Cooperative learning involves people of equal status working together to enhance their individual acquisition of knowledge and skills. It can be contrasted with two of its close relatives: tutoring and team training. Tutoring involves a clear distinction in status (expertise) among the participants; team training focuses on the enhancement of team (group) rather than individual outcomes.
Cooperative learning has a rich empirical and pragmatic history, although most of the systematic work has focused on children. Consequently, the general reviews and theoretical articles are heavily weighted by outcomes from experiments with learners in grades 2-9. (For reviews, see D.W. Johnson and R.T. Johnson, 1989; Nastasi and Clements, 1991; Totten et al., 1991 provide an annotated bibliography of 818 studies.) Studies with adultscollege students and technical traineesare included in the research, but they are not generally given special status. The assumption seems to be that cooperative learning principles, methodologies, and findings are applicable across ages and learning contexts.
This chapter explores some of the boundaries of that assumption by examining adult cooperative learning in light of the general literature. We begin with a description of the prototypical elements of cooperative learning and then use that description to guide our review of the theories and research and their limitations. We then consider in detail the research on cooperative learning in adult populations and its implications for future research and implementation.
In general, this chapter focuses on studies that compare cooperative and individual learning on the basis of dependent measures of individual achievement
or performance. It is concerned with the factors that make cooperative learning more effective so that it can produce results that are superior to some central condition of individual learning.1 This focus excludes an extensive body of research that compares group products (outcomes) with individual outcomes (for a review see D.W. Johnson and R.T. Johnson, 1989). It also excludes cooperative learning effects on nonperformance variables, such as intrinsic motivation (see, e.g., D.W. Johnson et al., 1985), self-esteem (see, e.g., D.W. Johnson and R.T. Johnson, 1985a), and attitudes toward minorities (see, e.g., Bossert, 1988). Positive gains on these variables may have indirect influences on subsequent individual achievement, but we did not find any published research on this connection.
KEY ELEMENTS OF COOPERATIVE LEARNING
Many different types of cooperative scenarios have been developed and explored; most can be characterized in either of two ways.
One is team learning, in which students are directed to assist each other in learning a body of material (e.g., DeVries and Edwards' teams-gamestournaments techniques, 1974; Slavin's student teams-achievement division, 1983). The other is expert groups, in which students become experts in a content area and then teach groupmates (e.g., Aronson's  jigsaw approach). Although there are tutoring episodes within this format, equal status is maintained across members of the learning groups over time.
The implementation and evaluation of most team learning and expert groups involve as many as four phases (see Figure 5-1): (A) precooperation instructions and activities; (B) the cooperation episode; (C) postcooperation activities (often not included); and (D) outcome assessment. Figure 5-1 also shows information about the general cooperative learning literature: the elements in the unshaded boxes labeled 1 in phases A, B, and D have been the focus of prior research; the elements in the shaded boxes labeled 2 and in phase C have been largely neglected.
Precooperation instructions and activities (phase Al) have received the most attention in cooperative learning research. Most of the work considers the establishment of positive interdependence among the group members as the critical step in promoting successful cooperative learning (e.g., D.W. Johnson and R.T. Johnson, 1989; Slavin, 1992). This interdependence is presumably heavily influenced by prescribed goals, incentives, tasks, cooperative instructions, and group assignments. The general objective of this phase is to cultivate a belief among group members that their individual success is positively linked to the success of the other group members. This positive interdependence presumably leads to productive activities during the cooperative episode and subsequently to positive outcomes on individual achievement measures.
It has been recognized that positive interdependence may not be sufficient if the participants do not possess the appropriate skills and strategies for interacting with one another (e.g., D.W. Johnson et al., 1990). To remedy this situation, pretraining for the cooperative episode (phase A2) involving social skills training (e.g., Mesch et al., 1988) or more specific training on enacting a cooperative script (e.g., O'Donnell et al., 1987) has been used. Unfortunately, there is very little empirical work on the effectiveness of these types of pretraining.
During the cooperative episode itself (phase B), the participants typically direct their own specific activities and interactions under the loose guidance of whatever cooperative format they are implementing. The general expectation is that the participants will engage in ''face-to-face promotive interactions" (R.T. Johnson et al., 1990) during this episode. For example, in some cooperative scenarios, learners are expected to promote each other's success by giving and receiving assistance, exchanging information, giving and receiving feedback about performance, challenging each other's ideas, building trust, mutually influencing each other, and reducing each other's anxiety about failure. Although this phase is considered to be the heart of cooperative learning, there have been relatively few efforts to directly examine, measure, or influence these interactions (phase B2) except for informal instructor monitoring. The use of explicit, detailed roles and scripts to guide the participants during the cooperation episode has received only limited attention (for a review, see O'Donnell and Dansereau, 1992). Also, there have been only scattered studies examining the effects on cooperative learning outcomes of visual aids (e.g., Patterson et al., 1992, 1993; Rewey et al., 1992), computer assistance (Hythecker et al., 1985; R.T. Johnson et al., 1986; Rocklin et al., 1985), and expert facilitators (e.g., teachers; Brown and Palinscar, 1989).
