Data collection and data management

Data management is defined as a systematic, coherent process of data collection, storage, and retrieval (Denzin and Lincoln 1998, 180). Data analysis, on the other hand, refers to data reduction, data display, and drawing and verification of conclusions. A part of these processes start before data collection, others occur during the design and planning of the study, during data collection as interim and early analyses are carried out, and after data collection as the final products are approached and completed (Denzin and Lincoln 1998, 180; Silverman 2000, 143). This section describes how the data for this study were collected and what was done to the data prior to the final analysis.

Preparation of the research design and planning of the interviews took quite a lot of time, even though I was familiar with the research subject from a practical point of view.

According to Gummesson (1991, 12), the pre-understanding of academic researchers traditionally takes the form of theories, models, and techniques; generally they lack institutional knowledge, like knowledge of the conditions in a specific company, industry, or market. A lack of pre-understanding will cause the researcher to spend a considerable amount of time gathering basic information, and in any case, most of the information will not be obtainable outside the company or organization being studied.

Thus, the term “pre-understanding” is used to refer to a researcher’s existing knowledge, insights, and experience prior to the research, while “understanding” is the knowledge that develops during the study (Gummesson 1991, 50). Understanding usually emerges gradually in the course of the fieldwork as the researcher overcomes the initial bewilderment with a new or unusual language and system of social meaning. Once the researcher attains an understanding of the interviewee’s point of view, the next step is to learn how to think and act within that perspective (Neuman 2000, 355). This is why Gummersson (1991, 51) considers it vital for academic researchers who study entrepreneurs, for example, to have personal experience from working in a similar position of responsibility in which they have had to make and implement decisions.

According to Straus and Corbin (1998, 59), personal experience and prior knowledge sensitize the researcher to significant problems and issues in the data, and allow him/her to see alternative explanations and to recognize the properties and dimensions of emergent concepts.

My own practical experience of entrepreneurship and familiarity with the research phenomenon proved very helpful indeed in various phases of the study – in planning the themes of the interviews, in discussions with the entrepreneurs, in interpreting the findings, and in drawing the conclusions. Still, apart from its many positive effects, personal experience may also have some negative influence. Sometimes it may be difficult to study a phenomenon that is very close to one’s own life. For instance, if the researcher’s own feelings are temporarily negative or if he/she has lost belief in the subject of the research – in this case, in family entrepreneurship – this may influence the researcher’s interpretation of the respondents’ interviews. In the case of this study, however, my personal experience was a definite asset because the lifestyle of the entrepreneurs was already familiar to me and the multiplicity of the phenomenon well realized. In any case, pre-understanding helps the researcher to get a clear picture of the problem area and to understand what the respondents are talking about.

In organizational or entrepreneurial research it is normally not easy to gain access to the studied firms or to understand the culture or language of the interviewees. That was not a problem in this study, since it was easy for me as an “insider” to discuss the various themes with the interviewees: the atmosphere was informal and natural, and the

discussion moved on freely. In her study on horticultural entrepreneurs, Levander (1998, 37) also points out that it is important for all parties in an interview to share common cultural ground. Neuman (2000, 355) sees such personal, subjective experiences as part of the field data, valuable both in themselves and in interpreting the events in the field.

Instead of trying to be objective and to eliminate personal reactions, field researchers should treat their feelings towards the field events as research data. A qualitative case study requires description of the whole research process, because there are no exact data to be analyzed. The entire process, covering the interviews and their transcriptions, the personal contacts, and the personality and experiences of both the respondents and the researcher, comprises the data that must be analyzed and interpreted.

It is also important to have informants to help to locate potential interviewees and to make the settings favourable for the interviews. An informant is an insider, a member of the group being studied. The informant is familiar with its culture and has years of experience in that culture, is a key actor in the fieldwork, and develops relations and gives information about the field (Gummesson 1991, 28; Denzin and Lincoln 1998, 59;

Hirsjärvi and Hurme 2000, 59; Neuman 2000, 374). The interviewees for this study were found with help of one principal informant, a K-retailer himself and also one of the respondents.

A letter was mailed to potential interviewees (Appendix 2), in which I introduced myself and my family’s experience as K-retailers. Reference was also made to the principal informant, who is very well-known among K-retailers. The telephone numbers of both the informant and the researcher were also attached in case the interviewees wanted more information about participating in the study. All those who were asked to take part were interested in giving an interview. Only two of the wives were unwilling to be interviewed although both were working in the family enterprise. They wanted their husbands to take care of the “public side” of the business and preferred to stay in the background themselves.

