In the qualitative method the data is usually either verbal or visual (Uusitalo 1991, 79), and are obtained by means of three kinds of data collection: in-depth, open-ended interviews, direct obser-vation, and written documents (Patton 1990, 10). Usually the data is so-called spoken data (Pyörälä 1995, 19). The interview as a method of data collection is flexible, allowing the inter-viewer to ask further, more defined questions, and clarify unclear questions and answers (Tashak-kori & Teddlie 1998, 102). On the one hand it allows the researcher to observe the body language and emphasis of the respondent, but on the other hand it allows also the researcher through his/her expressions, stress, or words, unintentionally to influence the answers (Tashakkori & Teddlie 1998, 102).
In this study the data collection was based on the individual face-to-face theme interview-ing method, but also on written documents (i.e. brochures, statistics, figures and plans given by the case firms, articles in newspapers, professional magazines, web pages, and register information).
A theme interview was suitable in this case, as the interest was to discover the interviewees’ opin-ions on the subject. Moreover, theme interviewing was suitable in this study because although the subject matter, MT work, was familiar, the themes within it were not. The first interviews opened
up several alternatives to be studied, but one theme, MT learning, emerged as the strongest. Thus theme interviewing and theme analysis offered the most suitable way for the gathering of deeper information about the cases. In addition, the informants were able to speak the theme in their own words. The flexibility of theme interviewing and analysis was also an important criterion in the selection of the suitable method. Rather than their responding to detailed questions, it was more important that the informants were able to talk freely about the subjects within the selected theme.
Moreover, the theme interview method fitted well to this study, because it tends to examine points of which the interviewees are not aware or to which they did not pay attention to. In order to attain a holistic picture (Hirsjärvi et al. 2004) and a deeper understanding of the phenomenon, it was es-sential to let the data speak. The empirical part of the study began with a pilot survey, which was also an important part of the learning process for the researcher. The main reason for the pilot sur-vey was, however, not only to test the form of the theme interview, but also, and particularly, to gain greater understanding of MT work in practice in a family-owned SME. The pilot case assists the researcher to develop relevant lines of questioning, possibly also even providing some concep-tual clarification for the research design (Yin 1994, 74). The pilot survey was carried out in a simi-lar firm in the same industry in October 2004.
The first extensive interviews were conducted with the CEOs in October 2004 and June 2005. The CEOs were asked, for example, to speak about themselves, their firm, work, MTs, and goals. The objective of this was to gain a holistic picture of the firm, the CEO, and the MT. The second interviews for the CEOs were carried out in August and September 2005 and they dealt with more specific information about their MTs, such as MT forming, MT meetings, and MT members. Thereafter, in autumn 2005, multiple informants were interviewed (every member of the MTs, including also certain previous MT members), and both semi-structured interviews and open questions were conducted. Open questions were aimed mainly at the CEOs but also at other MT members in the first interview. The objective of open questions was to allow the informants to talk about themselves, their firm, the MT, and the issues which they considered to be most important in their work: this produced a full and rich array of such issues. The interviews also helped to regen-erate and refine the research questions and to focus the research subject. Moreover, the previous interviews together with the first analysis based upon them, functioned as a base for the third in-terview with CEOs and for the second inin-terview with other MT members. At this time the more specific questions (i.e. the semi-structured interviews) could be formulated during the research process, as the uniqueness of the case and the phenomenon of interest became more familiar to the researcher. The second theme interviews, focusing on MT learning and succession, were con-ducted in autumn 2006, when the founders of the case firms were also interviewed.
Interviews with the MT members lasted from 30 minutes to 2 hours, and they were car-ried out mainly in the meeting rooms of the case firms. The place was familiar to the interviewees and it was they who also decided upon the suitable time for the interviews. After several visits to the firms and the first-time introduction to the premises and the factories by the CEOs, the re-searcher began to feel familiar with the case firms. One predecessor was interviewed at his home and two MT members were interviewed on the telephone on account of distance (one of them had left the firm and worked far away, and the other was on a maternity leave). However, all inter-views were recorded, and on each occasion the informants’ permission for the recording was re-quested. There were altogether 9 MT members in the case firms (5 MT members including the
CEO in Varkauden Metallityö Oy and 4 MT members including the CEO in Metallipojat Oy). The total number of interviewees was 11 and that of actual interviews was 21. The CEOs were inter-viewed altogether three times, other team members twice and the predecessors once (a table of the interviewees with the dates and duration as a piece of empirical research is presented in Appendix 4). Shorter follow-up sessions were conducted wherever necessary in order to clear up factual con-fusion and to verify perceptions and interpretations. Notes were taken both during and after the interviews, and the interviews were recorded using a mini-disc recorder. There were nearly 104 pages of transcripted text. Data collection in this study relied also on the reports of Asiakastieto Oy and Inoa Yrityshaku registers and written documents (e.g. brochures and other written material from the case firms and newspapers), as presented in Table 13, in addition to discussions on the telephone and by email.
