SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Method
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgement
  10. References
  11. Appendix

The need to model users' dynamic behaviour in Computer Supported Cooperative Work (CSCW) systems arises in many contexts. This study developed a probabilistic model of the usage of an awareness-maintaining mechanism in a collaborative hypertext database system. Longitudinal time series data of user-database interaction were studied. The study found that the recurring patterns in the occurrences of the awareness-seeking event were related to several contextual aspects of the CSCW system studied. The context-behaviour relationship is captured by a Poisson regression model. The analytical method can be applied to the study of situated actions in other CSCW systems.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Method
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgement
  10. References
  11. Appendix

The majority of Computer Supported Cooperative Work (CSCW) studies in recent years have been concerned with engineering aspects of building CSCW applications with advanced techniques. However, this technique-driven trend does not necessarily give us insights into the new phenomenon of collaboration with computer supported infrastructures (McCarthy, 1994). The design of a successful application calls for user-centred and multi-disciplinary design methodologies.

Improving users' awareness of social and historical issues in collaborative work has been of extensive research and engineering interest. It is one of the key issues to the success of CSCW systems (e.g. Dourish & Bellotti, 1992). The dynamic nature and the complexity of the coupling between situations and embedded actions have been emphasized in the research of CSCW systems (Suchman, 1987). The term situated actions emphasizes the interrelationship between an action and its context of performance.

A user-centred vision has resulted from the predominant ethnomethodology in the field of CSCW. Although ethnographical studies are widely accepted in the CSCW community, its results, usually, are not readily accessible to the designers and implementers of CSCW systems, or groupware applications. In particular, there has been a lack of quantitative and longitudinal studies on the use of CSCW systems.

This study analyzes the relationship between awareness-seeking behaviours and several contextual aspects of a collaborative hypertext database. People need awareness-seeking facilities to detect, monitor, and reduce various discrepancies that arise in using collaborative hypertext databases. The framework of the study is built on discrepancy reduction theories found in social psychology, such as the theory of cognitive dissonance (Festinger, 1957). The discrepancy reduction models address motivating factors behind people's behaviour, that when people recognise a discrepancy between an existing cognitive model and reality, people tend to act so as to reduce or resolve the discrepancies. Empirical work on cognitive dissonance particularly concerns with how people would change their cognitive models.

This study focuses on recurring behavioural patterns which might bear insights into the dynamics of collaborative writing with a shared hypertext database. Behavioural patterns of using a specific function INFO are analyzed as situated actions. The analysis first identifies the probability distribution of the occurrences of the INFO event and then develops a probabilistic model to capture the relationship between the occurrences of the event and several contextual aspects of the path of users interacting with a shared database. An improved understanding of the use of the database system in collaborative writing is expected in forms of the partial correlation relationships between the use of INFO and a number of explanatory variables which have influential impacts on the usage of INFO.

Background

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Method
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgement
  10. References
  11. Appendix

A recurring theme of research in the interaction between social structure and technology is the division of task-oriented and socioemotional aspects of collaborative work (Spears & Lea, 1992). This theme is found in the work of many researchers in computer-mediated communication. This division can be traced back to Bales' influential work in interaction process analysis (Bales, 1951).

A similar contrast can be found in recent studies of CSCW. A conceptualised process of collaborative working with computers consists of a spectrum of states or modes. At one end of the spectrum, users work with computers independently. At the other end of the spectrum, users require the capacity of face-to-face communication. Various efforts have been made to develop computer applications that will allow users to work in several different modes.

Working with a collaborative information system may involve the following three modalities:

  • • independent mode,

  • • loosely-coupled mode, and

  • • tightly-coupled mode.

