidence in results and blindness to emergent or marginal phenomena. As mentioned above, there are no procedures that guarantee objectivity. Instead researchers should always try to find a balance between interpretive openness and rigor. Huberman and Miles summarize the dialectic of qualitative inquiry in a general piece of advice to “Seek formalism, and distrust it” (Huberman and Miles 2002: 396). These issues are discussed further under the heading ‘Limitations and Validity’ below.
The appended studies (I – III) use an interpretive phenomenological method (cf. Smith and Osborn 2003). This method draws on hermeneutic and phenomenological traditions and seeks to access and describe how subjects experience specific phenomena in a structured way (e.g. Smith and Osborn 2003, Giorgi 1985). Phenomenologically oriented methods are common in psychological research (Giorgi 1985) and in applied fields such as nursing (Benner 1994) and pedagogy (van Manen 1990) where researchers’ and practitioners’ interests are guided by a deep interest in the subjective experiences of their subjects. To be a good nurse or teacher takes more than competent reading of vital signs, making diagnoses or delivering lectures and setting grades. Professions such as these also require empathy and intuition, and researchers have therefore used phenomenological methods to capture subjects’ lived experiences in ways that allow for sympathetic care and more attentive pedagogy. To this end phenomenological methods have been used to investigate aspects like the lived experiences of recovering from drug addiction (Banonis 1989), living with chronic pain (Hellström 2001), or schoolchildren’s feelings of loneliness (Kirova-Petrova 2000). Recently it has been argued that phenomenological methods are suitable also in entrepreneurship research (Berglund 2005, Cope 2005).
In this thesis, phenomenological methods are used to describe entrepreneurs’ personal experiences of risk, opportunity and self, including the cognitive and practical strategies used to make sense of and manage these experiences. The method has no ambition to dig beneath or replace these everyday understandings. In this way it differs from both the cognitive and discursive approaches, which seek to explain meaningful experiences by recourse to either underlying cognitions or linguistic and other cultural resources respectively (Berglund 2005). Instead the goal is to empirically examine the variety of ways in which risk, opportunities and the role of self are experienced and how this relates to action.
Studies I-III are based on individual interviews which are then analyzed and merged by the researchers into a composite whole. Study IV explores risk perceptions in a corporate innovation setting. This study builds on the same basic premises as the other three studies, i.e. it starts with individual experiences that are brought together to construct a composite picture. However, here the analysis and merger of the individual level results are done by the subjects themselves as part of a facilitated group exercise.
Below the methods are described under the headings of sampling, data gathering, analysis and results. Since the methods differ somewhat between studies, each heading includes presentations of multiple procedures.
The general population from which the samples were taken is entrepreneurs involved in technology venturing, i.e. individuals who had themselves initiated and taken an active role in developing a technology-based business idea. Sampling in the different studies was purposive with focus on identifying a manageable group of entrepreneurs suitable for exploring the issue at hand. The aim was thus not to present intrinsically interesting cases or to represent some general population but rather to gain a more detailed picture of the phenomenon.
Studies I and II were based on interviews with 12 and 15 Swedish entrepreneurs. The selection was purposive, meaning that inclusion criteria were set up and then used to choose the sample (Berg 2001). The criteria were that the entrepreneurs should have initiated, and still work in, technology-based ventures that were between one and three years old. The ventures were identified using lists of firms provided by different innovation support centers. The entrepreneurs that were identified had all taken a key role in driving the process of inventing, producing and marketing a technological innovation in the fields of either information technologies, biotech or advanced services. Some came from a research background, others from employment in large firms, and still others directly from university studies.
The sample in study III consisted of 19 founding entrepreneurs who alone or together with others started new firms in the Swedish mobile Internet industry between the years 1998 and 2000. These entrepreneurs were randomly chosen from a list consisting of 121 Swedish mobile Internet ventures started between 1995 and 2000. This list was compiled during 2000-2001 as part of a larger project and contained approximately 80% of the firms in the Swedish mobile Internet industry at the time. Not all ventures were based directly on technological inventions but they were all embedded in and dependent on a technologically sophisticated environment.
