Abstract
This paper is a proposal for dissertation research. The proposal was accepted Sept. 24, 2002. The research is focused on the question of how best to influence the accurate attribution of the meaning of silence for team members who are not co-located. The accurate attribution of silence can have a profound impact on the expenditure of effort. If we make more accurate attributions then we are more likely to make appropriate expenditures of effort, thereby affecting the performance of the group. The proposal calls for the creation of groupware that includes indications for three contextual elements – Physical Access, Information Access and Channel Richness. The groups will be taken through a task of sufficient complexity as to require cooperation and instances of silence for one member of each group will be introduced. The members will be surveyed post activity to determine their attribution of the periods of silence that occurred. It is expected that the groups that are given the cues will make more accurate attributions.
The Question
The phenomenon of interest is Attribution Accuracy, and specifically how the knowledge of contextual resources influences the accuracy of attribution of the behavior identified as silence.
Businesses today are incorporating more people into work teams than ever before. Due to the high costs of travel (and the risks associated with it) more teams are working in a distributed manner using a wider and wider variety of groupware tools. Groupware is a collection of utilities that may provide both structure and process to the teams that use them. The team members are often from varying departments, business units, parts of the country and very often, different countries (certainly different cultures).
While the industry that has sprung up to develop and support these tools is improving its understanding of the underlying social science of distributed team work, much still needs to be learned. One of the intriguing issues facing those working in distributed environments is how to deal with behaviors of remote partners when they are inconsistent with expectation. One such common and disturbing behavior is the act of falling silent (or failing to respond). Silence can have many meanings from the positive – “Good they are deep in thought about this thorny issue;” to the negative – “how can they not be paying attention to my important need for a response?” It can be attributed to the person – “that Sally, she’s trying to hurt our chances for success;” or to the situation – “perhaps Joe is having trouble with his connection again..” How the attribution is framed – situational or personal, positive or negative will impact the future interactions between the communicating parties, and therefore have a direct impact on team effectiveness. Cramton’s (2001) subjects found the interpretation of the meaning of silence to be “some problem or a serious problem” in all subject teams.
Who Cares
When we engage in communication behaviors we are constantly interpreting behavior to ensure effective communication. As a basis of the communication we make some assumptions about what is common ground between us. For example, if Bob and I meet on the street and he says “Hi,” he has made the base categorical assumption (probably based on the fact that we are both in the US and I appear to be of Anglo-Saxon heritage) that I speak English and his communication will be understood. In our efforts to accomplish some goal we exchange information--we interact. And through our interactions we test our assumptions, generally by determining whether the behavior of others is consistent with our assumptions about what they know. If their behavior is consistent, our assumption is held as accurate, if not, we engage in additional information exchange. So, if my response to Bob had been a quizzical look, he would have tested his assumption about my English speaking ability and may have made a different attempt.
The use of electronically mediated communication channels makes both information sharing and confirmation (assumption testing) more difficult. This is because of the reduction of non-verbal cues available in such communication channels. Without the cues, the interpretation of the communication is more difficult. We spend more speaking turns confirming the receipt of our message, and may require additional exchanges to clarify the message. This is true also for non-verbal communications (Clark and Schaefer 1989). This is believed to be true because most forms of mediated communication do not provide efficient back-channel feedback (the head nods, brief verbalizations, smiles and the like)(Yngve 1970). The non-verbal cues efficiently signal the state of mutual knowledge without taking over the speaking turn. Feedback lags associated with mediated communication and distributed collaboration have a significant impact on establishing and maintaining mutual knowledge (Krauss and Fussell 1990). Even a delay of 1.6 seconds is sufficient to disrupt real-time mediated communication, even though the back-channel response is eventually transmitted (Krauss and Fussell 1990).
We engage in the communication behaviors because we believe they will result in an increase in value, either to ourselves or to society as a whole. The extent to which our personal goals are aligned with the stated goals of the group in which we are operating is called goal congruence. The closer the alignment, the more likely we are to expend cognitive effort to continue the information exchange and assumption testing processes (Briggs, 1994). Feeding into Goal Congruence is Valence of Attribution (valence is defined in Webster’s as the degree of attraction or aversion that an individual feels toward a specific object or event and generally indicates the positive or negative connotation given to an attribution). Feeding valence is Certainty of Attribution – how strongly the attributor feels about the accuracy of the attribution. Moderating the relationship between valence and interaction is the Importance of Need. This is defined as how desired the requested behavior is to the attributor, the more the need the more likely the negative valence of the attribution. The end result of all this information exchange and assumption testing is mutual knowledge.