This model of collaborative work has been incorporated into several CSCW systems such as gIBIS (Conklin & Begeman, 1988) and SEPIA (Haake & Wilson, 1992). An essential function of a collaborative information system is to facilitate a smooth transition between an independent mode and a collaborative mode. Yakemovic and Conklin (1990) studied the use of a simple Issue-Based Information System (IBIS) over an extended period of time to understand the processes and the problems involved in capturing the design rationale behind large and complex computer systems. However, there is still a lack of empirical foundations as to how the existing collaborative information systems have been used in reality and how the tools and users change each other. As McCarthy (1994) recently suggested, the existing models of collaborative work are still to be verified by real people in real work settings. Systematic, longitudinal analyses are needed to evaluate the systems built on these models.

The theory of structuration (Giddens, 1984) focuses on the interdependency of human action and social structure. Structures are composed of rules and resources. The rules and resources provide contextual constraints that individuals draw upon when acting and interacting. The interplay of structure and action produces and reproduces social systems. This process is called structuration. However, one shortcoming of such a duality model as Giddens' is that the model does not easily lend itself to empirical examination (Fulk, Schmitz, & Schwarz, 1992). Structures and actions in Giddens' model cannot be analytically separated. Archer (1988) suggests replacing the restricted duality with an analytical dualism. The relationship between structure and action reveals itself in temporal cycles, which allow each component of the system to play its part autonomously of the other. Consequently, empirical analysis of the influences of context on action and action on context is possible.

Diaper and Addison, (1991) identify several theoretical, practical, and resource limitations related to the psychological behaviourism in user modelling. They propose that the logical behaviourism approach is more realistic so as to use observable behaviours to stand for users' mental states. The emphasis of the logical behaviourism in user modelling is on the role of the observations and descriptions of the actions performed by users in building a mental model of users.

Figure 2.1 shows interrelationships among the basic components concerned in the theoretical framework of this study. The sequence of actions {act(t)} and a collaborative hypertext database are evolving interactively.

image

Figure 2.1. Interaction process with a collaborative hypertext database.

Download figure to PowerPoint

Markova, (1987) gives a comprehensive account of human awareness issues in sociology. In a broad sense, human awareness can be defined as follows:

  • 1
     people's ability to recognise their own existence and experience, and the existence and experience of others;
  • 2
     people's knowledge of their own agency and of that of others; and people's ability to monitor events in their own lives, and to make decisions about their own future on the basis of that knowledge;
  • 3
     people's ability to communicate their awareness of themselves and others to other human beings.

It is clear from this definition that human awareness involves both awareness of the self and awareness of other people. Human awareness is a social process. Shared electronic workspaces need to provide users necessary mechanisms to facilitate such a process.

Method

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Method
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgement
  10. References
  11. Appendix

Interacting with a collaborative hypertext database is a dynamic process. Some recurring behavioural patterns of user-computer interaction can be expected to be captured with statistical analysis. The emphasis of the study is on understanding these recurring patterns and how they are affected by changes in the contexts. Quantitative approaches are employed in contrast to the ethnomethodological approaches that have proven useful in recent CSCW studies. Quantitative approaches appear to have the following advantages to the field of CSCW: (1) easier to repeat in different settings so as to transform empirical findings easily from one field to another; (2) more readily accessible to designers and engineers of CSCW systems; and (3) more cost-effective to fit into the iterative development cycles of a large and complex CSCS system.

The MUCH System

This study is based on the use of a multi-user hypertext database system, Multiple Using Collaborative Hypertext (MUCH). The design rationale of the MUCH system can be found in (Rada, 1989). The architecture of the MUCH system consists of a semantic network, a dynamic traversal engine, and a hierarchical interface of the semantic network generated by the traversal program. In terms of the Dexter Hypertext Reference Model (Halasz and Schwartz, 1994), the semantic network in the MUCH system corresponds to the organisation of the storage layer and the hierarchical interface in the presentation layer helps users to navigate through the underlying semantic network.

The traversal program uses a combined depth-first and breath-first search algorithm. This combined traversal first collected all the nodes from links emanating from a node and then chose the next node by a depth-first principle. This approach has the advantage of giving the user an overview of what is to come. A detailed description of the algorithm is given in (Rada, 1989). The system allows users to alter the default traversal path by configuring a number of optional parameters.

