Knowledge Sharing among Scientists

The performance of a firm is influenced by its utilization and accumulation of intellectual resources. Coase (1937), Demsetz (1967), Knight (1921), Stigler (1961), and, more recently, Baumol (2002) and Loasby (1999) have argued that the firm exists to organize the utilization of knowledge. Firm performance is realized through innovation or the entrepreneurial use of knowledge. This leads to an emphasis on the management of intellectual resources including non-codified knowledge within the firm. The intrafirm flow of knowledge is a means to gain efficiency and to innovate. Though many firms have adopted policies to encourage the collection, storage, and dissemination of codified knowledge that resides within the firm, what is less certain is how to foster the exchange of knowledge that resides in individuals. One item that may facilitate or retard the voluntary sharing of non-codified knowledge is the possible receiver’s reputation. The decision to provide another with assistance is based in part upon the signal that his/her reputation sends. That is, the person holding knowledge (the source) renders a decision to provide or not provide the requested knowledge based on an estimate of the past behavior and anticipated future actions of the would-be receiver (the recipient). Through this process, reputation affects the decision to share or not share personal scientific know-how.

This chapter analyzes extant theory and research on the subject of social mechanisms regulating behavior in resource exchange. The model of reputation and knowledge sharing proposed in chapter 1 is formally developed. Existing literature directs the formulation and exploration of the phenomenon in the present research, that is, exchange of resources in a social environment. From the literature on transactions within a social context, elements for the model are reconciled and explicitly drawn together and three primary hypotheses emerge. It is the restructuring of findings from previous investigations that direct this research.

This chapter examines the sample of knowledge-sharing events and shows the basis for the construction of the variables; an analysis of responses is conducted and independent, dependent, and contextual variables are reviewed. This examination not only presents the data but also arranges it in such a manner that it is more readily usable for logistic regression and testing of hypotheses (chapter 6). This systematic review of the data provides informative findings. While most findings confirm existing beliefs, some nuances are presented including areas for further exploration of the data (figure 4.1). Variable construction is based on (1) considerations of measurement and (2) construct validity. In particular, factor analysis is used to screen data for subsequent analysis rather than to develop hypotheses. Correlation tables generate evidence of linear relationships. Cronbach alphas measure reliability (i.e., how well questions coalesce).

This chapter examines the relationship between contextual variables, which serve as a standard of comparison, and the dependent variable knowledge-sharing occurrence. This chapter provides a point of reference for hypothesis testing conducted in chapter 6, where models are finalized and relationships among variables further explored. Descriptive statistics form the basis for analysis of relationships among variables in this chapter. Exploring the impact of individual baseline variables on the dichotomous variable—sharing of scientific know-how—is undertaken systematically. Figure 5.1 depicts the general classes for the variables of interest.

This chapter reports empirical tests of the primary hypotheses. Presented are the results of multiple and logistic regression analysis. Two specifications for distinct groups of models are developed. Analysis of these two specifications as well as progression from simple models to more complex models demonstrates robustness. Hypothesis 1 is explored first through multiple ordinary least squares regression. Hypotheses 2 and 3 are tested through binary logistic regression. Baseline models (Model 1 in both specifications) contain those contextual variables that hold statistically significant explanatory power. The second pair of models contains the past behavior dimension of reputation plus the contextual variables. The third set of models includes the expected action dimension of reputation in addition to the contextual variables. Finally, Model 4 in both specifications incorporates all of the variables.¹

The results of this study support the thesis that reputation plays a role in the decision to share technological knowledge. Among R&D scientists in the pharmaceutical industry it was observed that past behavior and expected action were considerations in making the decision to provide scientific know-how to a fellow R&D employee in the firm. This research corroborates and clarifies social exchange theory regarding the role of social factors in the exchange of intangible resources.