Mutual knowledge, also referred to as mutual understanding or common ground, is generally defined as knowledge that is shared and known to be shared. It is considered integral to the coordination of actions (Clark 1996, Clark and Carlson 1982). This is especially so for members of work groups that rely solely on mediated communications where their opportunities for direct experience as a mechanism for learning about contextual cues and building trust is extremely limited. For the most part they must rely on two other less precise mechanisms, interaction and categorical assumptions (Krauss and Fussell 1990).
Direct knowledge is created in the course of live experience with others. One can make relatively accurate assumptions about what others know and don’t know based on experiences shared with them and knowledge of their habits and environment gathered from firsthand observation. Unfortunately, it is unlikely that distributed team members will be able to gain direct knowledge. In lieu of direct knowledge, mutual knowledge can be established through interactions. Mediated interactions reduce the efficacy of information exchange, as these interactions tend to be less complete and more biased (Hightower and Sayeed 1995, Hollingshead 1996). Computer-mediated groups take more time and exchange less information overall than face-to-face groups. In addition, computer-mediated communication often proceeds at a slower rate than face-to-face verbal communications (Walther and Burgoon 1992). This is attributed to the additional time it takes to type rather than to speak. It is true that in many groupware tools, parallel communications are possible (with everyone contributing simultaneously), however, the members still must attempt to process each contribution to make sense of the whole, which takes time.
And finally, people make assumptions about other’s knowledge based on social categorizations they apply to them (Clark and Marshall 1981). Lacking other cues, people will “overattribute” based on what little social information they can glean (Lea and Spears 1991, 1992, 1993). They may also, lacking all information, make the assumption that the remote partner’s context is the same as their own (Cramton, 2001)
Mutual knowledge is fundamental to joint effort that is used to create value to both the individual and society as a whole. The establishment and maintenance of mutual knowledge is the result of cognitive effort expended in information exchange and testing of assumptions. The reason for expending that effort comes from goal congruence (how closely my own goals match those of the group). Goal congruence is derived from valence (the more positive the attribution, the more closely my goals and the groups goals appear to me). Inaccurate attributions of why people cooperate leads to mistakes in goal congruence and the inappropriate expense of effort. If the valence is too low, a member will not make effort that should be made to create mutual knowledge. If the valence is too high, effort will be made where it should not be and may result in actions not in the best interests of the member. Efficient effort is derived from accurate attributions.
The CIA model
According to Cramton, silence is a serious failure of information exchange, and she found that team members “often failed to guess which of the many features of their context and situation differed from the contexts and situations of remote partners. They did not communicate critical local information.” In addition, when the information was communicated it was often ignored. This is most likely due to the principle of information sampling (Stasser et al 1995), which holds that uniquely held information is less likely than commonly held information to be mentioned in group discussions and if mentioned, it is less likely than commonly held information to be salient to the group members. So just merely exchanging contextual condition information is insufficient for influencing the attribution of silence behaviors.
"The process of attribution is a person's attempt to explain the causes of an event." It is an intervening cognitive action and influences both attitudes and behaviors of the attributor toward the actor. Attributions are either personal (dispositional) or situational (contextual). Personal attributions are those where the cause is determined to be due to a personal trait or disposition. These could be intentional or unintentional. All of the items contained in the Knowledge of Trust Factors are personal traits. Personal attributions are also referred to as internal attributions when applied to the self. Attributions determined to be caused by the environment, or context, are named situational. These include the items contained in the Knowledge of Resources area. They are also referred to as external causes when applied to the self. Attributions affect our feelings about past events and expectations about future ones (Kelley and Michela 1980). The more accurately we make the attributions, the better we understand the actions of others and ourselves and our communication process is enhanced.
Contextual resources include all of the elements available to remote partners that make up their working environment. While it is recognized that silence can be used as a mechanism for manipulating communications, in this research we will limit silence to simply a result of failure to have a response recognized. In this then, we will limit the research to those factors dealing with contextual conditions that can lead to silence. These can be categorized into the following three areas:
- Physical access – the perception of the remote partner’s ability to gain physical and cognitive connectivity to the communication channel and would include a range from having ready, immediate, reliable access from anywhere at anytime to having access only at specific locations and times that may be unreliable.