Study IV drew on a sample of corporate innovators active in two large multinationals in advanced IT services and aeronautical development respectively. These two units were purposefully selected because they varied greatly in terms of things like work performed and corporate history. Access to the two units was obtained via industrial PhD students who were employed in other parts of the corporations in question. At each unit a cross-section of individuals with roles central to the innovation process were identified through discussions between the researchers and representatives of the respective organizations. The selected group of people covered all important aspects of the respective units’ key activities.
The first three studies used semi- to non-structured interviews which gave respondents room to speak freely and also allowed the interviewers to follow respondents’ leads into novel and unexpected areas as the conversations progressed. The interviews took place at the respondents’ offices or in their homes and averaged about two hours. In studies I and II between two and four interviewers were present at the interviews. One interviewer had the main interviewing responsibility and the others focused on taking detailed notes. Study III was part of a larger research program and the interviews were conducted by a team of interviewers. Here interviews were conducted by single interviewers. The interviews were tape-recorded and transcribed verbatim.
Study IV used group exercises that were conducted on location at the two corporate units. The exercises lasted for about half a day each and were based on the broadly articulated question: “What are the major risks/threats hindering us from achieving our objectives?” The exercise started with all participants individually reflecting on the theme question, writing down a number of risks (at both units participants asked if they could also include improvements, which they were encouraged to do). The participants then proceeded to jointly, as a group, discuss and merge the individual risks, producing a set of broad and thematically structured risk categories. During the discussion the categories were allowed to change, merge and split up, so that the end result represented a negotiated common understanding of each category.
In studies I and II, when there was more than one interviewer per session, the individual interview protocols were read by all the interviewers in order to establish interpretative flexibility and common meaning. In this way the interpretation of the general narratives, as well as of specific quotations, was agreed upon. The individual protocols were then re-read line by line and broken down into discrete parts, not according to grammatical rules such as sentences but with respect to visible changes in meaning, i.e. meaning units (MU) (e.g. Giorgi 1985). Each MU was associated with a tentative descriptive concept and broken out of the text together with its corresponding statements. When the whole text had been broken down in this way, the resulting list of MUs was re-read and discussed within the relevant research group. As the researchers worked their way through the list, MUs with similar meanings were cut out of the original document and pasted into a new document and given a tentative category heading. Each new MU on the list was thus either put in an existing category or given its own new category heading. This process generated a number of categories and during the process some categories which were found to be similar were merged and others split up, until all MUs had been clustered into categories that were agreed to capture specific homogeneous qualities of what was said by the participants. The categories and their interrelationships were then focused on in more detail, and similar themes were clustered into higher-order factors and in one case (study I) overarching super-factors.
Study III followed a similar procedure but since the interviews were conducted by a larger group of interviewers it was not possible to conduct a joint analysis. Instead the data analysis was carried out by the author with interviewers and transcribers being contacted when there was some doubt regarding what was meant or when details needed to be clarified.
In study IV the analysis was conducted as part of the group exercise as described above. Researchers were present but did not participate actively in the analysis and categorization or results.
During the analysis procedures in the interview studies, interpretations and judgments were continuously made by the researchers as categories and factors were developed. By returning to the original protocol and continuously questioning the bases of categorizations, the researchers actively sought to minimize the use of pre-existing theoretical categories and be true to the participants’ original expressions. If the MUs clearly coincided with existing theoretical categories, such categories were however used (cf. Smith and Osborn 2003).
Interpretive research is becoming more and more common in social science and management research. Unfortunately this tendency has not been accompanied by the development of appropriate criteria for evaluating results. Instead phenomenological arguments are often invoked to legitimize interpretive approaches only to be replaced by traditional positivist arguments during the validation phase (Kvale 1996).
As a result, qualitative investigations are often justified on the basis of correspondence with ‘the things in themselves’, e.g. ‘what risk really means to entrepreneurs’. The hermeneutic tradition does not acknowledge such an objective baseline. Consequently there can be no methodological algorithm that guarantees valid results. Instead, methodological procedures are seen to increase validity to the extent that the results of research address the original concern that prompted the research and allow for more comprehensive ways to understand and/or engage this concern (Packer and Addison 1989).
What triggered the present studies was a concern that key themes in the entrepreneurship discourse, especially the notions of risk, opportunity and self, were not theoretically understood in ways that improved our understanding of entrepreneurial action. The ambition was therefore to investigate these themes from the perspective of acting entrepreneurs in order to better understand the nature of entrepreneurial action. Results are therefore valid to the extent that this concern is addressed. Given the ontological and epistemological assumptions, no procedures can guarantee objective results. Still, there are ways to improve the trustworthiness of findings, i.e. the extent to which they address the initial concern that prompted the research. These are described and related to the present research under the headings of communicative, pragmatic and transgressive validity (Sandberg 2005).
