Research Methodology

The research method we chose to find answers to our research questions and to derive further research hypotheses is discussed in this section. We chose a qualitative approach to explore the nature of project uncertainty, project value opportunity, and the role of project stakeholders across different projects within different industries. Our empirical results also suggest that it is necessary to differentiate opportunities, and to better understand their gestalt. Also, it is not clear which contingencies are leading to specific types, and if opportunities with specific characteristics are preferred. The latter is a question that the literature in innovation management is concerned with. Some prescriptive studies demonstrate that managers prefer incremental innovations over radical innovations simply due to the misjudgment of the outcomes. Hence, this research addresses the gestalt of opportunities that occur during a project's implementation. Due to the lack of operational models to measure the nature of project opportunities, the only possible method for researching this concept is an exploratory, qualitative study.

Due to its exploratory nature this research employs a hybrid approach reflecting both positivist and interpretive perspectives (Kirsch, 2004). It was designed and conducted following the recommendations of Eisenhardt (1989), Yin (2003), and Corbin and Strauss (2008).

Data Collection and Sample Description

The primary means of data collection for this study was semi-structured interviews augmented with project documentation. The interviews were conducted between July 2010 and February 2012, and resulted in 41 in-depth cases. All interviewees were project managers of the projects under investigation. The interviews usually ran about 90 minutes and were conducted by at least two researchers. We refrained from taping the interviews because most interviewees felt uncomfortable with being recorded. To increase accuracy, notes taken during the interview were reviewed immediately afterward, and case write-ups were created and confirmed by the interviewees before the data interpretation was conducted.

The case study protocol and database were created according to recommendations for establishing construct reliability and validity (Yin, 2003). The case study protocol specified the procedures and the questionnaires for collecting data (see Appendix). The interviews were comprised of four segments. The first segment gathered general information about the industry, size, and general products of the organization. During this segment the interviewees described one of the projects they recently managed, in which they perceived unexpected events. Stakeholder structure, planned and actual budget, schedule, and scope information were collected. The interviewees were also asked about the degrees of innovativeness and complexity of the project. The second segment was an open discussion with the interviewees to identify up to three major situations of project uncertainty. In the third segment, the interviewees were asked to name two situations of uncertainty in which opportunities were discovered to improve the project's value. The sources of uncertainties and the means in which opportunities were discovered and exploited were discussed in detail. In the fourth segment, the interviewees discussed opportunities which were discovered but not exploited. The specific questions were modified slightly as the interviews progressed. For example, project managers were initially confused by the concept of uncertainty, so we added explanations to clarify the question.

The database for each case included raw field notes and detailed case study write-ups of approximately 10 to 15 pages, created after data collection, to summarize the facts of the cases. Case analyses of at least two hours, with at least three researchers, were conducted to establish the chain of evidence and to improve construct validity, resulting in two to four pages of interpretation comments. All case descriptions presented here are based on case write-ups and debriefing reports from the database. After closer analysis we chose to remove one case from the database because the information offered was insufficient.

The investigated projects are categorized in Table 2, following the classification of Archibald and Voropaev (2004). The schedule objectives of the projects range from three months (Case 29) to 36 months (Case 01), with a range of budget objectives from $200,000 (Case 59) to $69 million (Case 06). 23 projects (55%) were claimed to be highly complex, 11 projects (26%) had a medium complexity, and 8 projects (19%) were classified as low complex.

Table 2: Project Categories of Cases

Data Analysis Method

The study followed the suggested protocols for improving analysis speed and identifying needed adjustments to the data collection procedures by overlapping data collection and analysis (Eisenhardt, 1989; Corbin & Strauss, 2008). Debriefing meetings were arranged as soon as possible after the case write-up and accuracy was confirmed by the interviewee. It turned out that some cases needed more clarification, and the interviewees were contacted to assure a complete data set. All four researchers participated in the debriefings, and each essential construct was discussed and interpreted. The resulting interpretation report of the case data recorded the project managers' perceptions of the project and the researchers' perceptions of the mindset of project managers. The report also included the sources of uncertainty, discovered opportunities, knowledge areas, and any malmanagement or mismanagement in the project.

The case write-ups and interpretation reports were coded following a multistep coding process by Corbin and Strauss (2008): open coding, axial coding, and selective coding. Upon completion of an interview, open coding was conducted for each opportunity and uncertainty situation and was iteratively conducted for each case. Axial coding was taken to classify previously identified concepts into categories at the completion of ten cases. The findings in the first round of interviews then guided, but did not limit, the analysis for the next round, in order to confirm and complement existing findings. A selective coding process was used to integrate and refine those findings to create the theoretical contribution of this study. A cross-case analysis was conducted, and it is included in the case study results section below.