The MUCH system was designed to support asynchronous collaborative writing. The system has a client-server design architecture. The server of the MUCH system, called the monitor, handles transactions concerning concurrency control and various collaboration facilities. A theoretical model underlying the MUCH system has been developed for collaborative writing and a number of empirical studies have been conducted regarding effects of an evolving collaborative hypertext on users' behavioural patterns (Chen & Rada, 1995) (Chen, Rada, & Zeb, 1994).

According to adaptive structuration theory, designers incorporate some of existing social structures into the technology. Some of these structures are transformed by the technology and are instantiated in social life. There are structures in technology and structures in action (DeSanctis & Poole, 1994). Social structures introduced by the MUCH system are built on the following actions, which shape the way that users interact with a collaborative hypertext database:

  • • Unfold to expand the hierarchical view in a breadth-first manner (i.e., use Unfold function to retrieve and display all the child nodes of the focal node);

  • • Read to load and display the content of a node in the hierarchical view;

  • • Lock to lock the content of the focal node from concurrent editing, but other users still can use Read on the node;

  • • Info to display various information stored in an embedded object of each node;

  • • Update to save changes since last Update or since the latest lock and to release the updated content to concurrent users.

Each node in the MUCH system contains a chunk of multimedia information. The system persistently stores evolutionary information of each node and attached links. The maintenance information is accessible via a function called INFO, which is available from each node in the semantic network. A user can always check the history of the currently displayed node for information such as who modified the node recently and how many times the node has been visited by users other than the original node creator. Figure 3.1 shows the dialogue box of the INFO function. The number 2 in the field of Selection Credits indicates that the node has been visited twice by users other than the creator, who has the username much and the content of this node has not been changed since July 14, 1994.

Patterns of using the INFO function and associated contexts are of particular interest in this study. Calling the INFO function on a node indicates an interest of the user in mutual intelligibility issues in collaborative work. Recurring behavioural patterns with the INFO function are expected to reveal some insights into the nature of using this function in asynchronous collaborative writing and help the designers of the system to capture the most salient dynamics in the context-behaviour relationship.

Users and Tasks

Users of the MUCH system include a research group with more than 15 members and a class of 30 graduate students for a Master of Science degree course in Information Systems. Five databases in the MUCH system were used collaboratively throughout the 3-month observation period.

Four of these databases were used by researchers in the group for organisational management, project management, book writing, and archiving working papers of the research group. The database for the class contained a textbook based on material selected from one of the book writing databases. This paper will mainly present the work concerning the use of the class database. The class database was regularly used by the majority of student users. It is expected this intensiveness in regularity could highlight some issues arising with the collaborative use of a shared workspace.

Computer logged transactions from the class correspond to two independent phases. In the first phase (three weeks), the students used an electronic textbook in the MUCH database in the classroom where all the users were present. In the second phase (three weeks), the students formed eight 3-person groups to write a group essay into the MUCH system on one of the recommended topics on groupware. Students were allowed to use external communication tools along with the MUCH system. Each student was also requested to produce an individual report of the experience gained in the group writing phase.

Data Collection

The data were collected from two sources: (1) actions performed by users in the presentation layer and (2) a structural representation of the underlying semantic network in the storage layer. Each action in the presentation layer is recorded in terms of when it takes place, who activates the action, and where the user is at the time of the activation in the hypertext information space. The structural representation of the semantic network were used to match an INFO action with the node where the action is activated. The INFO action allows a user to view the information stored in the INFO object embedded in the current visited node.

The process of interacting with the shared database can be characterised by a sequence of events. Each event takes place in association with a particular node. Each node has a set of organisational or evolutionary attributes, such as its hierarchical position in the presentation layer and a list of users who recently modified the node. Events other than INFO were filtered out from the sequence. The resultant sub-sequence was aggregated on a chosen attribute in each time interval. For instance, let n.attr denote the value of the attribute attr of the node n and the sequence of nodes {ni1, ni2, … nik} correspond to the events occurred in the ith interval in the sub-sequence, the following equation defines a time series of an aggregated variable based on the original attribute level:

si(level) =j=1,2,…,k nij.level.

