Evaluating Cooperation in Group Work

Luis A. Guerrero Rosa Alarcdn

Depto. de Ciencias de la Computacio'n Depto. de Ciencia de la Computacio'n
Universidad de Chile Pontificia Universidad Cato'lica de Chile
Casilla 2777, Santiago, Chile Santiago, Chile
luguerre @dcc.uchi1e.d ralarcon @ ing.puc.cl

Ctsar Collazos Jose A. Pino

Depto. de Sistemas, Facultad de Ingenieria Depto. de Ciencias de la Computacio'n
Electrdnica y Telecomunicaciones Universidad Universidad de Chile
del Cauca, Colombia Casilla 2777, Santiago, Chile
ccollazo @ ucauca.edu.co jpino @dcc.uch!ile.cl
David A. Fuller
Depto. de Ciencia de la Computacibn
Pontificia Universidad Catdlica de Chile
Santiago, Chile
[email protected]

Abstract studies, and different concepts in others. The same

happens with the terms cooperutive work, collaborative
Several researchers in the area of cooperative work work and group work. This fact, perhaps, is due to the
take as a success criterion the quality of the group approach change that research in the area has undergone
outcome. Nevertheless, recent findings are giving more with time.
importance to the quality of "cooperation process" itself: Dillenbourg et a1 maintain that during many years,
This paper presents the design of a software tool that theories of collaborative learning have been focused on
allow us to capture information concerned with the group how individuals work in group, and recently, they have
work. Besides to reveal the final quality of the group focused on the group by itself, trying to establish when
outcome, this information permit us to verify the presence and under what circumstances Icollaborative learning is
of a cooperation indicators, which in turn allow us to more effective than individual learning [6]. In this
determine the quality of the work process. A conducted context, some independent variables have been identified
experiment is described, and the obtained data is and widely studied: the size and composition of the
discussed, comparing the quality of the collaboration group, the nature and the objectives of the task, the media
process with the quality of the results obtained by the and communication channels, the interaction between
groups. peers, the reward system and sex differences, among
others [I], [6], [13], [14]. Recenl research is setting more
Keywords: cooperative process evaluation, cooperation emphasis on the study of collaboration processes and
indicators, cooperation quality. their support [4], [5].
Traditionally, the evaluation of collaborative
learning has been made by means of examinations or
tests to the students to determine how much they have
1. Introduction learned. That is to say, we do a quantitative evaluation of
the quality of the outcome. Some techniques of
Several researches, both in the area of education and cooperative learning use this strategy (e.g. "Student Team
computer science, give different meaning to the terms Learning" [ 121, "Group Investigation" [ 111, "Structural
cooperation and collaboration. Thus, terms such as Approach" [ 101 and "Learning Together" [ 81).
cooperative learning, collaborative learning and even Nevertheless, little investigation has been done to
group learning are considered similar concepts in some