All the interviews were conducted during the spring of 1999 (February to April). The informant couple were interviewed at my house and during one weekend trip, one of the wives was interviewed in her home, and all the other interviews were conducted in the retailers’ store-offices. Grocery retailers tend to be very busy and their offices can be restless, but all the interviewees tried to calm the situation down for a couple of hours.

Although there were a few interruptions during some of the interviews, the discussion was usually resumed as if there had been no interruption.

The interviews were started by filling a “fact sheet” (Appendix 3), a page at the beginning of the written notes with information such as date and place of the interview, characteristics of the interviewee, content of the interview, and so on (Neuman 2000, 368). The fact sheet included some detailed data on the household, family, and enterprise of the interviewees. This first part of the interview was not recorded. According to Hirsjärvi and Hurme (2000, 66) it is not always necessary to discuss all the topics – some can be asked using a short questionnaire. Once the fact sheet had been filled in, the remainder of the interview was recorded. I also made some written notes during the interviews, which proved valuable as the tapes were sometimes of poor quality. Thus, in transcribing the interviews it was possible to check from the notes what was said in a certain situation, if the voice on the tape was blurred.

To build trust and to encourage the interviewees to open up, an interviewer can share his/her background with the respondents and encourage and guide them to express

themselves in the way in which they normally speak, think, and organize reality (Neuman 2000, 370). Accordingly, to make the atmosphere relaxed and unreserved, I first told the respondents about my own experience of entrepreneurship and why I had become interested in studying family businesses. It is easier to lead an informal discussion with interviewees about a topic if the researcher is familiar with the matters discussed (Varamäki 2001, 34). After some introductory “small-talk”, the tape recorder was switched on and the rest of the conversation was recorded. Soon the interviewees became relaxed and the tape recorder did not seem to disturb them. My role as a researcher also was swept aside after I told the respondents about my own experiences as an entrepreneur. The role of researcher was replaced by that of a business family member.

My feeling was that it was almost like as if I had met the respondents quite by accident and we had just started to talk about everyday life, as two members of business families would do.

Eighteen persons were interviewed for this study, and the interviews took two to three hours each. Only in two cases both spouses of the couple were interviewed together.

Consequently, the total length of the tape recordings is nearly 30 hours. Usually free conversation tends to yield almost too much material, but the use of preplanned themes in the discussion made it easier to handle and analyze the material. It has been pointed out that once a theoretical category has been saturated, it is unnecessary to collect further information (Glaser and Strauss 1967 ref. Piotrkowski 1979, 297). Ten cases proved to be sufficient in this study, since the “story of the retailer family” began to shape after these cases had been interviewed. As the conversations and the stories told by the respondents naturally followed the themes that had been assumed to be relevant beforehand, the saturation point was determined to be reached at these ten cases. The viewpoint of lifecycle thinking was present in all the interviews, since it was expected to be theoretically important. Afterwards, towards the end of the analysis process, the subject of succession came up as another potential topic of study. It might have been interesting to include some cases representing an ongoing succession process. However, it was not possible to continue with the interviews later on because the chain system had just been renewed and the respondents would have been in a far different situation than in 1999. It would perhaps also have been difficult to find these cases, since business families in a succession situation may be going through hard times and would hardly want anyone to probe into their private lives.

As described above, the interviews were semi-structured but in most cases it turned out that as the interviewees were telling their own stories, the themes came up automatically without any major prompting by the interviewer. Steyaerd and Bouwen (1997, 50-60) call this type of situation in which the interviewer does not follow a checklist of questions but suggests that the interviewees tell their story, a “storytelling interview”.

Interviewing can be seen as a situation where the researcher elicits stories from the interviewees: the interviewer intervenes and generates rather than collects data. But, as Dennzin and Lincoln (1998, 170) observe, even if the stories are so compelling that the respondents are permitted to speak for themselves, the researcher must not stop there because the researcher’s task is to discover and construct meaning into those texts. Field texts are, in general, not constructed with a reflective intent. Rather, according to Denzin and Lincoln, they are close to experiences, tend to be descriptive, and are shaped around particular events. Research texts are quite far from field texts and grow out of the repeated asking of questions concerning meaning and significance.

In this study, one respondent and his wife also interviewed me after the “official”

interview was over. They wanted to know how certain situations had been solved in our

family enterprise and household. The discussion was very fruitful for this study, although it was not recorded. It increased my understanding and gave more insight into the situation of the entrepreneur.

