Data Loading...
Anonymity In Group Support Systems Research: New ... Flipbook PDF
The authors would like to thank Anne Beaudry. This research was supported financially by FCAR (Quebec) and SSHRC (Canada
139 Views
111 Downloads
FLIP PDF 250.2KB
Anonymity In Group Support Systems Research: New Conceptualization And Measure Alain Pinsonneault
Nelson Heppel
École des Hautes Études Commerciales, IT Department 3000, chemin de laCôte-Sainte-Catherine, Montréal (Québec) H3T 2A7 [email protected]
Abstract By reducing inhibition, enhancing participation, and by increasing the number of ideas generated, anonymity is expected to improve communication and decision making and to enhance group performance. However, empirical evidence is inconclusive and provides a rather confusing understanding of the real effects of anonymity in group support systems (GSS). This paper addresses the state of empirical findings by examining the impacts of anonymity on groups through a new conceptualization of anonymity. In particular, this paper draws on the literature in social psychology and argues that anonymity has been too narrowly conceptualized as non-identification of participants, when in fact its most important dimension is subjective and perceptual. Measures of the new concept are presented and preliminary tests are conducted. This paper also argues that a better understanding of the effects of anonymity could be achieved by broadening the scope of research to include other situational factors such as attentional and accountability cues. A contingency approach for studying the effects of anonymity is presented.
1. Introduction Group work is becoming increasingly important in organizations. Managers spend more than half of their time in meetings [19], and most important decisions are made in groups [17]. Yet, group work is often ineffective: lack of participation, social blocking, self or socially-imposed censorship, hidden agendas, etc. By allowing anonymous communications, GSS are expected to reduce inhibition and evaluation apprehension, enhance participation, increase the number ideas generated, and improve the quality of decisions [35]. Yet, empirical evidence provides an unclear understanding of the effects of anonymity in computer-based interactions [36]. Anonymity was found to generate more critical comments [6, 22, 25,
44], to have no effects on inhibition [15, 18, 26] on group communications [40], and on group performance [15, 22, 44], and, surprisingly, to lead to more conservative decisions [18]. Some explanations have been advanced for this lack of convergence: effects of anonymity might depend on the importance of power and status differences in groups [35]; there might be lower evaluation apprehension and conformance pressures in some groups, which may reduce the impact of anonymity [35]; anonymity might interact with other variables, like leadership [15, 21], or evaluative tone [6]; and anonymity might have to be reconceptualized [27, 35]. The key commonality to these explanations is that the concept of anonymity seems much more complex than expected and that we do not clearly understand its essence. This paper attempts a theoretical explanation of the state of empirical findings by providing a new conceptualization and measure of anonymity and by analyzing its impacts in the broader context of disinhibition. First, it presents the empirical evidence about the effects of anonymity in computer-based communications. Then, it analyzes the empirical evidence by drawing on research in social psychology. Finally, it provides a new conceptualization and measures of anonymity and it presents a contingency approach for studying its effects. The results of preliminary tests of the measures are also discussed.
2. Anonymity in GSS research 2.1 Concept and expected effects of anonymity Anonymity in GSS research in particular, is defined as the inability of group members to identify the origin of messages they receive, and the destination of messages they send [15]. It is operationalized as a lack of identification, that is, no names are associated with messages sent and received by group members.
_________________________ The authors would like to thank Anne Beaudry. This research was supported financially by FCAR (Quebec) and SSHRC (Canada).
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
By divorcing statements from their authors, anonymity is expected to disinhibit behaviors, which has several implications for groups. First, because of the reduction in evaluation apprehension, more equal participation is expected and more comments are likely to be generated [6, 21]. Or, conversely, lack of identification will facilitate social loafing, and anonymous groups will generate less comments and there will be more unequal participation [18]. Second, lower conformance pressure and external social controls are expected to generate more critical and uninhibited comments [15, 18, 21, 22, 25, 40], especially for socially Study
N
Subjects
Method
sensitive tasks. This should improve the quality of ideas and of decisions, and group performance [6].
3. Empirical evidence Figure 1 presents the empirical studies on the effects of anonymity on groups. Contrary to the strong convergence in expected impacts of anonymity, except for Hiltz et al. [18], empirical evidence is rather inconclusive. Empirical evidence indicates that anonymity was found to have both no effect on inhibition and to increase criticalness [6, 15, 18, 21, 22].