A number of new time series are derived by aggregating observed values over fixed time intervals. These time series are collectively called an awareness-seeking path in the underlying database. These time series are further standardised over the whole process. For instance,

Zleveli= si(level) / s.d.

where s.d is the standard deviation of the time series {si(level)}. The resultant time series Zlevel = {Zleveli} indicates the average position of those nodes in the hierarchical view on which INFO actions have been frequently used. Several time series in this study are similarly generated (see Table 3.1). Note that some of these time series are not considered further as advised by the results of our preliminary analysis.

Table 3.1.  A list of time series obtained
VariableDefinitionNote
  1. *S - Standardised; A - Aggregated.

Houra particular time band imposed on the sampling windowS/A*
Phasea step function of reading and writing activitiesS/A
Zcreditaccumulated number of visits by users other than authors themselvesS/A
Zlevellevels of the nodes relating to INFO eventsS/A
Zlinktotal number of out-going links from the nodes involvedS/A
Zlinkthe size of nodes involved (link count)S/A
Zsizethe group size of current usersS/A
Zwordthe size of nodes involved (word count)S/A

Data Analysis

The occurrences of the INFO event were measured in non-negative integer numbers. Discrete data such as this event count cannot be analyzed with the classical regression analysis because the method requires a continuous dependent variable. A number of statistical models available for analyzing discrete dependent variables are called generalised linear models (McCullagh and Nelder, 1989). These methods model the probability of an event how likely the event is to occur. In particular, since the preliminary analysis found the occurrences of the INFO event follow a Poisson distribution, a Poisson regression model was used to capture the relationship between the occurrences of the INFO event and several contextual aspects of the collaborative hypertext database used by the students.

A Kolmogorov-Smirnov (K-S) test (e.g., Norusis, 1985) on the probability distribution of the occurrences of the INFO event yielded an adequate goodness-of-fit statistic to a Poisson distribution. A Poisson regression model can be specified via a log-linear model in which the dependent variable follows a Poisson distribution. A procedure GENLOG (for general log-linear model analysis) in Statistical Package for the Social Sciences (SPSS) was used for the modelling. The occurrences of the INFO event are the cell values of the contingency table associated with the log-linear model. Let Y(t) denote the occurrences of the INFO event in a one-hour time interval t, Xi(t) the value of the ith explanatory variable with its coefficient i in the same time interval t, the specified model has the following form:

ln(Y(t)) = constant +iXi(t)

These one-hour time intervals are consecutive for 24 hours a day and 7 days a week. The explanatory variables were transformed to their standard scores. An adequately fitted Poisson regression model was subsequently selected to represent the context-behaviour relationship in further analysis.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Method
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgement
  10. References
  11. Appendix

The nature of the situated action of the INFO event is studied via the relationship between the occurrences of the INFO event and several contextual, explanatory variables in the collaborative hypertext database. The occurrences of the INFO event are likely to indicate an increasing concern of some social and coordination issues of an ongoing collaborative process. The resultant model of the context-behaviour relationship is developed within the framework of the study so as to provide empirical evidence for the evaluation of existing models of collaborative work in CSCW studies.

Overall Usage Profile of Collaborative Hypertext Databases

The use of the five collaborative hypertext databases in the MUCH system resulted in thousands of transactions during the three-month longitudinal observation. The scale of use for each database varies. The class database was accessed by the largest group of users (User = 30) with the widest spectrum of functions from the presentation layer (Action = 25). Table 4.1 summarises the overall scale of use for each database, in terms of the total number of transactions observed, the total number of unique nodes involved, the size of the user pool, and the number of distinct functions required.