Communicative validity seeks to assure interpretive coherence in dialogue and may be sought during all stages of the research process. Before conducting interviews (studies I-III) and exercises (study IV), the general themes were introduced to the subjects to attain common ground. This entails a careful trade-off between making sure that interviewer and subjects understand one and other while at the same time trying not influence the subject too much.
The interviews (studies I-III) were conducted in a dialogical form. This allowed misconceptions to be surfaced and elaborated. It also offered a productive way to establish a shared understanding of the topic under investigation. In this way dialogical interviews improve communicative validity, especially compared with more structured questions and answers in which fundamental misunderstandings may remain unnoticed (Kvale 1996).
During the analysis phase (studies I-III), meaning units (MUs) and emerging categories were constantly compared with one and other in an effort to identify differences and similarities. The MUs were also evaluated against the original interview protocols to make sure that they were not taken out of context. This constant comparison between parts (specific MUs and emerging categories) and whole (other MUs, categories and the original interview protocols) produced more coherent result-structures and also increased the clarity and validity of specific categories and factors.
Communicative validity was also increased by discussing the empirical results with other researchers. Results were presented at internal seminars and conferences to expose them to dissenting interpretations. In the discussion sections of the four studies, the results were also compared and contrasted with previous literature. Such discussions and comparisons validate findings to the extent that they resemble previous theory. Comparison with previous literature also provides a baseline for assessing the plausibility of novel and unexpected findings.
Communicative validity mainly concerns interpretive coherence, but for a number of reasons there may be discrepancies between what people say in an interview situation and what they do in real life. This leads to the issue of pragmatic validity. The methods used did not entail naturalistic observation where accounts could be compared with actual behavior. Instead pragmatic validity was improved by asking questions of a practical nature. This included following up general and vague statements by asking for concrete examples and more generally asking questions in a way that embedded statements in concrete situations. The interviewer also provided practical interpretations and examples of general answers in order to provoke more practical responses (Kvale 1996). This allowed the entrepreneurs to correct the researcher’s interpretation and often led to further elaborations and counter examples. These responses were incorporated into the analysis which served to increase the pragmatic validity of the results.
Communicative and pragmatic validity criteria are often promoted as essential when justifying interpretive research (Kvale 1996). However, the focus on coherent and consistent interpretations may fail to capture the ambiguity, complexity and indeterminacy that are also part of lived experience (Sandberg 2005). Transgressive validity counters these tendencies by stressing how validity may be lost if lived experiences are presented as overly simple. Transgressive validity may be especially important in the case of innovative entrepreneurship, which is characterized by making judgments and taking actions under conditions of uncertainty. During the interviews, transgressive validity was improved by trying to stay attentive to marginal and non-obvious responses. Probing conflicting and paradoxical responses, e.g. between personal and professional priorities, also provided occasions for increasing validity (studies I-III). This effort was also reflected in the results presentation, where potentially conflicting results were presented and elaborated side by side in order to retain the richness of the results (all studies).
Generalizability has traditionally been upheld as a central criterion for validity in social science. Compared to quantitative studies, which rely on statistical inference, generalizability is a criterion that qualitative research always struggles to accommodate. The appended studies investigated the lived experiences of purposefully selected entrepreneurs with the ambition to probe the phenomena of risk, opportunity and the role of self. The purposive sampling and the focus on meaning rather than frequency imply that results may be analytically generalized to theory, rather than statistically to a certain population (Yin 1994). The findings are also general to the extent that they harmonize with and inform the experiences and practices of relevant stakeholders such as entrepreneurs, incubator managers and teachers (cf. Stake 1994). The philosophical basis for making generalizations in interpretive inquiry is a contested issue. Without entering into this debate it is simply argued that the results are general to the extent that they may inform theory development, and if they may be incorporated into the practice of entrepreneurship and related activities. There are of course several limitations based on the procedures for sampling, interviewing and analysis. Given the preceding discussion it is however left to the reader to decide what these limitations are.
Previous Chapter - Thesis Index - Next Chapter