Hiltz et al. (1989)
18 (5)
Managers
Field experiment
Independent Variables Anonymity (pen name)
Connolly et al. (1990)
24 (4)
Students
Laboratory experiment
Anonymity (no identification)
Dependant Variables Inhibition (=) Number of comments (=) Equality of participation (=) Consensus (=) Conservativeness (+) Number of comments (+) Quality of ideas (=) Criticalness (+) Number of comments (+)
Jessup et al. (1990)
20 (4)
Students
Laboratory experiment
Anonymity + critical tone Anonymity (no identification)
George et al. (1990)
30 (6)
Students
Laboratory experiment
Anonymity (no identification)
Satisfaction (+)
Laboratory experiment
Anonymity + leadership Anonymity (no identification)
Anonymity + face to face Anonymity + dispersed
Criticalness (+)
Jessup and Tansik (1991)
N:
18 (5)
20 (4)
Students
= 18 groups of 5 individuals Figure 1 Research on anonymity
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
Number of comments (+) Criticalness (+) Clarification efforts (+) Decision quality (=) Number of comments (=) Consensus (=) Time to reach a decision (=) Equality of participation (=) Satisfaction (=) Inhibition (=)
Clarification efforts (+) Number of comments (=) Criticalness (=) Satisfaction (=)
Satisfaction (+)
Empirical evidence also indicates that anonymity increases the number of comments generated [6, 22], but has no significant effect on the amount of participation [15, 18, 21]. Futhermore, anonymity was found to have no impacts on several variables where important effects were expected: equality of participation [15, 18]; consensus reaching and disagreement [15, 18]; and quality of ideas or decisions [6, 15]. Anonymous groups were not more satisfied, except when dispersed [21] or with an assigned leader [15]. Anonymous groups generated more clarification efforts than identified ones [21, 22]. Finally, surprisingly, anonymity was found to induce more conservative decision making within the context of a conservative corporate culture [18]. In sum, past research conducted indicates that anonymity has had limited impacts with settings using different technologies, different tasks, and subjects with different motivations Several explanations have been articulated for the state of empirical evidence. First, anonymity effects might depend on contextual variables like the importance of power and status differences in groups [35]. For example, it might interact with other variables, like leadership [15, 20], proximity [21], or evaluative tone [6]. Second, there might be lower evaluation apprehension and conformance pressures in some groups, which may reduce anonymity’s effect [35]. Third, anonymity might have been incorrectly conceptualized and may need to be redefined [27, 35, 43]. Along with Licker [27] and Nunamaker et al. [35], we favor this last line of thought. What seems to underly these explanations is the fact that anonymity is a much more complex concept than expected which we do not yet fully understand. Its impacts are varied and might depend on other characteristics of groups or other situational factors.
4. Anonimity and disinhibition in social psychology If anonymity plays a role in computer-based interactions, it has been assumed that it would break down social barriers and conformance pressures and lead to disinhibited behaviors (criticalness, amount and equality of participation, clarification efforts, flaming, etc.). But empirical evidence does not support this proposition.
4.1 Why disinhibition does not always occur ? Disinhibition is caused by two main subjective states (public and private self-awareness), each of which has
its own antecedent conditions (accountability and attentional cues respectively) [38]. Public self-awareness involves attention to oneself as a social object and concerns appearance and the impressions made in social situations [38]. It reflects the importance given to being evaluated and judged by others and to social evaluation in general [1, 13, 12,]. Lowering public self-awareness reduces concerns with social standards, conformity, and social evaluation. Not expecting retaliation and censure, individuals become disinhibited [11, 9, 48]. In these situations, people are aware of their actions, they simply do not expect to suffer any negative consequences for their actions, as disinhibited behaviors will go unpunished. Public self-awareness is determined by accountability cues, or the extent to which people think they will be held accountable for their actions. These are made of group size, similarity of group members or knowledge of others [9], the presence of recording material and the number of observers in the room [4, 9, 14], diffused responsibility [28, 37, 38, 46, 48], and anonymity [37, 38]. All these factors have been found to affect public self-awareness and disinhibition. But, anonymity was found to interact with other situational variables in unpredictable ways, making its effects quite varied and contradictory and its relationship with disinhibition is very complex. Anonymity was found to increase uninhibited behaviors [24, 45], but also to have no effects [11], or even to decrease uninhibited behaviors [47, 48, 9] argues that one possible explanation of contradictory effects is that anonymity manipulations in experiments might convey the message to participants that the actions they are about to perform are undesirable, and thus might heighten self-awareness. Also, anonymity was found to interact with other situational factors, such as group unity, in quite unpredictable ways. Third, anonymity will disinhibit behaviors only if threats of punishments is a major source of inhibition and if the condition is perceived by individuals as an effective method for protecting themselves from these threats. This might explain why anonymity has had limited effects on disinhibition in experiments with students working on fictitious cases. They were liberated from constraints that were not a major source of inhibition. Anonymity might increase or decrease inhibition when inhibition originates from internal standards (strong values for example), depending on whether anonymity heightens or lessens self-awareness [9]. Then, Hiltz et al.’s [18] finding that anonymity increased conservativeness in a conservative culture could be explained by the fact that individuals working in such a culture are more conservative by
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