Table 4.1.  Usage of the five collaborative hypertext databases over 3 consecutive months.
DatabaseTransactionNodeUserActionStartedEnded
Groupware120897501122Tue Apr 19Mon July 25
Class92474743025Wen Apr 20Mon July 15
Management70686981622Wen Apr 20Thu July 21
OSCAR3148315718Tue Apr 19Mon July 25
General1590961021Tue Apr 19Fri July 15

The percentage use of each function available in the presentation layer is given in the appendix for each database (Table A.1). Some heavily used functions appear to entail some cognitive and behavioural overheads due to particular design decisions for the presentation layer of the MUCH system. For example, the percentage use of the UNFOLD function is predominant in all cases but the General database. On the other hand, a heavily used UNFOLD function in a particular database may indicate that the database has been mainly used for navigating and reading tasks rather than substantive authoring tasks.

The average percentage usage of the INFO function is about 1% over the five databases. It is rather small in contrast to the percentage use of functions such as UNFOLD, READING, and LOCK. The dispersed occurrences of the INFO event suggest that the INFO function is only required for some needs to do with higher degrees of coordination coupling in collaborative writing. The independent working mode is primarily supported by the MUCH system as the system is designed to facilitate asynchronous collaborative writing. Although it is difficult to define the range of tasks to be enclosed by the independent mode, it appears that tasks can be fulfilled via the INFO function go beyond this range.

Awareness Behaviours and Contextual Characteristics

The occurrences of some rare event in the world are generated by Poisson processes. The probability distribution of the occurrences of the event associated with a Poisson process is called a Poisson distribution. The Kolmogorov-Smirnov (K-S) goodness-of-fit test of the null hypothesis, that the data are drawn from a population with a Poisson distribution, indicated that the occurrences of the INFO events follow a Poisson distribution with the mean = .0308 (K-S Z = 7694, p = 5946 two-tailed). Table 4.2 shows the occurrences of the INFO event observed in one-hour time intervals.

Table 4.2.  The occurrences of the INFO event in one-hour time intervals in using the class database
Occurrences01234567
Intervals226919621131

In the vast majority of the time intervals observed, INFO events did not occur at all. In 19 time intervals, the INFO event occurred once. The higher the count of the INFO event in an interval, the fewer number of such intervals can be found, except one case (Occurrences = 6). The highest count of the INFO event for the class database is 7.

Several Poisson regression models were fitted to the time series data of the occurrences of the INFO event and corresponding contextual characteristics of the awareness-seeking path of users in a given time interval. In particular, the model specified as follows resulted in an adequate goodness-of-fit statistic for the data:

ln(count) =0 + 1*Zcredit +2*Zlevel +3*Zlink +4*Zword +5*Zsize +6*hour

The explanatory variables, or independent variables, were specified as the covariant variables in the model, except the hour variable, which is categorical. The GENLOG procedure in SPSS revealed that the model adequately represents the data set (Pearson 2= .0960, DF= 1, P = 76). The Poisson distribution of the count of the INFO event was specified in the model. Table 4.3 shows the results of the estimates for the model and associated significance tests.

Table 4.3.  A Poisson regression model of the relationship between INFO events and contextual characters of the path activated in the same time intervals. (Source Database: Class; Observation: April 20 - July 15, 1994).
ParameterEstimateSEZ-valueMarginal
xkk *p < 0.05Effect
Zlevel19.27867.56542.55*0.5938
Hour.2060.3023.680.0063
Zword.93979.7437.100.0289
Zlink-.62851.4958-.42-0.0194
Zcredit-11.21204.1423-2.71*-0.3453
Zsize-15.24722.6603-5.73*-0.4696
Constant1.40173.1006.450.0436

For a given explanatory variable x, a positive sign of the estimate suggests the likelihood of the INFO event count increases with the level of x, with the other xs held constant. A negative sign suggests that the likelihood of the INFO event count decreases with the level of x. An insignificant test at a conventional level such as 0.05 suggests that the effect of an x on the response variable is not statistically different from zero.

