Communications of the ACM
Communications of the ACM
Sharing Knowledge
Knowledge management systems (KMS) have taken on a prominent role in both information systems research and practice. Such systems promise the potential for contributing to an organization's strategic advantage by unlocking knowledge heretofore housed only in the minds of certain organizational members. This potential has led to a rapid increase in the volume and types of systems for sharing knowledge. Recently, many issues and strategies relevant to KMS have been identified [7].
Sharing knowledge throughout an organization has intuitive appeal. If organizational members share valuable information freely with other members, the organization's responsiveness and effectiveness can be greatly augmented by preventing those members from having to repeatedly solve the same problems. In an environment of organizational sharing, a KMS can readily save time and money for both providers and users of knowledge. For example, more than a decade ago Microsoft created its "Knowledge Base" (KB), which was the first place employees checked when a software problem surfaced. Microsoft's decision to move the KB to a customer-focused extranet is ample testimony to its usefulness; now customers query the system without having to contact a fee-based support center, saving customers the fees and/or call, and saving Microsoft from having to build, staff, and manage a response center.
Because a KMS requires individuals to share their knowledge, success is not guaranteed; several factors can interact to hinder knowledge sharing [2]. One significant problem is that individuals often gain considerable power when they hold unique knowledge [5], especially when it is perceived to be of high value. They are sometimes hesitant to give up their competitive advantage by sharing it. Therefore, KMS adoption and use might require substantial encouragement by management (prompting), intense loyalty to the firm (group identification), or perhaps even simply personal values that support sharing (social value orientation).
Prior Research
Researchers have studied how sharing behavior relates to managerial prompting, group identification, and social value orientation [4]. Here, we describe the previous research on these three facilitating factors, the design of our experiment, our findings, and managerial implications.
Knowledge sharing. Knowledge in a KMS can be considered to be a public good, an asset available to all members of a community or organization regardless of whether the members contributed to its constitution [9]. Examples of public goods include public television, public parks, and clean air. Unfortunately, the fundamental issue with public goods is the "free rider" problem, whereby certain individuals will utilize a public good without contributing their fair share. Thorn and Connolly [11] asserted that information in a shared database will generally be undersupplied, as is the case with other public goods. Individuals are not likely, therefore, to share their valuable, personally held information through KMS.
Positive motivators have been identified that entice individuals to contribute even when personal costs of doing so are high. Factors that are likely to affect knowledge sharing include repeated direction (prompting) by managers, identification with a group engaged in a common task, and the person's fundamental predisposition to sharing. Each factor is described in more detail here.
Managerial prompting. To influence employees' attitudes, behaviors, and activities, organizations sometimes exert control [1] to achieve particular goals. Prompts are cues provided by managers to remind workers what is expected of them and can be informally described in highly descriptive yet colloquial language as "nagging." We do not attach value judgments about such prompts; they are simply an attempt by management to influence the behavior of individuals within an organization. As described here, prompting workers with meaningful reminders can have a strong desired effect on behavior [4]. Therefore, we expect that in situations where management encourages knowledge- sharing behavior by prompts or reminders, employees will contribute more knowledge to a KM system.
Group identification. Identification with a group will create allegiances that motivate individual behavior to benefit the group. Group identification has been examined in public goods research to determine why some individuals share more willingly than others in resource dilemmas [8].
Group identification accounts for the contributions from people to a common good [4]. Although usually considered an individual difference variable, it was found in De Cremer and Van Vugt [4] that, when certain managerial prompts reminded individuals about their group membership, "selfish" individuals increased their cooperation. Therefore, we expect to find that people who strongly identify with their team will share more knowledge for the benefit of the whole group. Individuals who do not feel they belong to the group will withhold valuable knowledge and contribute less to a KMS.