nature and that anonymity heightened self-awareness and conservativeness. Fourth, the concept of anonymity is multidimensional. It can be actively sought or imposed and there are different methods for becoming anonymous (costume, non-identification, pen name in written communications). Fifth, anonymity is much more than lack of identification. It is a condition that frees individuals from social evaluation or scrutiny [46]. There is an important subjective dimension to it, and it is the perception of whether the offered condition adequately frees individuals from social evaluation that really makes people feel anonymous and disinhibit behavior. Disinhibition is also caused by private selfawareness, which refers to a focus on personal aspects of oneself, like perceptions, thoughts, and feelings [38]. Lessening private self-awareness leads to deindividuation, when a group of individuals become transformed into a united entity that seem to develop a collective mind. People become immersed in a group rather than focusing on self-regulation [9, 38] argues that when private-self awareness is decreased, people are not as aware of their internal processes, standards, and values and are less able to retrieve them to compare and adjust their behavior [9]. Group members thus regulate their behavior based on group norms and standards rather than on interval values, self regulation, and behavioral constraints. When deindividuated, behaviors depend on other emotional, motivational, and situational factors (like group norms) present at the time [9]. Thus, loss of private self-awareness might lead to disinhibition of behaviors, when, for example, group norms and standards favor its emergence, or it might also lead to more conservative, prosocial, and altruistic behaviors. Private self-awareness is determined by attentional cues (fostering group cohesiveness and arousal, and focusing attention away from oneself), but not by accountability cues (of which anonymity is part). Thus, according to Prentice-Dunn and Rogers [38] deindividuation is not likely to be obtained by anonymity. Also, it is rather unlikely that deindividuation occurs in adhoc groups, unless there is a period of time before the experiment during which group members could become a cohesive group [9]. Non-anonymous computer-based communications have been found to heighten private self-awareness and to lessen public self-awareness [30, 31, 32, 25] found that electronic mail attenuated social context cues, which substantially deregulated communications and favored the transmission of information that would not have been conveyed through another medium. Although the messages were clearly identified, the feeling of anonymity lead to more uninhibited behaviors, such as
flaming. Consequently, it may be argued that the marginal impact of anonymity in computer-based interactions is very limited because computer-based communications already lessen public self-awareness and diminish social context cues. In sum, the following observations can be made based on research in social psychology. First, there are numerous factors leading to disinhibition, and anonymity is only one of them. Hence, its expected effects on disinhibition are likely to be marginal. Also, social context cues are already attenuated and behaviors uninhibited by non-anonymous computer-based communications, leaving little to be gained by anonymity. Furthermore, anonymity only disinhibits behaviors when the major source of inhibition is fear of threats. Anonymity will have no significant effect when inhibition is caused by internal personal standards or values (related to private self-awareness). Second, anonymity has been found to interact with other situational factors, making its relation to disinhibition Depending on quite unpredictable and complex. different factors, anonymity could increase or decrease inhibition. This heightens the importance of measuring and controlling for such factors as private selfawareness, group unity, group size, presence of observers, and so forth. Finally, the very concept of anonymity is multidimensional, continuous, and of subjective nature. It is how individuals perceive the anonymous condition to protect themselves from threats that will lead to disinhibition of behaviors. Sorting through these explanations remains difficult because most empirical research in computer-based interactions has focused on anonymity alone as a single cause of disinhibition. In the next section, we develop a new conceptualization of anonymity and a contingency framework based on research in social psychology.
5. Anonymity and disinhibition in GSS revisited 5.1 The anonymity concept revisited Research in social psychology indicates that anonymity is the degree to which individuals feel liberated from social evaluation and from threats of punishment. Liberation from social evaluation is a multidimensional concept made of five components. First is the lack of identification associated with actions, comments, ideas, and so forth [9, 37, 38, 46]. This is an experimental condition and is not part of our perceptual measure of anonymity. People are either identified or non-identified. The second component is diffused
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
responsibility [28, 46, 48]. As discussed in the preceding section, empirical evidence indicates that when responsibility of behaviors is felt to be diffused to all members of the group, individuals feel more anonymous [11, 29, 48]. The third component of interest is proximity [9]. Proximity is especially important in the context of computer-based communications, because it can greatly influence the degree to which people feel being or not being observed by group members or by observers. For example, if people are seated too closed to one another, they might feel people can read what they write on their screen or look at their keyboard, and thus trace their comments. Although there are no names attached to comments, anonymity will be lessened in such situation. The fourth component is knowledge of other group members [9]. Knowledge of other group members directly affects the feeling of being observed and the suspicion that others could recognize comments by their style, content, wording, and so forth [46]. Group size is related to this variable, because in larger groups, people will feel it is less likely anyone can identify their comments [9]. The fifth component of anonymity, the confidence group members have in the system, comes from experience in conducting electronic meetings and not from the social psychology literature. People are suspicious of computer systems, and this is especially true when it comes to sharing critical information or discussing sensitive issues. Although great care is taken to ensure anonymity to participants, they often question whether or not the system is really anonymous, or if, for example, a number is not attached to comments so they can be traced back to individuals. Thus, although anonymity is provided, people might be suspicious and they will behave according to their perception.