In the transcription of the tapes, I was assisted by a research assistant, a university student who had previous experience of transcription work. Using another person’s help in transcription is one way to improve the reliability of the study (Varamäki 2001, 35). I transcribed two of the interviews myself, since making one’s own transcripts gives an important opportunity to relive the interview and become substantially more familiar with the data (Maycut and Morehouse 1994, 101). The data have to be listened to over and over again, and the study material has to be collected and analyzed partly at same time (Koskinen 1995, 60-63). The qualitative research process is not so simple as in causal quantitative studies, as the different stages of the process are overlapping or cannot be separated from each other. The function of the literature also differs, and in qualitative research the literature is mainly read after data collection while the analysis is going on. There is not even a precise, clearly defined research problem at the time the collection and analysis of the data starts. The fact that qualitative research cannot be planned beforehand usually leads to a huge body of collected data. According to Ehrnrooth (1990, 39), the general tendency is to collect too much data too early. He states that it is important to have preliminary questions, to decide the rules by which to handle the data, and also to create preliminary rules for interpreting the data before collection is started.

According to Holstein and Gubrium (2002, 125), writing up the findings from interview data is, in itself, an analytically active enterprise. They say that rather than adhering to ideal of letting the data “speak for themselves”, the active analyst empirically documents the meaning-making process. With ample illustration and reference to records of talk, the researcher describes the complex discursive activities through which respondents produce meaning. The goal is to explicate how meanings, their linkages and horizons, are constituted both in relation to, and within, the interview environment. The researcher’s reports do not summarize and organize what interview participants have said, as much as they deconstruct participants’ talk to show the reader both the hows and whats of the narrative dramas conveyed, which increasingly mirrors an interview society.

Coding data has also a different meaning and role in qualitative than in quantitative research. The raw data must be organized into conceptual categories and the themes or concepts created. These categories, concepts, and themes are then used in analyzing the data. Qualitative coding is an integral part of data analysis, which is guided by the research problem and leads to new questions. This frees the researcher from entanglement in the details of the raw data and encourages higher-level thinking about them. It can also help with generalizations and with identifying the theory. (Neuman 2000, 420) In this study, no computer program was used in analyzing the interviews.

Manual coding and analysis was preferable, since the aim was not to seek any precise concepts. The transcriptions were read over carefully and the different themes were marked with different colours. It was difficult to use pre-structured coding of the data because the interviewees told their stories in their own order and in their own words. In the interviews the themes became mixed, and in a situation in which “everything affects everything” it is hard to make any precise structural patterns on the basis of the stories.

The case study method and informal interviews generate masses of information, and it may sometimes be difficult to determine what data to include in the study and what to discard (Chetty 1996, 77, 81). Alasuutari (1999, 40) suggests that in reducing the data they must be examined from only one theoretic-methodological viewpoint at a time. In

doing this, it is important to pay particular attention to what is essential from the point of view of the theoretical framework of the study and from the viewpoint of each research question. While much of the analysis process consists of reduction and division of data into smaller pieces, the final goal is the emergence of a larger picture (Creswell 1994, 154).

The process also involved combining the interviews of the spouses of each case. After this, the data were rewritten, first theme by theme in each case. Then the themes from all the cases were collected together (cross-case thematizing) to see what had been said about each theme. The preliminary idea was to analyze and report the findings by themes, but it turned out that this kind of analysis did not answer the research questions.

Reporting had to be done differently to really find the answers to the questions. The new reporting model was to describe the cases, analyze them, and then make a cross-case analysis by examining each research question separately.

Part of the data were collected by means of SWOT-analyses (the acronym stands for strengths, weaknesses, opportunities, and threats). A SWOT-analysis was compiled for each case on the basis of separate SWOT-analyses made by each spouse. This analysis method is commonly used in marketing, and the idea is to find the point at which the strengths of a firm coincide with the best opportunities in the marketplace. The analysis outlines a unit’s differential strengths and weaknesses and its present and future threats.

(Blois 2000, 444 and 462) The SWOT-analyses of the cases are presented in connection with the case vignettes, which describe the cases based on the three-circle model of family entrepreneurship. First, there is a description of the family structure, then of ownership structure, the family members and their work roles, and the division of labour in each of the cases. Finally, the SWOT-analyses made by the spouses are presented in the case vignettes. The case vignettes describe those characteristics that were considered most informative in each individual case. The next section introduces the cases by giving a short case profile of each one of them. These initial profiles are rechecked and modified in the interpretation process into the final case vignettes in Chapter 5.