Social value orientation. Individuals can be described by three different profiles: as collectivists, competitors, or individualists [10]. Collectivists try to ensure they are sharing any returns in an equal manner with others. Competitors try to maximize their return by producing the maximum amount of separation between themselves and others. They are not as concerned with maximizing their own utility in the traditional sense of making a maximum amount of revenue. Rather, they obtain their utility by maximizing the distance between themselves and other competitors. So, competitors seek to increase the distance between themselves and others. In comparison, individualists try to maximize their own return regardless of what others are doing.
In many studies focusing on individuals' contributions, the three categories are often combined into two groups: individualists and competitors are labeled proself and collectivists are labeled prosocial. Proself individuals are most concerned that the outcome in any social dilemma is based on their ability to make a gain. Prosocial individuals are satisfied when group collective goals are met in social dilemmas. It is proposed here that a person's orientation (prosocial or proself) has an impact on the decision to share in a public goods dilemma. Therefore, we expect that prosocial individuals will contribute more knowledge to a KMS than proself individuals.
Other factors. Managerial prompting, group identification, and social value orientation are not the only factors that could be considered when studying elements that impact knowledge sharing. However, they are related through the work of De Cremer and Van Vugt [4] because they are common elements in most organizational environments. Therefore, they are pertinent factors in the study of organizational dynamics in modern business environments.
Interactions. It has been shown in other contexts that the three principal variables described here affect the amount of sharing by individuals when examined in isolation. It would be more interesting to investigate how these variables interact among themselves and how they together as a whole affect the amount of knowledge sharing. It is expected that proself individuals will make fewer contributions for the good of the group, but the institution of managerial prompts can lead them to contribute. On the other hand, it is plausible to expect that prosocial individuals do not find the prompts for sharing necessary, as they already understand and appreciate the need to maximize the good of the group. Therefore, we propose that managerial prompts will be more effective in increasing sharing with proself individuals than with prosocial persons.
Similarly, proself individuals are expected to be affected more by group identification cues than prosocial individuals. Prosocial individuals are likely to already be thinking about the group when faced with group identification cues, while proself individuals might be led to consider indirect rewards from being on a winning team with such cues. This effect was shown in another context in De Cremer and Van Vugt [4]. Therefore, group identification cues will be more important to proself individuals than prosocial ones when knowledge sharing is to be maximized.
For prosocial individuals, who already are expected to contribute, managerial prompting and group identification cues are expected to be largely unnecessary, even ineffective. It is therefore expected that the differences in knowledge sharing will be even greater when proself individuals will be given group identification and managerial reminders at the same time.
Method. A laboratory experiment was conducted utilizing computers at two large U.S. universities to test the previously described predictions. Seventy-six undergraduate business students participated in the experiment in exchange for extra credit toward their course grades. Students were seated in a computer lab and told the number of points to be awarded would be determined by their performance, composed of their own reward plus a share of their group's reward. The sessions were carried out in class under the researchers' supervision.
The study was set up as an investment game in which participants were to make a series of profit-maximizing decisions. The decisions focused on subjects' need to balance contributions of information for the common good against retaining that information for their own good, as we will describe in more detail here. Although individuals were told they formed groups of six members each, all decisions were made individually with no possible interaction with other participants.
The total profit of a player consisted of the player's own gains plus a bonus awarded if the player's team achieved a predetermined level of contributions (40% of the total). Whether to keep information for oneself for probable direct rewards or to contribute it to the team for probable indirect rewards was the dilemma at the heart of the investment decision. On one hand, players were enticed to keep information to themselves and increase their individual gains; on the other hand, they were drawn toward contributing the information to the team and securing the group bonus.
The relationship between profit and information sharing was implemented by a series of decision-making situations (rounds) offered to every player. In each round we displayed investment information and two probabilities the player was facing: a probability of individual gain should the player keep the information and a probability of a team gain should the player decide to share the information with another team member.
After two practice decisions, the players were presented with 16 consecutive information-sharing decisions, illustrated in the figure here. The rounds were identical except for the varying probabilities assigned to individual and team profit successes. The complete table of probabilities is shown in Table 1. Upon completion of the decision rounds, subjects were asked to provide demographic and other information.