28
$10.00 0 2000 IEEE
0-7695-0828-6/00
evaluate the quality of the collaboration process. 1. Design the content and main tasks objectives to be
Some authors settle a difference between collaboration accomplished by cooperative groups (pre-process).
and cooperation according to where the emphasis has 2 . Specify the size of the groups. It is suggested to be
been put: this can be in the process or in the final result. up to 6 people depending on the nature of the task
In this paper, we do not make differences between the and the time available (pre-process).
terms cooperation and collaboration. We are focused on 3. Arrange the groups. Designate the students to
the comparison of the quality of the collaboration conform each group or allow them to form the groups
process against the quality of the group outcome. by their own (pre-process).
Taking into account characterization of cooperative 4. Arrange the room for the cooperative learning
learning presented by Johnson & Johnson, we propose a activity. The facilitator must be “attainable” by every
set of indicators and an experiment with a tool group and their members can seat together with out
instrumented to gather information that permit us to interrupt other groups (pre-process).
measure the quality of cooperation process. Also, we do 5. Distribute the instructional material. This can be
not introduce a conversation protocol, so the achieved of several forms (pre-process).
communication occurring in the task is spontaneous, 6. Design roles, such as: speaker, facilitator, recorder,
requiring semantic analysis. executor, and observer (pre-process).
In section 2, we present the Johnson & Johnson 7. Specify the task directives: define the game rules
characterization of collaborative learning process. Next, (pre-process).
we divide it into three phases according to the nature of 8. Apply strategies like positive interdependence of the
the task. In section 3, we propose an evaluation goal, motivation of the peers and support to learning.
instrument. In section 4 we describe the cooperation Create a product related to a goal system where
indicators as well as a method that permit us to evaluate rewards are based on individual and group results (it
some key points identified in the phases of collaborative is defined in the pre-process, but evaluated in the in-
learning. Section 5 describes the experiment design. In process phase).
section 6 we analyze the obtained results, and in section 7 9. Organize the intra-group cooperation, that is to say,
we compare the collaboration process versus the group define the collaboration strategies that are going to be
outcome. Finally, section 8 presents some conclusions used by the members of the group (pre-process, the
and proposals for future work. definition of cooperation strategies occurs in the in-
mocess phase).
2. Stages of cooperative learning process 10. Test the success criteria explaining the guidelines,
limits and roles (pre-process, in-process and post-
process phases). The success criteria must be defined
A cooperative learning process is typically at the beginning of the activity, and must be reviewed
composed of several tasks that must be developed by the during the activity to check if the common goal is
cognitive mediator or facilitator, and by the group of being reached, and after the activity, to check if the
apprentices, defining naturally two categories of tasks. common goal was reached.
In order to evaluate the cooperative learning process, 11. Determine the desired behavior (pre-process,
we divide it into three phases according to its temporal definition of desired behavior occurs in the in-process
execution: pre-process, in-process and post-process. phase).
Thus, pre-process tasks are mainly coordination and 12. Monitor the students, for example, verify that the
strategy definition activities and post-process tasks are previous point is fulfilled (phase of in-process).
mainly work evaluation activities. Both phases, pre- 13. Provide assistance when someone asks for it (in-
process and post-process, will be accomplished entirely process phase): it is provided to the whole group by
by the facilitator. The tasks concerning to the in-process the facilitator or peers.
phase will be performed, to a large extent, by the group 14. Intervene when groups have problems to collaborate
members. It is here where the interactions of cooperative (in-process phase).
work process takes place, so that, our interest 15. Terminate an activity (post-process phase).
concentrates in the evaluation of this stage. In order to 16. Evaluate the quality of learning accomplished by the
specify this division, we present the structure of a students (post-process phase).
cooperative learning activity proposed by Johnson & 17. Encourage students to perform an evaluation on how
Johnson in [l], and next we classify each activity well does the group work altogether (at the end of the
according to the stage we have identified’: in-process phase).
18. Provide and foster feedback. Discuss how the
activities could be improved (at the end of in-process
phase).
Johnson &Johnson do not make this phase differentiation.

29
 Table 1. Activities of a cooperative learning process.

Pre-process In-process Post-process

Design the contents Application of strategies (positive Inspect success criteria
interdependence of the goal, motivation
between pairs, aid to learn)
Specify the group sizes Intra-group cooperation Present the activity closure
Arrange the groups Probe the success criteria Evaluate the quality of
learning
Arrange the room Monitoring
Distribute the material Provide help (from facilitator and from peers)

Design the roles Intervention in case of problems

Specify the game rules Account of the group
Define the success criteria Feedback
Determine the desired behavior

Table 1 summarizes the activities and specifies the collaborator (the three remaining). Also, each player is
corresponding phases. These activities define the identified by a color, and the colored square besides the
structure of any cooperative learning activity that takes quadrant indicates who is the coordinator (5).
place in small groups, and in synchronous learning A coordinator must take decisions concerned to the
scenarios (face to face, same time, same place). We are movement of the mouse. The collaborators must support
interested in the evaluation of the activities that the coordinator to ensure that her decisions are accurate.
correspond to the in-process phase. Based on these, we When a coordinator leads the mouse to the traffic-light
will define some collaboration indicators. icon of her quadrant (or the cheese in the last one), her
The following section presents a software tool used role is switched to collaborator and the coordinator role
as an instrument to evaluate the presence or absence of then, is assigned to the next player (clockwise).
our indicators of collaboration. In section 4 we define The arrows to drive the mouse (4) are only enabled
these indicators. when the user is performing the role of coordinator. In
each quadrant there are two types of obstacles through
where the mouse cannot pass: general obstacles or grids
3. Chase the Cheese
(6) and colored obstacles (7). Though grids can be seen
by every player, colored obstacles can only be seen by
“Chase the Cheese” is a group game, implemented in the player who has that assigned color. As an example,
a software tool that allows us to capture group Figure 1 corresponds to the interface of the player
information that occurs within a cooperative work assigned with the yellow color. For that reason, only the
interaction. In this section we describe the functionality general grids (6) and the yellow obstacles can be seen (7)
of the system and the nature of the information registered in her interface.
when a group uses the tool. Since each participant has a partial view of the
labyrinth, he or she must interact with their peers to solve
3.1. System’s functionality the problem. In order to comrriunicate with them, each
player has a dialogue box (8) from which she can send
The game is a labyrinth divided into four quadrants. messages for each of them explicitly (once at a time)
Figure 1 shows the game interface. To win the game, the through a set of buttons associated to the color of the
group (four players) must lead the mouse (1) to its cheese destiny (9). For example, in Figure 1 , he or she can send
(2). It is the common goal, and to achieve it, every group messages to the players with ccdors blue, red and green.
member must fulfill a partial goal that is accomplished Also, each player can only see the messages that the
when every one of them “solve” their own quadrant. That other players send to him or her in their mailbox
is to say, the mouse is driven to a traffic-light icon (3) messages (lo), one for each player. Every mailbox has a
allowing the group to advance to the next quadrant. This title that displays the name of the player, as well as its
is a decision making problem game. color.
Each player has two predefined roles: coordinator At the beginning of each quadrant, the coordinator
(only one per quadrant and randomly assigned) or has a partial score (11) of 100. Whenever the mouse