Three estimates in this model are statistically different from zero. Among these, Zlevel has a positive coefficient, whereas both Zcredit and Zsize have negative coefficients. For example, the estimate of level= 19.28 (p < 0.05) for Zlevel suggests that the occurrences of the INFO event in a given time interval are likely to increase as the Zlevel in the same interval increases.

The Poisson regression model can also be interpreted in terms of the marginal effect of an x on the expected INFO count. Liao (1994) discusses five ways of interpretation generalised linear models, including the two ways utilised in this paper. The marginal effect of xk on expected y is given by k, where = 0.0308 is the expected occurrences of the INFO event in a given time interval.

The marginal effect of Zlevel on the INFO count is (0.0308)(19.2786) = 0.5938. In other words, if the value of Zlevel increases by one, the expected INFO count will increase by 0.5938, other things being equal. In contrast, the marginal effect of Zcredit is (0.0308)(-11.2120) = -0.3453, suggesting that the expected INFO count will decrease by 0.3453 as the Zcredit increases by one standard deviation.

Preliminary analyses found the contextual variable Zlevel is negatively correlated with the frequency of UNFOLD and it is positively correlated with the frequency of LOCK (Pearson's r= -0.529, p < 0.001 and r = 0.187, p = 0.001, respectively).

Finally, if a dummy variable Phase is included in the model to specify the two phases of reading and authoring, the Phase variable would have the most predominant effect on expected occurrences of the INFO. The writing phase considerably increases the probability of a higher INFO count. The frequency of the INFO events was indeed increased when group essays were emerging in the MUCH system. Students were more interested in new nodes coming into the shared repository.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Method
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgement
  10. References
  11. Appendix

Early experiences with collaborative authoring hypertext systems found maintaining a mutual intelligibility among coauthors is an essential part of a successful collaboration (Irish and Trigg 1989). The positive sign of the Zlevel coefficient suggests that users' interests in social and historical issues in an evolving collaborative hypertext database tend to focus on the substantive work and less on high-level organisational work. Given that the MUCH system has been mainly used for asynchronous collaborative authoring, this finding conforms with the organisational role of the MUCH system as perceived by the members in the research group.

The Poisson regression model suggests that the INFO function is more frequently needed on nodes involved in substantive work. The use of the INFO function tends to be triggered by specific problems in an independent working mode. A field study of the use of an IBIS system also found a similar tendency.

INFO events occurred rarely with all the five MUCH databases that have been studied. The occurrences of the event were found to follow a Poisson distribution (mean = 0.0308). This finding in its own right may lead to some insights into the awareness-seeking behaviours in asynchronous collaborative authoring with the MUCH system. The knowledge of the probability distribution of the event count of the INFO is necessary for applying appropriate tools and making sense of the empirical data observed.

The relationship between the INFO events and some contextual characteristics of the paths of users interacting with a shared hypertext database reveals some patterns related to the interaction between users and the database. These factors affect users' behaviours of using the particular function of the INFO and they are likely to bear important information to users who may be motivated as to change from one mode of collaboration to another. The total number of words involved in the awareness-seeking path was expected to correlate with the occurrences of the INFO event positively, but the estimate is not significantly different from zero (word= 0.9397, p > 0.05). The overall linkage, the total number of parent child links, involved in a given interval was expected to correlate with the INFO count negatively, but the significance test did not provide sufficient evidence (linkage= -0.6285, p > 0.05) for the proposition. On the other hand, the effects of the overall level, the overall credit, and the total number of concurrent users were all statistically significant from zero. These patterns help us to understand what information is important to users. The users may be motivated to seek further social interaction and move to a working mode with a higher degree of coupling or they may remain working in the current collaboration mode with updated views to the world.