5.2 Anonymnity and disinhibiton: a contingency approach
effects on disinhibition. Hence, a given level of anonymity will have different effects on disinhibition depending on how important social evaluation is as a source of inhibition for individuals (public-self awareness). A high level of anonymity is likely to have limited effects on disinhibition for people with a low level of public-self awareness (for whom social evaluation is not an important source of inhibition), whereas it is likely to produce high disinhibition for people with a high level of public-self awareness (for whom social evaluation is a very important source of inhibition). A low level of anonymity is likely to have no significant effects on disinhibition when public selfawareness is low and minimal effects when it is high because people will only feel partly liberated from their inhibiting factor. It is important to note that when public selfawareness is high, a correspondingly high level of anonymity will be required in order to liberate people from social evaluation. That is, when people fear social evaluation, disinhibition will occur only when they feel fully protected from it by anonymity. Conversely, a lower level of anonymity will be required to disinhibit behavior when fear of social evaluation is low because people will require a lower protection against something they do not fear as much (yet, as discussed above, the amount of potential disinhibition is limited because public self-awareness is not a major source of inhibition). In sum, the literature in social psychology leads us to expect that the effects of anonymity on disinhibition and the amount of disinhibition one can expect largely depend on public self-awareness. Although anonymity might interact with other situational factors, public selfawareness is likely to be an important variable to help better understand the complex relation between anonymity and disinhibition.
6. Measures and a preliminary test Empirical evidence on the effects of anonymity in GSS is quite contradictory. Reconceptualizing anonymity as a degree of liberation from social evaluation, threats of punishment and retaliation, and as a subjective, continuous, and multidimensional concept might help better understand this phenomenon and how it relates to disinhibition. However, research in social psychology also indicates that public self-awareness comes into play between anonymity and disinhibition. Anonymity can only significantly affect disinhibition when social evaluation is an important source of inhibition (Public self-awareness) [9]. Liberating people from social evaluation when social evaluation is not an important source of inhibition is likely to have very little
6.1 Measures Two concepts are important: anonymity and public self-awareness. Recall that anonymity is made of five components (identification, diffused responsibility, proximity, knowledge of other group members, and confidence in the system). Identification, being an experimental condition, is not included in our perceptual measure of anonymity. Some work has been done in social psychology on anonymity, but no measure has been proposed that could be applied in a GSS context. Figure 2 presents the measures we developed based on
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
the conceptualization of anonymity we proposed earlier and on the literature in social psychology [23, 29, 34, 39, 41]. Public self-awareness/public self-consciousness have been studied quite extensively in social psychology and scales have been developed and tested [2, 3, 14, 16]. The two concepts are similar with the exception that that public self-consciousness refers to a general awareness of the self as a social object, and public self-awareness is a state, that is, an awareness of the self as a social object as a result of transient situational variables [14]. Measures of public self-awareness in social psychology are simply measures of public self-consciousness, but in a given context. We are most interested in the specific context of computer-based interactions at a given time, and, consequently, public self-awareness is more relevant to us. However, most work has been done on the public self-consciousness in social psychology, which is then adapted to particular situations to Fenigstein et al. [14] were among the first to proposed a public self-consciousness scale. Their measure consisted of seven items: “One of the last things I do before I leave my house is look in the mirror” “I’m usually aware of my appearance”. “I’m concerned about my style of doing things” “I’m concerned about the way I present myself” “I’m self-conscious about the way I look” “I’m concerned about what other people think of me” “I usually worry about making a good impression” Each item was rated on a scale of 0 (Extremely uncharacteristic) to 4 (extremely characteristic). [14] tested their scale with 432 students and retested it with 150 college undergraduates. Principal components analysis yielded strong factor loadings (all above .45), and had a test-retest reliability of .84. Burnkrant and Page [3] tested Fenigstein et al. [14] scale in two studies using confirmatory factor analysis performed by LISREL V. In the first study, 360 women answered the questionnaire and the confirmatory factor analysis yielded a chi-square goodness to fit measure of 37.50, p=0.00, indicating a poor fit with the data. Two items, “One of the last things I do before I leave the house is look in the mirror” and “I’m usually aware of my appearance”, had very low reliabilities (less than .20) and were dropped from the scale. One hundred ninetyeight college students participated in the second study, in which the reduced scale produced a chi-square of 7.74, p=0.17 which is an acceptable fit. Gould [16] tested the five-item scale of Burnkrant and Page [3] with 169 adults in a large Northeastern metropolitan. Confirmatory factor analysis (LISREL VI) indicated that
all items loaded on the construct with a score of over .558. We preferred adapting the five item public selfconsciousness measure rather than] Prentice-Dunn and Roger’s [38] adaptation because the former was validated several times. Figure 2 presents the five item scale of public self-awareness.