30
 8

Figure 1: "Chase the Cheese" game interface.

hits an obstacle, her partial score decreases by 10 units. strategies, intra-group cooperation, checking the success
When the coordinator reaches the traffic-light icon of criteria, monitoring and the ability of providing help. So
her quadrant (the cheese in the last quadrant), the partial that, we defined four categories of messages:
score is added to the total score of the group (12). The coordination, work, strategy and lateral messages.
objective is not just to reach the cheese, but to arrive with Subsequently, we established the Index of Cooperation
the highest score (the maximum would be 400). If some (IC) as the average of these five indicators. This index
partial score goes down to zero, the group loses the game. allows us to evaluate the work process of the groups. In
Both scores, partial and total, are hidden, if a player order to determine the values of our indicators, it is
wants to see them, he or she must pass the mouse over necessary to make semantic analysis of the messages.
their icon (1 1 or 12), unfolding the score by two seconds. In the first category, "coordination", we recognize
those messages that correspond to activities which its
3.2. Information gathered main purpose is to regulate the dynamics of the process,
and are characterized by prescribed future actions, such
as "I will move six to the right". In the "work" messages
The application records every message sent by any
category, we include those messages that help the
member of the group. Along with each message, it
coordinator to make the most suitable decisions. Those
registers the time of occurrence, sender, addressee and
are sentences in present tense and have the aim to inform
current quadrant (the mouse location when the message
the group about the current state of the group task. In the
was sent). In addition, it records the partial scores and
experiment, it is referred to the state of the quadrant and
total score by quadrant. The tool also registers the start
the game. For example: "Stop, there is an obstacle in B3".
and finish time of the game, the time spent in each
We identify as "strategy" messages all those that
quadrant, and the number of times that each player looks
propose guidelines to reach the group goal, for example:
at the partial and total scores by quadrant.
"Let's label the columns with letters and the rows with
numbers". Finally, the "lateral" message category
4. The indicators includes the kind of particular messages (i.e. social
messages, comments) and conversations that are not
Based on the structure of a cooperative learning focused on the solution of the problem, for example:
activity explained above in section 2 (in-process phase), "Come on, hurry up, I'm hungry!!!!!!! 'I.

we define five indicators in order to evaluate the Next, we define the five indicators, along with the
cooperation degree in the group interaction. These mechanism provided by the game to determine its
indicators are based on the following activities: use of presence or absence during the group interaction.

