Gagne de la cryptomonnaie GRATUITE en 5 clics et aide institut numérique à propager la connaissance universitaire >> CLIQUEZ ICI <<

3.4. Data collection

Non classé

Having established that a qualitative methodology was most appropriate for this study, the in-depth interview was considered an appropriate investigation tool for this research. The use of interviews in my case study helped me to gather valid and reliable data that are relevant to my research questions and objectives (Saunders et al., 2003). Also sensitivity in face-to-face discussions can elicit information from respondents that would otherwise not have been prepared to give. Such respondents would be reluctant to provide key qualitative statements on written questionnaires or in verbal answers to telephone calls.

According to Patton (1990), the in-depth interview provides a framework within which respondents can express their own understandings in their own terms; it assists in valuing the meaning and significance which people give about their experiences. Jones (1985) concurs that this method enables the researcher to understand other persons’ constructions of reality, by allowing interviewees to respond in their own language. Besides, according to Tull and Hawkins (1990), the in-depth interview is the most suitable method to probe in detail an individual’s behaviour, attitudes or aspirations.

When implementing a qualitative methodology, interviewer and interviewee biases are important issues regarding the use of interviews. Interviewer bias relates to the possibility that comments, tone or non-verbal behaviour creates bias in the way that interviewees respond (Saunders et al. 2003). According to Easterby-Smith et al. (2002), it is possible that the researcher may be biased in the manner in which responses are interpreted. This may be caused by perceptions about the interviewer or in connection with perceived bias from him. However, Interview bias is not always related to the interviewer, there may also be instances in which the interviewee does not wish to explore certain themes and thus only gives partial responses in order to make his company’s position appear more positive. It was therefore important for me to approach each interviewee personally to establish trust, consent and rapport. In order to overcome interviewee bias, Saunders et al. (2003) propose to follow some key measures, some of which are:

 Visiting each potential interviewee in person to establish rapport, to explain the reasons for the study and to familiarize oneself with the situational context in which the interview will take place.

 Providing a letter of introduction presenting an overview of the research and indicating my issues of interest (See appendix 1).

 Asking the interviewee to contact me, should there be any concerns or doubts the interview or the agreed meeting.

 Allocating time at the beginning of the interview to thank the interviewee for his participation, explaining the confidentiality and anonymity of the research and providing a brief overview of my background. These assurances increase the level of trustworthiness and reduce the possibility of response bias (Healey and Rawlinson, 1994).

 Treating each interviewee with respect and sensitivity

 Being dressed appropriately for each interview.

 Using active listening and clarifying any points that may be unclear.

 Taking handwritten notes that should be compiled into a full record shortly after each interview to control bias and produce reliable data for analysis (Robson, 2003).

The interview type, which was chosen for this research is a semi- structured interview. Completely unstructured interviews were considered inappropriate because they might have inhibited the flow of information and might have prevented the emergence of useful and revealing insights (Jones, 1985). There is also a higher risk that the interviewer would fail to glean the desired information (Robson, 2003). Whereas, semi-structured interviews enable the informant to discourse freely on topics capturing his interest; both flexibility and structure are achieved (Johannessen and Dolva, 1995). With such a method, I exploited the key aspects, of the offshoring practice, emerging from the literature review to guide the interview process. Riley (1996) supports this approach, by recommending that formally structured questions should be minimised and any subsequent interventions should be made in the form of prompts and probes, based on the informant’s words. Thus the development of the interview questionnaire was perhaps the most difficult task of the whole research project. It was essential that the questions asked were open-ended and did not limit the responses of he participants. The sequence of the questions was carefully analysed to ensure that the preceding questions did not influence the respondents’ answers.

Due to the limited scale of the phenomenon of IT offshoring in Morocco and given time and cost constraints, a non-random purposive sample of four businesses was chosen. Purposive sampling ensured that cases were varied and rich in information (Marshall and Rossman, 1989). When using purposive sampling, the composition of the sample is determined only after the researcher’s entry into the field (Hill and Wright, 2001). For example the first firm was chosen for convenience, as I had a work experience in this company and I was already familiar with employees and managers. Having completed this initial case study, I used snowball sampling to compile a list of potential firms from which the next company was chosen. I tried to select a heterogeneous group of firms in order to describe and explain key themes within these companies. I managed to conduct four case studies: in June 2006, I visited two companies in France that have been involved in a offshoring IT projects in Morocco. In August I conducted two on-site interviews in Morocco with Spanish and French companies that opened affiliates on Morocco to conduct offshoring activities.

The logic underlying a multiple cases approach in is one of ‘reproduction’, where one seeks to show that a particular phenomenon is likely to be observed, being given a combination of specific factors predicted by the theory (literal reproduction). Moreover, the idea is also to show that this same phenomenon will not be observed in the absence of the specific combinations of factors (theoretical reproduction) (Miles and Huberman, 1994).
Each Semi-structured interview lasted about 30 to 45 minutes. And a mix of open, probing and closed questions were deliberately utilized to gain the relevant information for the study; a set of factual questions are included as an introductory section (see Appendix 2). Not all questions in the protocol were addressed.

Where necessary, I guided the interviewee to discuss my themes of interest. Relevant issues that were raised by previous respondents were also introduced when the interviewee did not discuss such issues ‘voluntarily’. In this way, a degree of deduction or confirmation and/or disconfirmation was introduced as recommended by Carson et al. (1998). The respondents were encouraged to add any nuances or comments that they deemed appropriate. Follow-up questions and requests for precision complemented the interviews.

Moreover, the interviews were recorded on a digital device; except in the case of one interview, where only notes were taken because respondent was not willing to be recorded. When the interviews were completed, the notes were written up within 24 hours. This yielded a large volume of accurate data.
These interviews provide preliminary evidence to support the proposed theory. Findings based on the interview transcripts from the three case studies are reported in the next chapter.

In addition to primary data collected from interviews, secondary data such as information on the software industry in Morocco and an overview of the offshoring activities in the country has been obtained through contact with other major practitioners of offshoring activities and regular scans of business newspapers and websites. Information collected enabled to fill some gaps in the case studies and gave me a better overview of the current situation. The use of varied data sources also permits a certain level of triangulation (Benbasat et al., 1987) and helps to counter certain biases, such as social desirability or selective memory of events.

Page suivante : 3.5. Validity of the research