6.2 Method A preliminary test of the different measures was conducted to assess construct validity and to identify any ambiguous items. In order to achieve this, we followed a method proposed by Davis [7] [8] and by Moore and Benbasat [33]. We asked judges to sort the various items into categories which represented the constructs of interest. Fifteen second-year graduate students in management information systems participated in the study (judges). Each item presented in Figure 2 was printed on one 4 X 6-inch index card randomly numbered from one to twenty-nine. Each judge was provided with a set of 29 cards randomly shuffled, clear instructions describing the sorting procedure (with precise examples), and six cards indicating the grouping categories and the definition of the variables (“Diffused responsibility”, “Proximity”, “Knowledge of other group members”, “Confidence in the system”, “Public selfawareness”, “Ambiguous/Does not fit”). The “Ambiguous/Does not fit” category was provided to allow judges to disregard items they estimated did not belong to any particular category or that were confused. This prevented judges from force fitting items into categories. A sorting trial, unrelated to the present study, was performed prior to the actual sorting to insure they understood the procedure. Judges were encouraged to ask as many questions they wanted to make sure they clearly understood the procedure. Judges were then asked to independently place items in categories which best reflected the underlying construct. An indicator of construct validity was convergence and divergence of items placed in the different categories. An item consistently classified in the same category by the different judges is considered to indicate convergent validity with the related construct and discriminant validity with the other constructs [33]. Three indicators were used to assess the reliability of the sorting procedure. The first indicator is which is the number of items that were placed by the panel of judges in the “appropriate” theoretical construct. Each item was designed to measure a particular theoretical construct; this indicator assesses the frequency with which items were assigned to the intended construct.
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
Concepts/Variables ANONYMITY
Definitions Degree of liberation from social evaluation Degree to which responsibility is believed to be equally distributed to all group members
Items N/A
Proximity
Perception of distance between people
I believed it was impossible to read on other group members’ screen (# 3) I believed my neighbours could not read my comments on my screen (# 16) I thought other group members could clearly see my keyboard when I was writing (# 11) I believed other group members knew when I sent a comment (# 24) I was able to read comments on my neighbour’s screen (# 2) I thought others read my comments on my screen (# 22)
Knowledge of other group members
Degree to which people know each other
I believed others could identify my comments (# 13) I believed that group members did not know each other well enough to identify the authors of comments (# 25) I believed I had a distinguishing characteristics that allowed other group members to identify my comments (# 21) I believed it was possible to identify the origin of the comments based the author’s personal charac (# 7) I recognized the author of most comments (# 15) I believed the group was large enough that it was impossible for any one to identify my comments (# 19) I believed the group was large enough that nobody could trace comments back to their authors (# 10)
Diffused responsibility
During the decision making process....1 All group members were equally accountable for the final decision (# 4) I believed it was impossible to make one group member responsible for the final decision (# 8) I believed it was impossible to ask me personally to justify the final decision (# 9) The decision making was everybody’s affair (# 12) I believed it was impossible to blame me personally for the final decision (# 26) I believed it was impossible to make me more responsible than others for the final decision (# 28)
1
The scales were designed for a decision making process. Some modifications are needed in the wording for idea generation processes. Some items were originally in French.
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
Confidence in the system
PUBLIC SELFAWARENESS
Degree to which people believed in the anonymity of the system
I believed the system could malfunction and identify my comments (# 29) I believed it was possible to identify my comment using the system (# 20) I believed that the system attached a code to comments so that their author could be identified if needed (# 18) I believed that no names were attached to comments by the system (# 5) I believed that my comments were not identified for other group members (# 1)
Attention to oneself as a social object
During the experiment.... I was concerned about my style of doing things (# 17) I was concerned about the way I presented myself (# 27) I was self-conscious about the way I looked (# 6) I usually was worried about making a good impression (# 14) I was concerned about what other people thought of me (# 23)
(# : item reference number) Figure 2 Concepts and initial measures The higher the percentage of “appropriate” placements, the higher the degree of inter-judge agreement. Also, Moore and Benbasat [33] suggest that a high degree of items being appropriately placed in a category indicates high construct validity. Second, the level of agreement was estimated using Cohen’s Kappa [5]. An assessment was made by comparing the categorization of all pairs of judges. Although there are no established guidelines for Kappa, scores greater than .65 have been considered acceptable in recent studies [33. The third indicator we used is the frequency of “correct” placements for each item. This is the percentage of time each item was placed in the correct category. This can be used to determine whether an item is ambiguous or not.
7. Results Table 1 presents the average results for the placement ratio for the 15 judges for the original scale (with the 29 items) and for the revised scale (with 26 items) (detailed placement ratios are presented in Appendix 1). Each of the 15 judges had to place 29 items, a total of 435 placements. These placements can be analyzed by a theoretical versus actual matrix (Appendix 1). Expected placement for each construct is
the number of judges multiplied by the number its number of items. For example, responsibility has six items. A perfect placement ratio would be 90 out of 90, or 100% (15 judges X 6 items). The diagonal in Appendix 1 (in bold) indicates the hit ratio, that is the number of correct placement for each construct and as a total. Table 1 indicates four of the five constructs of the original scales have acceptable placement ratios (all above 86%) and that the overall placement ratio is 88%. The low score of public-self awareness (73.33%) might indicate that some items are confused. The off-diagonal numbers in Appendix 1 indicates that there were five placements that were made in other constructs and that that were 15 placements rejected, a relatively high number. There was a good inter-judge agreement for all constructs, except for public self awareness. The rawagreement score was .79 and the Kappa score was .75 also indicating a good inter-judge agreement. To identify potentially confused items, the frequencies of item placements were calculated. Table 2 indicates that most items have very good placement ratios, except items 1, 17, 24 with scores of 53%, 40%, and 53% respectively.