31
4.1. Applying strategies 4.4. Monitoring

Because each player has a partial view of the game The monitoring is understood as a regulatory activity. The
obstacles, to solve the problem they have to interact objective of this activity is to oversee that the group
closely with their group peers. Due to this necessity, the maintain the chosen strategies to solve the problem,
game presents a strict positive interdependence of goals. keeping focussed on the goals and the success criteria. If a
If the group is able to solve the game, we can say that player do not sustain the expected behavior, the group
their members have built a shared understanding of the will not reach the common goal. In this sense, our fourth
problem (see Dillembourg definition of collaboration cooperation indicator ((214) will! be related to the number
[6]). They must have understood the underlying problem: of coordination messages, where a few amount of
the coordinator does not have all the information needed messages means a best coordination. It is expresed in a
to move the mouse in her quadrant without hitting any range from 0 to 1 where 1 indicates a better coordination.
obstacle, so she needs the opportune assistance from each
collaborator. According to Fussell [7], the discussion of 4.5.Providing help
the strategy to solve the problem, helps the group
members to construct a shared view or mental model of
their goals and tasks required to execute, this mental This activity is our fifth cooperation indicator (CI5), and
model can improve the coordination, because each is based on the ratio between the number of work
member knows how their task fits into the global team messages and the total amount of messages spawned by
goals. In this context, due to its high degree of inter- the group. Each player requires, help from their peers to
dependence, it’s enough to consider the success or failure reach her partial goal when acting as a coordinator, the
in solving the labyrinth to know if the group has applied ratio represents the involvemerit and concern degree of
strategies. This will be our first cooperation indicator players in their role of colaborator (coordinator assistant).
(CIl). CII has a discrete value: 1 if they can solve the CI5 has a range from 0 to 1, wlhere 0 means less interest
labyrinth (global goal) and 0 if they cannot. and help (few information prcwided to the coordinator
regarding the status of quadrant) and 1 , a high degree of
commitment to provide enough help.
4.2. Intra-group cooperation The activity “intervention in case of problems”, is
not considered because we don’l have a facilitator, except
This activity corresponds to the application of at the beginning of the game to explain the rules, so the
collaborative strategies during the process of group work. help request through questions regarding how to play is
If each group member is able to understand how her task not registered by the instrument).
is related with the global team goals, then every one can At the end of the game a didog window is presented
anticipate their actions, requiring less coordination efforts. to the players with the question: “What do you think of
Because of this, our second cooperation indicator (CI2) is this experience?” An analysis of these answers would
related to the “work” messages category: a good indicate if the group is able to self-evaluation, if they are
coordination procedure and a shared strategy, should be able to understand the game, that is to say, to construct a
reflected in an efficient and fluid communication were the shared understanding of the problem. However, due to its
messages are precise, accurate and opportune, thus subjective nature, the self-evaluation and feedback
requiring fewer messages. To determine CI2, we consider activities are not evaluated in this experience.
a scale from 0 to 1, were 1 means fewest work messages
sent between the group members. 5. Experimental design
4.3. Success criteria review The experiment has four phases. In the first phase the
group receives a brief description of the software tool, in
In the game, the success or failure of the group, related to the second phase, group members are assigned to
the partial and global goals, is shown in the scores network workstations, in separate rooms (synchronous
obtained (partial and global scores). The third cooperation distributed interaction). From now on, all the
indicator (CI3) reflects the interest in the individual and communication is mediated by computer. During the
collective performance. The more concerned the player is third phase the group will try to solve the labyrinth.
with the goals of the team, the most queries to the scores Finally, the fourth phase corresponds to the gather and
she will do. C13 is defined through a range from 0 to 1 analysis of data recorded in the tool logs. We made also
where 1 means a high number of score queries. a final interview to the participants to foster a self-
evaluation of the experience, thiis will give us a general
overview of the problem perceived by each member of

32
the team. By the time, we have applied the experiment to messages decreases throughout the game, not allowing
seven groups, as we describe: them to end the game (common goal).

A group of post-graduated students, from the course 6.2. Work messages

“Collaborative Systems” at the Pontificia
Universidad Cat6lica de Chile, with some experience Work messages of both groups, 6 and 3, are shown
on collaborative work techniques (group 0). in Figure 3. The curves pattern is similar to that of Figure
A group of people, randomly selected, wich have not 2. It suggest that there is a close relationship among those
been meet among them and, of course, they never categories, where coordination activities regulate the
have worked as a group before (group 3). tasks, but require a feedback from the players.
0 A group of friends that has worked in group many
times before the experience and has a good personal
6.3. Strategy messages
relationship (group 4).
Four groups of high school students from Cumbres
Figure 4 shows the tendency of strategy messages
de Santiago with an average age of 15 years old. Two
curves for both groups, 6 and 3. As in the previous charts,
of these were randomly selected (group 1 and 2) and
the remaining ones were friends (group 5 and 6). the curve corresponding to group 6, presents a high
degree of regularity, but a decreasing tendency in the
group that was not able to solve the problem (group 3).
6. Results analysis
6.4. Lateral messages
We define the Index of Cooperation (IC) as the
average result of the previous identified indicators: IC = Figure 5 presents the results obtained in the lateral or
[CIl + CI2 + C13 + C14 + CI51 / 5). sided messages category for gopus 3 and 6. Notice the
This section present a quantitative analysis of the similarity of both curves, as in the tendency as well as in
data recorded, corresponding to the group wich has the total amount of messages sent throughout the four
obtained the best IC (group 6) and the group with the quadrants. This suggest that both groups develops, in a
worst one (group 3). We present, also, a preliminar simmilar degree, a social and casual interaction behavior,
interpretation of these results. Table 2 shows the results but this does not influence significantly the performance
obtained by the groups. obtained by the group, that is to say , a greater amount of
messages does not imply a high degree of contribution
6.1. Coordination messages and commitment to reach the common goal.