We simulated Managerial Prompts by delivering three strategically timed pop-up messages while the task was under way for subjects in the "prompted" treatment. The first message appeared before round 1 and stated that the system would monitor the overall number of contributions to the team by its members, and should these contributions be insufficient to reach the bonus, each member would be notified. Regardless of the individual subject's level of sharing, the second message appeared after round 6 and warned the player that contributions to the team were too low and needed to be increased in order to obtain the bonus. The third message appeared after round 12 and displayed the same warning as the second message.
The manipulation of Group Identification was achieved by the indication to half of the subjects by the researcher that the game was being played in a competition with a named rival university. It was hoped that subjects who received such indication would identify with their group of classmates as an "us against them" situation. A similar manipulation was used successfully in Van Vugt and De Cremer [12].
Social Value Orientation reflected an individual preference and therefore was measured as part of the demographic variables. The Decomposed Games instrument adapted from Messick and McClintock [10] was used to classify the respondents into proself (individualistic and competitive) and prosocial (cooperative) categories.
Results
It appears that managerial prompts indeed resulted in more information sharing. Both the total sharing and three individual segments of sharing opportunities were examined, because the prompts were offered twice. Interestingly, the first segment did not stimulate differences, which would indicate that subjects did not pay much attention to the first managerial prompt. However, the other segments and the total did elicit significant differences. Overall, subjects who were provided with managerial prompts shared on average 7.34 times while subjects without such prompts shared on average 5.93 times (out of 16).
Group identification subjects (those exposed to a message about beating a rival school) did not seem to share any more than non-group identification subjects (6.84 vs. 6.32, respectively). While the means are in the opposite direction, they are not significantly different, and no conclusions can be drawn about group identification as a main effect.
Unexpectedly, 47 of the 76 subjects responded as proself individuals, while only 10 responded as prosocial individuals. The other 19 were indeterminate. Consistent with our expectations, the small number of prosocial subjects did contribute significantly more information (8.10 times) than the proself majority (6.26 times). When turning to interactions among the factors, we predicted that proself individuals would be affected more by managerial prompts than would prosocial individuals. Results were consistent with that expectation: the sharing scores for prosocial individuals receiving managerial cues were not significantly different (8.4 with prompts vs. 7.8 without prompts; p=0.362), and in contrast, the sharing scores were significantly different for proself individuals receiving managerial prompts (7.15 for managerial prompt subjects and 5.59 for subjects without prompts).
We asserted that proself individuals would be more sensitive to group identity cues than prosocial individuals. We were unable to test this assertion because of distribution of subjects in the sample. There were only two prosocial subjects in the no-group-identity cell.
Finally, a three-way interaction among the three experimental factors was tested. The small number of prosocial subjects in the non-group identification cell leads us to interpret any results from this test with caution. However, even with the sample difficulties, the three-way interaction was indeed significant. R2 was 0.301 and adjusted R2 was 0.210, indicating that the ANOVA explained a substantial amount of variance for the small sample; the means are shown in Table 2.
Conclusion
This experimental study explored the direct and interaction effects of managerial prompts, group identification, and social value orientation on knowledge sharing. The results indicate that knowledge sharing in organizations can be encouraged by management's reminders of the importance of the goal, as well as reminders about rivals. It is also important for organizations to hire personnel with prosocial traits if they want to depend on them to share. This builds on previous literature that identifies important organizational cultural variables as important elements in knowledge sharing [3].
Organizational culture might also interact with national culture in potentially interesting ways, and additional research might investigate if the phenomena are related. For instance, a future study might find that in societies that are more collectivist in orientation (such as several countries in Asia), different tactics could be used to foster prosocial attitudes than those in more individualistic societies (such as the U.S.) [6].