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
Table 1. Average placement ratios
Theoretical constructs
Original Scales Total Placements 90 90 105 75 75 435
Diffused responsibility Proximity Knowledge of other members Confidence in the system Public self-Awareness TOTAL
Hits
Average (%) 95.56 88.89 92.38 86.67 73.33 88.05
86 80 97 65 55 383
Revised Scales Total Placements 90 75 105 60 60 390
Hit s 86 72 97 57 49 361
Average (%) 95.56 96.00 92.38 95.00 81.67 92.56
Table 2. Frequency of item placements Item number Diffused responsibility (A)
Proximity (B)
Knowledge of gr. members (C)
Confidence in the system (D)
Public self-awareness (E)
4 8 9 12 26 28 2 3 11 16 22 24 7 10 13 15 19 21 25 1 5 18 20 29 6 14 17 23 27
Categories A 13 15 15 15 15 13
B
D
E 1
1 15 14 14 15 14 8
2
C
1
Total Rejects 1
1
1 1 1 1 15 13 14 14 12 14 15 1 1
1 1
1 1
1 1
1 1
1 1 1
1
3 2
8 12 15 15 15
1 2
5
11 14 6 14 10
3 6 5
Overall average
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15
Average (%) 86.67 100.00 100.00 100.00 100.00 86.67 100.00 93.33 93.33 100.00 93.33 53.33 100.00 86.67 93.33 93.33 80.00 93.33 100.00 53.33 80.00 100.00 100.00 100.00 73.33 93.33 40.00 93.33 66.67 88.05
Item 1 ”I believed that my commens were not identified for other group members knew when I sent a comment” were rejected by numerous judges. These items were removed from the scales because they were too ambiguous. Following these changes, the overall placement ratio increased to 92.56% (Table 1) and the placement ratios for proximity, confidence in the system, and public selfawareness were improved substantially to 96%, 95%, and 81.67% respectively (Appendix 1 presents the detailed ratios). Also, the inter-judge agreement score was raised from .79 to .87 and the Kappa score from .75 to .84.
8. Discussion 8.1 Scholarly implications From being an objective (identification or not), unidimensional, and dichotomous construct, anonymity was reconceptualized as a subjective, multidimensional, and continuous construct: the degree of liberation from social evaluation. This new conceptualization has several implications for research. First, because anonymity is subjective, its measurement has to be performed as soon as possible after the anonymous condition. For example, in the context of a GSS experiment with multiple tasks, this entails measuring anonymity after every anonymous task. Second, numerous factors affect the degree of anonymity (e.g. proximity, screen dispositions, etc.), and consequently greater care will be required to assure sufficient degree of anonymity. Third, anonymity has been found to interact with other situational variables in social psychology such as group unity and self-consciousness that will have to be controlled and measured. Fourth, any study ofself awareness. The contingency framework proposed in this article might help better approach these phenomena and better understand their complex relationships. This heightens the importance of being careful when conducting experiments with students because public self-awareness might not be important for them in that context, and, consequently, anonymity will not have important effects. Furthermore, great care must be taken when generalizing results from adhoc groups to established ones. Adhoc groups are already partly anonymous because their members do not know each other well. Also, deindividuation is likely to be minimal in adhoc groups and important effects in established groups, which most probably affects disinhibition.
8.2 Future research The new scales need to be further tested and refined. The preliminary tests indicate that construct validity is acceptable, but more test must be performed to assess reliability and validity (internal and external). Heterogeneous samples should be sought to ensure generability. Also, the overall anonymity measure is relatively long and a shorter version of the scale would facilitate its usage in experiments. The framework also needs to be tested empirically and, maybe, further refined. Laboratory experiments should be conducted to determine the impact of anonymity on group processes and outcomes as it is reconceptualized in this article and to determine its interaction with public self-awareness. Also, the effects of anonymity in adhoc and established groups should be studied. The bottom line implication of this article is that it forces us to rethink the way we study GSS’s and how they affect group work. Anonymity is a subjective and very complex construct that interacts with other situational variables in quite unpredictable ways. More attention ought to be given to these situational variables, one of which is public self-awareness, to better understand anonymity.
9. References [1] Abrams, D. and R. Brown. (1989). Self-Consciousness and Social Identity: Self-Regulation as a Group Member, Social Psychology Quarterly, (52:4), 311-318. [2] Bernstein, I.H., G. Teng, and C.P. Garbine. (1986). A Confirmatory Factoring of the Self-Consciousness Scale, Multivariate Behavioral Research, (21), 459-475. 3] Burnkrant, R.E. and T.J. Page, Jr. (1984). A Modification of the Fenigstein, Scheier, and Buss Self-Consciousness Scales, Journal of Personality Assessment, (48:6), 629637. [4] Buss, A.H. (1980). Self-Consciousness and Social Anxiety, W.H. Freedman and Company, San Francisco, CA. [5] Cohen, J.A. (1960). A Coefficient of Agreement fro Nominal Scales, Educational and Psychological Measurement, (20), 37-46. [6] Connolly, T., L.M. Jessup, and J.S. Valacich. (1990). Effects of Anonymity and Evaluative Tone on Idea Generation in Computer-Mediated Groups, Management Science, (36:6), 689-703. [7] Davis, F.D. (1989). Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology, Management Information Systems Quarterly, (13:3), 319-340.