Figure 2 compares the results obtained by groups 6 7. Process versus outcomes

and 3 in the coordination messages category throughout
the four quadrants. It is important to notice that the curve
Baeza-Yates and Pino [3] made a proposal for the
corresponding to the number of this sort of messages of
formal evaluation of collaborative work. They take in to
group 6 (best coordinated according to IC), keeps a
account three aspects: Quality (how good is the result of
flatten curve throughout the game., wich differs from
collaborative work), Time (total time elapsed while
group 3. In this case, the number of coordination
working) and Work (total amount of work done).
So that, in regard to our experiment, Quality can be

Table 2. Results of the Cooperation Indicators

Group CI1 CI2 C13 C14 CI5 IC

0 1 .oo 0.80 0.20 1 .oo 1 .oo 0.80
1 0.00 0.40 0.20 0.60 0.80 0.40
2 1.oo 1.oo 0.20 1 .oo 0.40 0.72
3 0.00 0.20 0.40 0.20 0.60 0.28
4 1 .oo 0.40 0.80 0.40 0.20 0.56
5 1.oo 1.oo 1.oo 1 .oo 0.20 0.84
6 1 .oo 1 .oo 1 .oo 1 .oo 0.60 0.92
 MESSAGES CATEGORY COORDINATION MESSAGES CATEGORY STRATEGY
~

I 20 7

j.&qR*i
+Group 6
Quadrants

Figure 2. Coordination messages of groups 6 and 3 Figure 4. Strategy messages of groups 6 and 3
throughout the labyrinth quadrants. throughout the labyrinth quadrants.

MESSAGES CATEGORY: LATERAL

6
2 7 1 5
Quadrants

Figure 3. Work messages of groups 6 and 3 throughout Figure 5. Lateral messages of groups 6 and 3 throughout
the labyrinth quadrants. the labyrinth quadrants.

measured by three factors: fewest errors done by the Table 3, is the group 6, which is the group that obtained
group (related to the best score), achievement of the main the best IC in the collaboration process evaluated in the
goal (the group can solve the labyrinth) and fewest previous sections (Table 2).
movements of the mouse (related to efficiency). The tool
records the play-time since the first event (movement of 8. Conclusions and further work
the mouse or message sent by any player), until they
reach the goal (cheese) or loose the game (a partial score
In this paper we presented a software tool that
goes down to zero). In this view, the “best” group does
allowed us to enable a collaborative work computer-
the Work faster. The Work is reflected by the amount of
based activity, and to gather information concerned to it,
messages sent by the members of the group.
in order to evaluate the cooperation proccess that occurs
in the group work. For the evaluation of the cooperative
Table 3. Final product evaluation. procces, we have identified five cooperation indicators,
wich in turn allow us to define a cooperation index. As
well, to evaluate the outcome of Ihe group work, we used
three factors: Quality, Time, and ‘Work.
As a major conclusion we can say that the process
quality of the work group is not directly related to the
final product quality. In this experiment, the group that
0.62 I 0.42 I 0.95 1 0.66 has the best results in the process quality evaluation has
5 I 0.74 I 0.83 1 0.27 I 0.61 the worst in the product quality measurement.
6 1 0.56 I 0.81 I 0.19 I 0.52 It is also important to headline that the cooperative
work process is influenced by the personal style and
Table 3 summarizes the results (average value) obtained individual behavior of every member of the group. We
by the groups, notice that the “worst” group, according to can observe a stability in the performance of the tasks