Although group identification in this sample did not seem to make a difference, the results should be taken only as a directive for closer study. Many organizations use tactics to foster such an image, such as posting mottos on a wall, having retreats and annual meetings with motivational speakers, or communicating regularly the importance of the organization's strong standing in the market. Such tactics to make employees desire to beat the competition could provide productivity benefits, while employees seek improved job security, larger bonuses, or even satisfaction that comes with success. Group identification, therefore, remains an intuitive and promising goal in spite of the nonsignificant findings in this study.
This research also has some limitations. The research method was experimental. The method was selected in order to adequately manipulate the variables and environment. However, this method does sometimes suffer from the ability to generalize results across a population when there are multiple other factors that can also influence behavior. Students' expectations of reward might have played a role in their behavior during the study, and incentives in a real situation need to be examined to see if they are consistent with the behavior in the study.
While managerial prompting might indeed enhance contributions in actual organizational situations as it did in this study, its effectiveness in an employment setting might need further study. This experiment was meant to discover if prompting had any promise, and future studies must be conducted to determine its efficacy over time. In a situation where a pop-up screen is shown multiple times per hour, such a tactic might be completely ignored or ineffective after several days. Personal prompting might be most useful in a variety of situations, but there could very well be systems that intermittently monitor performance and provide occasional advice through prompts that are used sparingly. Users would quickly learn the system is keeping track of their individual progress in a formal way and might adjust their behavior according to appropriate incentive systems.
Future research is needed to discover other important antecedents of knowledge sharing, both separately and in combination with those explored here. The full promise of KMS will follow a firmer understanding of how to foster the sharing of knowledge by organizational members. If individuals learn how to overcome natural tendencies to retain personal competitive advantage based on proprietary information, we might realize such large organizational gains that the individuals will be even better off than they would be if everyone jealously guarded their valuable knowledge.
References
1. Birnberg, J. Some reflections on the evolution of organizational control. Behavioral Research in Accounting 10 (1998), 2746.
2. Bock, G., Zmud, R., and Kim, Y. Behavioral intention formation in knowledge sharing: Examining the roles of extrinsic motivators, social psychological forces, and organizational climate. MIS Quarterly 29, 1 (2005), 87111.
3. Davenport, T., DeLong, D., and Beers, M. Successful knowledge management projects. Sloan Management Review (1998), 4357.
4. De Cremer, D. and Van Vugt, M. Social identification effects in social dilemmas: A transformation of motives. European Journal of Social Psychology 29 (1999), 871893.
5. Goodman, P. and Darr, E. Computer-aided systems and communities: Mechanisms for organizational learning in distributed environments. MIS Quarterly 22, 4 (1999), 417440.
6. Hofstede, G. Culture's Consequences: International Differences in Work-Related Values. Sage Publications, Beverly Hills, CA, 1980.
7. King, W., Marks, P., and McCoy, S. The most important issues in knowledge management. Commun. ACM 45, 9 (Sept. 2002), 9397.
8. Kramer, R.M. and Goldman, L. Helping the group or helping yourself? Social motives and group identity in resource dilemmas. In D.A. Schroder, Ed., Social Dilemmas: Perspectives on Individuals and Groups. Praeger, Westport, CT, 1995, 4967.
9. Marwell, G. and Oliver, P. The Critical Mass in Collective Action: A Micro-Social Theory. Cambridge University Press, New York, 1993.
10. Messick, D. and McClintock, C. Motivational basis of choice in experimental games. Journal of Experimental and Social Psychology 4 (1968), 125.
11. Thorn, B. and Connolly, T. Discretionary data bases: A theory and some experimental findings. Communication Research 14, 5 (1987), 512528.
12. Van Vugt, M. and De Cremer, D. Leadership in social dilemmas: The effects of group identification on collective actions to provide public goods. Journal of Personality and Social Psychology 76, 4 (1999), 587599.
©2008 ACM 0001-0782/08/0200 $5.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2008 ACM, Inc.
No entries found