CRITICAL SUCCESS FACTORS OF OFFSHORE SOFTWARE DEVELOPMENT PROJECTS

Index

Research Objectives
Research Methodology
Research Process

Within the following sections, we first outline the research objectives of the chapter at hand. Then, we introduce the research methodology applied and describe the research process implemented in order to reach these goals.

In German-speaking countries, OSD still represents a relatively new and emerging area of research. As a result, barely any research studies exist, which focus solely on the CSF of OSD projects (compare with the conducted literature review in section 5.2). For this reason, the main objective of this chapter is the identification of those factors which significantly influence the successful implementation of an OSD project. In this context, we focus on the perspective of German-speaking clients of software services. In doing so, primarily the following research question will be addressed: “Which CSF do companies have to consider when implementing OSD projects?”

In order to identify CSF, a wide array of research methods is used. Among them are, for instance, the realization of case studies (e. g., Sumner, 1999), group interviews (e. g., Khan­dewal and Miller, 1992), structured interviews (Rockart and Van Bullen, 1986), as well as the analysis of relevant literature (e. g., Esteves and Pastor, 2000). According to Shah and Siddiqui (2002), the most frequently used method to identify CSF is the realization of a questionnaire.

In order to create a deeper understanding of the CSF of OSD projects, we decided to use a qualitative research design. Due to the fact that important influencing factors currently still remain unknown (Creswell, 2003) and that existing theories and models cannot be applied to the precisely examined topic (Morse, 1991), we felt this was an appropriate approach.

For the qualitative research design, we selected the so called Grounded Theory in accordance with Glaser and Strauss (1967). According to Pandit (1996), this method is particularly suitable when a theory cannot or should not be established deductively. Rather, a theory developed in line with this type of research method is “inductively derived from the study of the phenomenon it represents. That is, discovered, developed, and provisionally verified through systematic data collection and analysis of data pertaining to that phenomenon. Therefore, data collection, analysis, and theory should stand in reciprocal relationship with each other. One does not begin with a theory, then prove it. Rather, one begins with an area of study and what is relevant to that area is allowed to emerge.” (Corbin and Strauss, 1990, p. 23)

The fact that, when applying the Grounded Theory method, a theory is not defined in the front end and then continuously verified, but instead developed over the course of time, allowed for an impartial approach during the investigation of the different CSF.

For us, another important attribute of the Grounded Theory method was that it simplifies “the generation of theories of process, sequence, and change pertaining to organizations, positions, and social interactions” (Glaser and Strauss, 1967, p. 114). This proved to be of importance, especially regarding the investigation of CSF for OSD projects, as these are many a time influenced in great deal by organizational aspects. This assumption is confirmed by both the results of respective CSF studies in English-speaking countries (e. g., Delmonte and Mc­Carthy, 2003) as well as the results of the chapter at hand.

In regard to the collection of relevant data, we came to the decision that the conduction of expert interviews was the most appropriate method for examining the CSF of OSD projects. According to Gläser and Laudel (2004), the conduction of expert interviews is a specific research method which aims to access the knowledge of experts about certain circumstances. This method is particularly used in social sciences and is characterized by two important attributes:

1.      Experts are a medium by which researchers want to obtain knowledge about relevant issues. Here, experts are not themselves the object of study, rather they are or were witnesses of important processes which refer to the object under study.

2.      Experts have an outstanding and sometimes exclusive position in the context under investigation. For instance, employees are interviewed in order to collect information on the internal structure and processes of their companies.

In consideration of these two attributes, the authors define expert interviews as the reconstruction of situations or processes in an effort to find scientific explanations. In this context, the role of the experts is to make specific knowledge accessible. For this reason, according to Gläser and Laudel (2004), expert interviews are the method of choice when specific circumstances should be reconstructed.

For data analysis within the scope of the Grounded Theory method, the so called coding process can be used. This process can be composed of up to three consecutive types of coding, namely open, axial, and selective coding (Corbin and Strauss, 1990). According to Pandit (1996), these types must neither be executed in a particular succession, nor must all three be executed completely: “There are three types of coding: open coding, axial coding, and selective coding. These are analytic types and it does not necessarily follow that the researcher moves from open through axial to selective coding in a strict, consecutive manner.” (Pandit, 1996)

By means of open coding, usually the first step of the coding process, various concepts and categories are developed from the collected data. Hereby, data is compared, and identical or similar statements are combined to form specific concepts. Closely related concepts are then pooled once more to form categories (Pandit, 1996). According to Pandit (1996), the formulation of the various concepts proves to be of the highest importance in this context, as they represent “the basic building blocks in grounded theory construction”. Axial coding is responsible for the structured assembly of the newly formed data. Thereby, categories and their sub-categories are connected and set in relation to one another (Pandit, 1996). Through selective coding, the identified categories are connected to one another. Here, one category is defined as the core category and the others are connected to this one core category (Esteves, 2004).

It must be noted that, in the course of our data analysis, only open and axial coding was used.

Figure: Iterative research process implemented

The implementation of the Grounded Theory method is characterized by an iterative process (Pandit, 1996). This is reflected particularly in our selected procedures for the preparation of the interviews as well as for the data collection and analysis. A total of three iterative loops, all building up on one another, were run through these pivotal research steps (compare figure above).

Regarding the research process, the following table outlines the five major research steps.