34
accomplished by each one of the group members, as
much in their coordinator role like in their collaborator [3] Baeza-Yates, R. and Pino, J.A. A First Step to Formally
role. This stability is also observed in the personal styles Evaluate Collaborative Work. ACM International Conference
and communication skills. on Supporting Group Work, GROUP’97, Phoenix, AZ, USA,
The results suggest that the shared construction of a NOV.1997, pp. 55-60.
strategy to fullfill a group work, understood and adopted
[4] Barros, B. and Verdejo, M. F. An Approach to Analyse
by every member of the group, is related to a successful Collaboration when Shared Structured Workspaces are Use for
process, to the individual construction of cognitive Carrying Out Group Learning Processes. In S.P. Lajoie and M.
context, and to the experiences shared by the group Vivet (editors), Artificial Intelligence in Education, 1999.
members. Also, it enhances the elaboration process of
strategies and facilitates its application. This fact is [5] Brna P. and Burton M. Roles, Goals and Effective
reflected in the performed lenguage utterances: those are Collaboration. Proceedings of the IV Collaborative Learning
homogeneous, direct and unambiguous when refered to Workshop in the 8‘h World Conference on Artificial Intelligence
the common problem features. in Education, Kobe, Japan, 1997.
There is a direct relationship between the control
activities (coordination messages) and the execution of [6] Dillenbourg, P., Baker, M., Blake, A. and O’Malley, C. The
Evolution of Research on Collaborative Leaming. In Spada, H.
the task (work messages) beyond its final result (success
and Reimann, P. (editors), Learning in Humans and Machines,
or failure). This suggests that those variables do not 1995.
forecast a satisfactory result for the group, it is a clear
understanding of the problem and the solution strategy [7] Fussell, S., Kraut, R., Lerch, F., Scherlis, W., McNally, M.
which lead to a success process. In this way, if every and Cadiz, J. Coordination, Overload and Team Performance:
member of the group understands the problem, they will Effects of Team Communication Strategies. Proceedings of
require less communication efforts. Besides, too much CSCW’98, Seattle, Washington, USA, 1998.
communication, not focused in the major goals, has a
negative effect into the accomplishment of the goal [8] Johnson, D. and Johnson, R. Leaming Together and Alone.
(disturbance, cognitive overload, etc.). Cooperation, Competition and Individualization. Prentice Hall
Finally, is important to notice that the four categories Inc. Englewood Cliffs, New Jersey, 1975.
identified (coordination, work, strategy, and lateral
messages) give us significative insights into the [9] Inaba, A. and Okamoto, T. The Intelligent Discussion
Coordinating System for Effective Collaborative Learning.
interaction process embedded in a group work, wich Proceedings of the IV Collaborative Learning Workshop in the
pursuits a goal and a concrete objective, and requires the 81h World Conference on Artificial Intelligence in Education,
attention and participation of every member of the group. Kobe, Japan, 1997.
As for further work, we’ll improve the interface
design of the tool and will use it in more experiments. [IO] Kagan, S. The Structural Approach to Cooperative
This will allow us to obtain a significative amount of data Leaming. Educational Leadership, Vo1.47, No.4, 1990, pp. 12-
to analyze and compare, in order to generalize our 15.
observations.
[I 11 Sharan, Y. and Sharan, S. Group Investigation Expands
Cooperative Learning. Educational Leadership, Vo1.47, No.4,
Acknowledgements 1990, pp.17-21.

This work was partially supported by grants 1000870 [12] Slavin, R., Madden, N. and Stevens, R. Cooperative
and 196-0960 from FONDECYT (Chile) and grant No. I- Learning Models for the 3 R’s. Educational Leadership. Vol. 47,
015-99/2 from Direccidn de Investigacidn y Desarrollo No.4, 1990, pp.22-28.
(DID), Universidad de Chile.
[ 131 Slavin, R. Synthesis of Research on Cooperative Learning.
Educational Leadership, Vo1.48, No.5, 1991, pp.71-82.
References
[14] Underwood, G., Mc.Caffrey, M. and Underwood, J. Gender
[l] Adams, D. and Hamm, M. Cooperative Leaming, Critical Differences in a Cooperative Computer-based Languaje Task.
Thinking and Collaboration Across The Curriculum. Second Educational Research, Vo1.32, No.1, 1990, pp.44-49.
Edition, 1996.

[2] Baecker, R. Readings in Groupware and Computer

Supported Cooperative Work: Assisting Human-Human
Collaboration. San Mateo, CA. Morgan Kaufmann Publishers,
Inc., 1992.

35