Experiences of students in group essay writing showed that the most coordination-demanding phase was the initial brainstorming. The initial brainstorming was followed by task allocations among the members of a group. The majority of the groups followed a process similarly to the ones studied by Kraut, Galegher, Fish, and Chalfonte, (1992). The detailed transactions show an increased frequency of the INFO events as the submission deadline was approaching. The regression models suggest that students needed more awareness assistance in the group writing phase than in the browsing phase. The regression model must be understood in a broader context of users, tasks, and tools.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Method
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgement
  10. References
  11. Appendix

This study reveals that a specific feature of a shared hypertext database system (e.g. INFO) may have impacts on both the process of social interaction and users' awareness-seeking behaviours. In this sense, every individual computer system shapes the process of people's work and inevitably the social structure which embeds their work.

The use of quantitative methods is not to replace qualitative methods. Rather, the intention is to explore a wider spectrum of combined methodologies to deal with the complexity of understanding the phenomenon and encoding this understanding into practical CSCW systems. Recurring patterns of the INFO event require a longitudinal observation. The generalised linear model approach could be applied to similar field studies on collaborative writing.

An important approach to organising and coordinating collaborative writing is assigning roles explicitly to group members (e.g. Neuwirth, Kaufer, Chandhok, & Morris, 1990) (Leland, Fish, & Kraut, 1988). Further work is needed to apply the probabilistic modelling method to the study of collaborative authoring with specific allocated roles. Probabilistic models may take into account additional variables such as roles as well as behavioural and cognitive variables. This modelling approach may be strengthened by incorporating knowledge elicitation methodologies such as structured interviews. The behavioural patterns of using the computer system and the dynamics in the social structure of the group in terms of allocated roles need to be further combined so as to result in an adequate understanding of the phenomenon as a whole.

Acknowledgement

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Method
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgement
  10. References
  11. Appendix