Table 13: A summarising table of the collected data
Data Hour (app.) Page (app.) Piece
In this study information from the cases was collected in order to gain a more profound under-standing of the MT learning of two different firms, with the intention of comparing the two MT learning cases and examining their differences and similarities. It was interesting to see to what extent the reasons for the MT learning were similar, and in which aspect they differed from each other. Thus the comparative case study method was suitable for this study. Data collection and analysis were inseparable from each other, because every attempt to organize and label the empiri-cal material required some kind of interpretation. After the first interviews the data was catego-rized thematically, interpretation being included at this stage. According to Metsämuuronen (2003, 196) it is very common in qualitative research for the data collection and analysis to take place simultaneously. Miles and Huberman (1994) identify three phases interacting with the actual data collection: reducing the data, classifying, and making conclusions. The first of these phases in-volves the reduction of a large quantity of information to more manageable proportions by means of thematization. The next step is to examine the answers of these themes followed by classifica-tion of the material in order to draw the conclusions. In practice this means comparing the various topics and the answers in order to identify the differences and/or the similarities between them.
Finally conclusions are draw from the data in order to clarify the total meaning, and to evaluate the reliability of the interpretations (Miles & Huberman 1994, 10-12). This is the method by which the data was processed in this study. Processing qualitative data is largely manual work carried out by
the researcher, thus it is important for the credibility of the study that the processing of the data, in addition to the choices and their justifications, should be described in detail (Pyörälä 1995, 21).
The data analysis of this study began after the first interviews (those with CEOs and other MT members) with the analysis of both firms separately (i.e. within-case analysis). The purpose of this description was to construct meaning by linking empirical patterns (themes, events and proc-esses) in order to form a holistic configuration, i.e. the case (Stake 1995, 78). The data was exam-ined in as empirically-based a manner as possible, in order to allow different themes to emerge (Eskola & Suoranta 1998, 153). The data was systematically studied, and each topic was marked with a different colour during, as an aid the structuring and clarification of the contents. The cod-ing was to facilitate the examination of the variety of data, i.e. to study which topics found in the data were related to the research topic. Coding facilitated the processing and analysis of the data, as similar parts of the text were easily found (according to the code). The coding was done by hand with colours and side notes, the aim of using different colours being to identify different top-ics in the text. The side notes showed the key words of the contents. Manual coding seemed a natural choice to the researcher, there being no programs available for the processing of qualitative data. Dividing the interview text into themes or discussion topics helped towards the recognition of central matters and the creation of connections between them. Gathering the rejoinders under the same theme simplified the interpretation and summarised the opinions of the respondents on the different themes in both firms. The same piece of text may have been classified under several categories, if it referred to several themes.
Many interesting themes were discovered in the data, such as reasons for MT work and forming a MT, responsibility, commitment, flow of information, co-ordination, firm growth, de-velopment of MT work, ownership in MT, turnover, conflicts, power in MT, and decision-making:
these were classified on the basis of the themes. Many of these, although interesting as such were widely studied phenomena. At first MT learning seemed to be somewhat on the periphery of the discussions but repeated attention to the data revealed that the subject recurred in connection with certain other themes. MT learning seemed to be an extraordinary and important phenomenon in both the case firms run by new successors and their MTs. This fascinating phenomenon demanded examination, and it thus formed the main research question of this study and the theme of the final interviews.
The analysis of the final data (including the entire interview material) was conducted as follows. The taped text was transcribed, by which is meant that all interviews were transcribed verbatim. The relevant was separated by classifying the sections of the interviews into many dif-ferent groups and sub-themes which clarified the research problem and helped towards structuring the text. This was done by picking out those sections of the text which discussed the current theme, and by arranging them under the different themes. This was conducted with the copy-paste function, which also allowed the use of the same text under several themes. It was important to organize the material chronologically in order to show and maintain the order of the events. Over time, changes had occurred both in the MT work and in their membership, and it was important to be able to examine these changes in the correct order. However, although the data collection was made temporally, the analysis of the cases did not follow the same chronological order, because it was not possible to present chronologically the themes arising from the discussion. However, chronological order was used where possible. The data was organized gradually, and the
theoreti-cal framework assisted in the processing of the empiritheoreti-cal observations. At the same time the rela-tionship between the separate answers and the theoretical framework was beginning to take shape.
Quotations were used as a basis for interpretations, for example, to justify an interpretation, to de-scribe the data, to invigorate the text, and to simplify the story (Eskola & Suoranta 1998, 173).
After this it was easier to identify the issues which the MTs had learned, and to see the whole body of data from the viewpoint of the study, i.e. MT learning.