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
[8] ----------. (1986). A Technology Acceptance Model for Empirically Testing New End User Information Systems: Theory and Results, Unpublished Doctoral Dissertation, Massachusetts Institute of Technology, Boston, MA. [9] Diener, E. (1980). Deindividuation: The Absence of SelfAwareness and Self-Regulation in Group Members, in P.Paulus (Ed.), Psychology of Group Influence, Lawrence Erlbaum and Associates, Inc., Hillsdale, NJ:, 209-244. [10] Diener, E., R. Lusk, D. DeFour, and R. Flax. (1980). Deindividuation: Effects of Group Size, Density, Number of Observers, and Group Members Similarity in Self-Consciousness and Disinhibited Behavior, Journal of Personality and Social Psychology, (39:3), 449-459. [11] ----------. (1976). Effects of Prior Destructive Behavior, Anonymity, and Group Presence on Deindividuation and Aggression, Journal of personality and Social Psychology, (33), 497-507. [12] Fenigstein, A., and P.A. Vanable. (1992). Paranoia and Self-Consciousness, Journal of personality and Social Psychology, (62:1), 129-138. [13] Fenigstein, A. (1979). Self-Consciousness and Overperception of Self as a Target, Journal of personality and Social Psychology, (37:1), 75-86. [14] Fenigstein, A., M.F. Scheier, and A.H. Buss. (1975). Public and Private Self-Consciousness: Assessment and Theory, Journal of Consulting and Clinical Psychology, (43:4), 522-527. [15] George, J.F., G.K. Easton, J.F. Nunamaker, Jr, and G.B. Northcraft. (1990). A Study of Collaborative Group Work With and Without Computer-Based Support, Information Systems Research, (1:4), 394-415. [16] Gould, S.J. (1986). The Self-Consciousness Scale: A Confirmatory Analysis, Psychological Reports, (59), 809-810. [17] Hackman, J.R. and R.E. Kaplan. (1974). Interventions into Group Process: An Approach to Improving the Effectiveness of Groups, Decision Sciences, (5), 459480. [18] Hiltz, S.R, M. Turoff, and K. Johnson. (1989). Experiments in Group Decision Making, 3: Disinhibition, Deindividuation, and Group process in pen Name and Real Name Computer Conferences, Decision Support Systems, (5:2), 217-232. [19] Hymowitz, C. (1988). A Survival Guide to the Office Meeting, The Wall Street Journal, June 21, 35. [20] Jessup, L.M., and J.F. George. (1996). Theoretical and Methodological Issues in Group Support Systems Research: Learning from Groups gove awry, Working paper, Indiana university. [21] Jessup, L.M., and D.A. Tansik. (1991). Decision Making in an Automated Environment: The Effects of Anonymity and Proximity with a Group Decision Support System, Decision Science, (22:2), 266-279.
[22] Jessup, L.M., T. Connolly, and J. Galegher. (1990). The Effects of Anonymity on GDSS Group Process With and Idea-Generating Task, Management Information Systems Quarterly, (14:3), 313-321. [23] Johnson, D.J., and L.L. Downing. (1979). Deindividuation and Valence of Cues: Effects on Prosocial and Antisocial Behavior, Journal of Personality and Social Psychology, (37:9), 1532-1538. [24] Jorgenson, D.O., and F.O. Dukes. (1976). Deindividuation as a Function of Density and Group Membership, Journal of personality and Social Psychology, (34), 2429. [25] Kiesler, S., J. Siegel, and T.W. McGuire. (1984). Social Psychological Aspects of Computer-Mediated Communication, American Psychologist, (39:10), 1123-1134. [26] Lea, M., and R. Spears. (1991). Computer-Mediated Communication, Deindividuation and Group DecisionMaking, International Journal of Man-Machine Studies, (34), 283-301. [27] Licker, P.S. (1992). You Said What? Understanding the Role and Function of Anonymity in GDSS, Proceedings of the Conference of the Administrative Science Association of Canada, 98-107. [28] Mann, L., J.W. Newton, and J.M. Innes. (1982). A Test Between Deindividuation and Emergent Norm Theories of Crowd Aggression, Journal of Personality and Social Psychology, (42:2), 260-272. [29] Mathes, A.W., and T.A. Guest. (1976). Anonymity and Group Antisocial Behaviour, The Journal of Social Psychology, (100), 257-262. [30] Matheson, K., and M.P. Zanna. (1990). ComputerMediated Communications: The Focus in on Me, Social Science Computer Review, (8:1), 1-12. [31] ----------. (1989). Persuasion as a Function of SelfAwareness in Computer-Mediated Communications, Social Behavior, (4), 99-111. [32] ----------. (1988). The Impact of Computer-Mediated Communication on Self-Awareness, Computer in Human Behavior, (4), 221-233. [33] Moore, G.C. and I. Benbasat. (1991). Development of an Instrument to Measure the Perceptions of Adopting an Information Technology Innovation, Information Systems Research, (2:3), 192-222. [34] Nadler, A., M. Goldbert, and Y. Jaffe. (1982). Effect of Self-Differentiation and Anonymity in Group on Deindividuation, Journal of Personality and Social Psychology, (42:6), 1127-1136. [35] Nunamaker, J.F., A.R. Dennis, J.S. Valacich, D.R. Vogel, and J.F. George. (1991). Electronic Meeting Systems to Support Group Work, Communications of The ACM, (34:7), 40-61. [36] Pinsonneault, A., and K.L. Kraemer. (1990). The Effects of Electronic Meetings on Group Processes and Outcomes: An Assessment of the Empirical Research, European Journal of Operational Research, (46), 143161.
