Chapter Listing | Return to 1997 Topical Index


METHODOLOGY

The Sample

    Given the difficulties in comparing performance across industries (Hirsch, 1975, Chakravathy, 1986) only firms from a single industry, biotechnology, were chosen for this study. All of the firms have faced approximately the same pressures and regulations from the external environment.  The biotechnology industry was chosen for two reasons.  First, it has a relatively short history (most firms are less than 20 years old); this allows the investigation of the entrepreneurial phases of the individual firms.  Second, the industry has a large number of strategic alliances engaged in new product development; all firms were either in alliances (96%) or wishing to enter alliances (4%).

Data Collection

    Denzin defines a survey as a “methodological technique that requires the systematic collection of data from populations or samples through the use of the interview or the self–
administered questionnaire” (1989: 144).  Both methods were used in this study.  Initially, unstructured interviews were conducted with four young biotechnology firms engaged in alliances with a larger partner.  This was undertaken to identify areas to investigate in questionnaires and more formal interviews.  A questionnaire was designed using Dillman’s (1978: 119–159) Total Design Method; these were mailed to all of the firms in the sample.  Seventy six completed questionnaires were returned for a response rate of 32%.  This is consistent with the 30–35% response rate which was expected (Milliken, 1990).

    Additional data was collected by in–depth, on–site interviews with business development executives in 25 of the questionnaire respondents; the interviews consisted of 40 questions modified from Larson (1992) and Murray & Siehl (1989).  The interviews have been used to address the question of what resources were employed in the alliance exchange over time, why they were employed, and what their value was to the exchange.  We have relied on readings of the transcripts, and are aware that these quotations do not fulfill the rigorous requirements of computerized content analysis (Morris, 1994).  Common method bias was also a concern.  The many types of measures used, however, limits the potential for common methods bias (Eisenhardt & Tabrizi, 1995), and as Wagner and Crampton (1993) observed in their review of numerous studies, the problem of common methods variance is often overstated.

Analysis

    An exploratory factor analysis was performed to identify the constructs underlying our set of observed variables (Hughes, Price & Marrs, 1986); five factors were identified. This solution provided the best fit for the data.  Originally a six factor solution was obtained based on obtained eigenvalues of over 1.0.  However, the six factor solution provided no additional interpretive power when compared with a five factor forced orthogonal solution.  In the six factor solution partner’s managerial and technical skills loaded onto two separate factors; these skills were combined into a single factor when the five factor solution was obtained.  A varimax rotation was performed for ease of interpretation, with the five factor  solution explaining 70% of the overall variance; the five factor solution produced highly stable data.  Table Four shows the variables loading on each of the five factors after the varimax rotation.

    The analysis produces a clean factor structure with items loading on the appropriate factors; one item was deleted because of incorrect loading.  The measures of the constructs show excellent validity.  Additionally, internal reliability tests showed strong Cronbach alphas ranging from .5445 and upward.  As a further validity check, the sample was split randomly and Cronbach alphas were recalculated.  Alphas continued to be excellent with a range of .5445 and upward (Hair, Anderson & Tatham & Black, 1995).

    The second part of this analysis was to test the hypotheses performing an ordinary least squares regression.  The standardized factor scores were entered as the independent variables and survey questions related to success of the alliance and performance of the entrepreneurial firm were entered as the dependent variables.

Top of page | Chapter Listing | Return to 1997 Topical Index


1997 Babson College All Rights Reserved
Last Updated 03/23/98