Chaomei Chen wishes to thank Dr. R. J. Bhansali, Department of Statistics and Computational Mathematics at the University of Liverpool, for his valuable advice on Poisson regression models and Dr. D. Diaper for his in-depth comments on an earlier draft of the paper. Roy Rada acknowledges the support of The Gallup Organization.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Method
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgement
  10. References
  11. Appendix
  • Bales, R. (1951). Interaction process analysis. Cambridge , Mass. : Addison-Wesley Press.
  • Chen, C. & Rada, R. (1995). Understanding collaborative authoring in shared workspaces. Proceedings of IFIP TC13 Fifth International Conference on Human-Computer Interaction (INTERACT'95), 27–29 June, 1995, Lillehammer , Norway .
  • Chen, C., Rada, R. & Zeb, A. (1994), An extended fisheye view browser for collaborative writing, International Journal of Human-Computer Studies, 40(5), 859878.
  • Conklin, J. & Begeman, M. (1988). gIBIS: a hypertext tool for exploratory policy discussion. ACM transactions on office information systems, 6(4), 303331.
  • De Sanctis, G. & Poole, M. S. (1994). Capturing the complexity in advanced technology use: adaptive structuration theory. Organization Science, 5(2), 121147.
  • Diaper, D. & Addison, M. (1991). User modelling: the task oriented modelling (TOM) approach to the designer's model. D.Diaper and N.Hammond (Eds.) People and computers VI: Proceedings of the HCI'91 Conference, pp. 387402. August 20–23, 1991. Cambridge : Cambridge University Press.
  • Dourish, P. & Bellotti, V. (1992). Awareness and coordination in shared workspaces. Proceedings of the ACM Conference, CSCW'92, pp. 107114, New York : ACM Press.
  • Festinger, L. (1957). A theory of cognitive dissonance. Standford University Press.
  • Fulk, J. Schmitz, J. A. & Schwarz, D. (1992). The dynamics of context-behaviour interactions in computer-mediated communication, in MartinLea (ed.), Contexts of computer-mediated communication (pp. 730), Harvester Wheatsheaf.
  • Giddens, A. (1984). The constitution of society: outline of the theory of structuration, Chicago , IL : Polity Press.
  • Haake, J. & Wilson, B. (1992). Supporting collaborative writing of hyperdocuments in SEPIA. Hypertext'92 proceedings, pp. 138146, New York : ACM Press.
  • Halasz, F. & Schwartz, M. (1994). The Dexter hypertext reference model. Communications of the ACM, 37(2), 3039.
  • Irish, P. & Trigg, R. (1989). Supporting collaboration in hypermedia: issues and experiences. Journal of the American Society of Information Science, 40(3), 192199.
  • Kraut, R., Galegher, J., Fish, R. & Chalfonte, B. (1992). Task requirements and media choice in collaborative writing. Human-Computer Interaction, 7, 375407.
  • Lea, M. (ed.) (1992). Contexts of computer-mediated communication, Harvester Wheatsheaf.
  • Leland, M., Fish, R., & Kraut, R. (1988). Collaborative document production using Quilt. Proceedings of CSCW'88, pp. 206215. September 26–28, 1988, Portland , OR . New York : ACM Press.
  • Liao, T. F. (1994) Interpreting probability models: logit, probit, and other generalized linear models. Sage University Paper series on quantitative applications in the social sciences, 07–101. Thousand Oaks , CA : Sage.
  • Markova, I. (1987). Human awareness: its social development, London : Hutchinson.
  • McCarthy, J. (1994). The state-of-the-art of CSCW: CSCW systems, cooperative work and organization, Journal of Information Technology, 9, 7383.
  • McCullagh, P. & Nelder, J. A. (1989). Generalized linear models (2nd ed.). London : Chapman & Hall.
  • Neuwirth, C., Kaufer, D., Chandhok, R., & Morris, James. (1990). Issues in the design of computer support for co-authoring and commenting. Proceedings of CSCW'90, pp. 183194. October 7–10, 1990. Los Angeles , CA . New York : ACM Press.
  • Norusis, M. J. (1985) SPSS-X advanced statistics guide. SPSS Inc.
  • Rada, R. (1989). Guidelines for multiple users creating hypertext: SQL and HyperCard experiments. NoelWilliams (Ed.) Computers and writing: models and tools, Blackwell/Ablex Publishing, pp. 6189.
  • Suchman, L. (1987). Plans and situated actions: the problems of human-machine communication. Cambridge : Cambridge University Press.
  • Yakemovic, K. C. B. & Conklin, E. J. (1990). Report on a development project use of an issue-based information system. Proceedings of ACM Conference CSCW'90, pp. 105118, New York : ACM Press.

Appendix

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Method
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgement
  10. References
  11. Appendix
Table A.1.   The percentage use of each function used with the five collaborative hypertext databases.
Actiongware(%)class(%)man.(%)oscar(%)gen.(%)
TOTAL12089100.09247100.007068100.03148100.001590100.0
Unfold438336.26304732.95278139.35104033.0423714.91
Reading313625.94378340.91199528.2378825.0339124.59
Lock130110.763323.595467.7238312.1734321.57
Fold121010.018469.156689.4532510.32674.21
Update11889.832322.514446.2934711.0233421.01
Delete1841.5222.2415.21401.2710.63
Quit1741.442372.562793.95782.48664.15
Modify1501.2410.1134.4814.443.19
Info90.741391.5046.6525.797.44
Rename84.7010.1116.23331.052.13
Save66.5574.8014.20351.11191.19
Createnode63.5270.761181.6722.70553.46
Send20.171041.1231.440.006.38
Refresh15.1232.3516.239.2913.82
Word8.0748.522.030.000.00
Preview4.0367.733.041.0313.82
Print4.0341.4415.210.0013.82
Users3.0258.637.103.107.44
Generating2.0210.114.062.061.06
Check2.022.021.010.001.06
Author1.0122.240.002.061.06
Browser1.0012.1331.441.031.06
Assessment0.0040.432.030.000.00
Last0.007.080.000.000.00
Reply0.002.020.000.000.00