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
[37]
Prentice-Dunn, S., and R.W. Rogers. (1989). Deindividuation and the Self-Regulation of Behavior, in P. Paulus (Ed.), Psychology of Group Influence, Lawrence Erlbaum Associates, Inc., Hillsdale, NJ:, 87109. [38] ----------. (1982). Effects of Public and Private SelfAwareness on Deindividuation and Aggression, Journal of Personality and Social Psychology, (43:3), 503-513. [39] Propst, L.R. (1979). Effects of Personality and Loss o Anonymity on Aggression: A Reevaluation of Deindividuation, Journal of Personality, (47:3), 531544. [40] Siegel, J., Dubrowsky, V., Kiesler, S., and T.W. McGuire. (1986). Group process in Computer-Mediated Communication, Organizational Behavior and Human Decision Processes, (37), 157-187. [41] Solomon, H., L.Z. Solomon, J. Maiorca. (1982). The Effects of Bystander’s Anonymity, Situational Ambiguity, and Victim’s Status on Helping, The Journal of Social Psychology, (117), 285-294. [42] Sproull, L., and S. Kiesler. (1986). Reducing Social Context Cues: Electronic Mail in Organizational Communication, Management Science, (32:11), 14921512. [43] Valacich, J.S., L.M. Jessup, Dennis, A., and J.F. Nunamaker. (1992). A Conceptual Framework of
Anonymity in Group Support Systems. Group Decision and Negotiation, (1), 219-241. [43] Valacich, J.S., L.M. Jessup, Dennis, A., and J.F. Nunamaker. (1992). A Conceptual Framework of Anonymity in Group Support Systems. Group Decision and Negotiation, (1), 219-241. [44] Valacich, J.S., A.R. Dennis, and J.F. Nunamaker. (1991). Anonymity and Group Size Effects on Computer Mediated Idea Generation. Proceedings of Academy of Management Meeting. [45] Watson, R.I. (1973). Investigation into Deindividuation Using a Cross-Cultural Survey Technique, Journal of Personality and Social Psychology, (25), 342-345. [46] Williams, R. (1988). Reflections on Anonymity, Perceptual and Motor Skills, (67), 763-766. [47] Zabrack, M., and N. Miller. (1972). Group Aggression: The Effects of Friendship Ties and Anonymity, Proceedings of the 80th Annual Convention of the American Psychological Association, (7), 211-212. [48] Zimbardo, P. (1970). The Human Choice: Individuation, Reason, and Order Versus Deindividuation, Impulse, and Chaos, in W.J. Arnold and D. Levine (Eds.), Nebraska Symposium on Motivation, University of Nebraska Press, Lincoln, NE:, 237-307.
Appendix 1. Placement ratios Original scales Categories
Diffused responsibility (A) Proximity (B) Knowledge of gr. members (C) Confidence in the system (D) Public self-awareness (E) Overall
(A) 86 0 0 2 1
(B) 0 80 0 1 0
Total (C) 0 1 97 2 2
(D) 1 3 3 65 3
(E) 1 1 3 0 55
Rejects 2 5 2 5 15
90 90 105 75 75 435
Hits
Average (%)
86 80 97 65 55 383
95.56 88.89 92.38 86.67 73.33 88.05
Hits
Average (%)
86 72 97 57 49 361
95.56 96.00 92.38 95.00 81.67 92.56
Revised Scales Categories
Diffused responsibility (A) Proximity (B) Knowledge of gr. members (C) Confidence in the system (D) Public self-awareness (E) Overall
(A) 86 0 0 0 1
(B) 0 72 0 0 0
Total (C) 0 0 97 1 0
(D) 1 3 3 57 2
(E) 1 0 3 0 49
Rejects 2 0 2 2 8
Proceedings of The Thirtieth Annual Hawwaii International Conference on System Sciences ISBN 0-8186-7862-3/97 $17.00 © 1997 IEEE
1060-3425/97 $10.00 (c) 1997 IEEE
90 75 105 60 60 390