Table: Description of research steps

In the following, the research steps mentioned above will be described in more detail.

Literature Review

Regarding the procedure used in analyzing the state-of-the-art of CSF research in the field of IT outsourcing, reference can be made to section 5.2.

Preparation of Interviews

Phase 1

During the conception of the interview guidelines, an effort was made to consider all of the relevant aspects previously identified with the help of the literature review.

Generally, the interview guidelines consisted of two parts: The more general first part relied on a variety of both open- and close-ended questions and was responsible for the collection of general information on the interview partners. On the contrary, the second part of the questionnaire was more specific and focused on the CSF of OSD projects. Here, open-ended questions were used primarily to prevent the manipulation of the interviewees and to ensure that the statements made possessed a certain grade of authenticity.

Phase 2

After having finished and evaluated the first ten interviews, a modification of the interview guidelines, which affected exclusively the more specific part, was carried out. In the process, frequently mentioned CSF were added to the compiled list. This resulted in a partially structured questionnaire which meant to motivate future participants to extend the preliminary list with the addition of not yet mentioned CSF.

Phase 3

In order to verify the evaluation results, we developed a questionnaire in a MS-EXCEL format during the second analysis phase. The basic reason for this was to confirm that the interview partners accepted the identified list of CSF. Here, the participants also had the opportunity to either comment on the individual CSF and/or suggest additional ones.

Data Collection

Within the framework of the survey, a total of 22 participants were interviewed. When selecting the interview partners, it was taken into consideration that, in order to create a most comprehensive understanding of the successful implementation of OSD projects, a variety of company perspectives (client, provider, and consultancy) needed to be taken into account. Additional requirements, which influenced the selection of the participants, were their position in a company and an OSD project respectively as well as their personal experience with OSD.

Phase 1

The first round of questioning consisted of ten 30 to 60 minute interviews, all of which were conducted by phone within a time span of two weeks. In order to reduce the loss of information, two members of the research team took part in the interviews at all times. In an effort to ensure the open nature and the authenticity of the statements, we decided to keep written records of the relevant contents, rather than to record them (Urquhart, 2001).

Phase 2

After having finished and evaluated the first phase of interviews, twelve further interviews took place within the following three weeks. Eight of these were carried out by phone, while the other four were face-to-face interviews.

What really stood out in this second round of questioning was the fact that a certain saturation regarding the suggestion of not previously mentioned CSF was able to be observed. Especially towards the end of the second round, hardly any new insight was gained from the interviews. This resulted in the finalization of the collected data after 22 individual interviews. According to Carvalho, Esteves and Ramos (2002), this particular procedure is typical for the implementation of the Grounded Theory method as soon as saturation is observed.

Phase 3

In order to complete and verify the first results of the analysis, the interviewed participants were confronted with the developed list of CSF, sent to them via e-mail in the form of a questionnaire.

Data Analysis

Phase 1

Due to the iterative nature of the Grounded Theory method (Pandit, 1996), a temporary analysis was conducted after the first ten interviews in order to give way to possibly necessary changes within the research design.

Orlikowski (1993) applies a similar iterative procedure in her studies on CASE-Tools, and, in doing so, emphasizes the advantages of such a procedure: „Data collection, coding, and analysis proceeded iteratively (Glaser and Strauss, 1967) with the early stages of the research being more open-ended, and later stages being directed by the emerging concepts, and hence involving more strategic selection of informants and more structured interview protocols.” (Orlikowski, 1967, p. 313)

It must be noted that this initial data evaluation was partially influenced by the perceptions made within the literature review.

Phase 2

The evaluation of the collected data took place in accordance with the Grounded Theory’s coding process. This process “represents the operations by which data are broken down, conceptualised, and put back together in new ways. It is the central process by which theories are built from data.” (Corbin and Strauss, 1990, p. 57)

First, all of the 466 statements made in the 22 interviews were entered in a MS-ACCESS database. Next, the CSF list was derived from the collected data by means of open and axial coding. The implementation of open and axial coding allowed for the conversion of the collected data into a categorized CSF list (compare example in table below). Here, in a first step, the collected statements, which varied only slightly in regard to their syntax, were aggregated (first aggregation level). Thereby, a list of 250 different statements was gained. This list, in turn, was used as a basis for further evaluation. In a second step, the list was reduced even more through the aggregation of semantically identical statements (second aggregation level). The result was a list of 90 statements, which again was scaled down in a third and last step to a list of 28 CSF (third aggregation level).

Table: Example for the implementation of the coding process

As was the case during the first data evaluation process, the perceptions made within the literature review were also considered during the second and final evaluation of the interview data.

Phase 3

The questionnaire, sent to the various interview partners in an effort to verify the initial list of CSF, was meant to give insight into whether or not the participants considered the list to be representative or if they felt important factors were missing from the list.

Twelve of the 22 interview partners filled out the questionnaire and sent it back to us. This equals a return rate of 54.54 percent. Due to additional information gained from this feedback, the list of CSF was adjusted and individual formulations were modified.

Comparison with Related Studies

Based on the analysis of similar studies, which dealt with the CSF of IT projects and the issues and problems of OSD projects, three contrastable studies were first selected. Second, the content of these studies was directly compared to our CSF list. Third, due to the additional insight gained, one further CSF was added to our original list.

References