- Information access – the perception that the remote partner has physical and intellectual capacity to access and use the information necessary to satisfy information needs of the requestor. This may include group history, task related data and channel related information.
- Channel richness – the standard media richness issue of channels that provide sufficient cues for the attribution of contextual condition. The theory is that richer channels provide more cues and therefore support more accurate attributions.
Propositions:
Assumption: The more we know about the subject at hand the better decisions we make regarding it.
Proposition 1: Accuracy of Attribution is a function of Knowledge of Resources
Proposition 2: Accuracy of Attribution is a function of Knowledge of Trust Factors
H1: The participants who receive salient measures of partner contextual resource condition will make more accurate attributions of silence caused by those conditions than participants who do not receive the measures.
The method
Simple Experiment. Build two groupware toolsets, one with functionality to expose and emphasize the contextual conditions of all members and one without this functionality. Using a script, take the teams through a distributed exercise (teams could be in separate rooms connected by technology). Silence can then be simulated by manipulating each of the three conditions (physical access, information access and channel richness) and survey about their attributions regarding the silent behavior of remote partners.
It is argued that technology treatment (providing salient information regarding the contextual resources) will not significantly affect Knowledge of Trust factors as they relate to attribution accuracy. In other words, attributions about trust will be the same regardless of the presence of the new functionality.
The tool will identify to all partners the quality of each of the three conditions. These values will be captured at change to allow for assessment of the values as they load on type and quality of attribution made (personal/situational, positive/negative). Give the group some task with sufficient cognitive load so that the task cannot be accomplished individually and set a reward sufficiently high so that there is something at stake.
Both groups need to be informed of the importance of contextual condition, but the control groups will not be given a tool to make it salient during group interactions.
Survey the groups as to their attributions for the periods of silence that occur to collect data covering the knowledge of contextual conditions, attribution type (personal/situational) and the valence of attribution (positive/negative). The survey can be set up to provide specific reasons for silence and both silent partners and their remote counterparts will make a determination as to the cause for the silence (rather than asking open ended questions and analyzing the written responses to categorize them).
I would also like to collect data as to importance of need and certainty of attribution as well as measures of goal congruence (for future study).
× Briggs, R. 1994 The focus theory of group
productivity and its application to development and testing of electronic group
support systems. Unpublished dissertation, University of Arizona.
× Clark, H. 1996. Using Language. Cambridge University Press, New York.
×¾, Carlson, T 1982. Speech acts and hearer’s beliefs. N. Smith, ed. Mutual Knowledge. Academic Press, New York 1-36.
× ¾, Marshall, C. 1981. Definite reference and mutual knowledge. A. Joshi, I Sag, B. Webber, eds. Elements of Discourse Understanding. Cambridge University Press, New York. 10-63.
× ¾, Schaefer, E. 1989. Contributing to Discourse. Cognitive Science 13 259-294
× Cramton, C. 2001. The Mutual Knowledge Problem and its Consequences for Dispersed Collaboration, Organization Science, 12:3, 346-371.
× Hightower, R., Sayeed, L. 1995. The impact of computer-mediated communication systems on biased group discussion. Computers in Organizational Behavior 11:1 33-44.
× Hollingshead, A. 1996. Information suppression and status persistence in group decision making: The effects of communication media. Human Communication Research. 23:2 193-219.
× Kelley, H., Michela, J. 1980. Attribution Theory and Research. Annual Review of Psychology 31 457-501.
× Lea, M. Spears, R. 1991. Computer-mediate communication, deindividuation and group decision-making. International Journal of Man-Machine studies. 34 293-301.
¾ 1992. Paralanguage and social perception in computer-mediate communication. Journal of Organizational Computing. 2 321-341.
¾ 1993. Love at first byte? Building personal relationships over computer networks. J. Wood, S. Duck eds. Under-studied Relationships. Sage, Thousand Oaks, CA. 197-233.
× Stasser, G., Stewart, D., Wittenbaum, G., 1995. Expert roles and information exchange during discussion: The importance of knowing who knows what. Journal of Experimental Social Psychology 31 244-265.
× Walther, J. Burgoon, J. 1992. Relational communication in computer-mediated interaction. Human Communication Research. 19:1 50-88.
× Yngve, V. 1970. On getting a word in edgewise. Papers from the Sixth Regional Meeting, Chicago Linguistics Circle. Chicago Linguistics Circle, Chicago, IL.