Closely related to pretraining is the postcooperation activity of "group processing" (phase C). In this activity, group members (often guided by an instructor) examine how well they are functioning in the group and how this functioning might be improved prior to engaging in another cooperative episode. In essence, group processing acts as pretraining for future episodes. As indicated by the shading, only a few studies have explicitly examined the impact of group processing on subsequent outcomes (e.g., R.T. Johnson et al., 1990; Yager et al., 1985).
The final phase, outcome assessment (phase Dl), has typically involved the evaluation of the direct acquisition of knowledge and skills from the tasks and materials provided by the instructor. An exemplary study by Fantuzzo et al. (1989a) compared four groups of college students on a comprehensive final exam in an abnormal psychology course. Students were assigned to one of four conditions: (1) in the dyadic structured format (DS), students intermittently developed and answered test questions during
a structured series of reciprocal tutoring sessions with a randomly assigned partner; (2) in the dyadic unstructured format (DU), students intermittently prepared for and discussed general course topics with a randomly assigned partner (no specific structure for these discussions was provided); (3) in the independent structured format (IS), students developed test questions and answers in the same manner as the DS group, but they did not interact with a partner; and (4) in the independent unstructured format (IU), students prepared discussions on general topics in the same manner as the DU group but did not interact with a partner. The means and standard deviations on the comprehensive course exam: DS, M = 84.8, SD = 10.1; DU, M = 70.1, SD = 15.5; IS M = 69.0, SD = 17.5; IU M = 66.3, SD = 18.8. The results showed significant main effects for the dyadic formats versus the independent ones and for the structured formats versus the unstructured ones.
This study is exemplary in its use of a collection of appropriate comparison groups: both individual learners and partners implementing an alternative cooperative strategy. Many of the studies reported in the literature use only an individual comparison group without strategy instructions.
To summarize, cooperative learning as studied typically involves the specification of a task, a cooperative technique, the goals and incentives for the cooperative episode, and the assessment of knowledge and skills directly acquired from the learning episode. Pretraining, direct support and scripting of the cooperative episode, postcooperation activities, and assessment of transfer to new individual learning tasks are only rarely included.
GENERAL THEORETICAL PERSPECTIVES AND FINDINGS
Two Theoretical Perspectives
Two general theoretical perspectives appear to have guided research on cooperative learning. The first one, social-behavioral, emerges primarily from social psychology and, to a lesser extent, from behaviorism. Investigators who emphasize this approach tend to focus on establishing conditions that promote effective cooperative learning (the elements in phase Al of Figure 5-1). Within this perspective are two distinct viewpoints that emphasize different types of preconditions: social interdependence, which emerges directly from group research in social psychology and focuses on the impact of different types of cooperative, competitive, and individualistic goal structures; incentive structuring, which is based on behaviorism and focuses on the use of group rewards to promote cooperation. These socialbehavioral approaches tend not to focus directly on the activities occurring during the cooperative activity itself (phase B of Figure 5-1). The second
general theoretical perspective, cognitive-developmental, focuses less on the establishment of conditions and much more on the activities taking place within the cooperative episode and of the individual learners. This two-category scheme certainly simplifies the work to date, but it is useful in presenting the findings in this chapter. The two theoretical perspectives are also not mutually exclusive; they have served primarily to guide the selection of experimental variables.
A framework for viewing these theoretical perspectives, presented in Figure 5-2, incorporates some of the terminology developed by D.W. Johnson and R.T. Johnson (1989). The top strand of this figure represents socialbehavioral perspectives. Group cohesion (positive outcome interdependence) is fostered by appropriate goals, tasks, and, under some conditions, incentives or reward structures. This cohesion leads to increased motivation to achieve individually and to help other group members achieve. This increased motivation may lead directly to positive outcomes by enhancing individual efforts or indirectly by increasing promotive (positive) interactions among group members. The bottom strand of Figure 5-2 illustrates the cognitive-developmental perspectives. In this case, the specification of information processing activities and the provision of processing supports creates explicit process interdependence (e.g., depending on others for feedback), which, in turn, increases the promotive interactions among group members. The increase in promotive interactions may lead to enhanced outcomes directly by increasing learning, indirectly by increasing general motivation due to the occurrence of successful and rewarding interactions, or both.
With a few exceptions (see D.W. Johnson and R.T. Johnson, 1987; D.W. Johnson et al., 1991a, 1991b), all the books and articles on cooperative learning tend not to draw sharp distinctions between findings based on children and adults. Because studies done with children represent the bulk of the research, the adult findings tend to be relegated to the background in examinations of the general literature. (For excellent recent reviews, analyses, and annotated bibliographies of the general cooperative learning literature, see D.W. Johnson and R.T. Johnson, 1989; Knight and Bohlmeyer, 1990; Nastasi and Clements, 1991; Slavin, 1987, 1990, 1992; Totten et al., 1991).
As noted above, evaluations of the effects of cooperative learning on performance variables require an examination of individual rather than group outcomes. The most extensive review of cooperation is reported in D.W. Johnson and R.T. Johnson (1989). In examining over 575 experimental research studies since 1898, they identified 104 studies that compared coop-
erative and individualistic eff0rts 0n individual achievement (D.W. Johnson and R.T. Johnson, 1989:Table 3.5). The mean effect size for cooperation was 0.51 with a standard deviation of 0.60. However, in an analysis of 32 cooperative learning studies involving durations of at least 4 weeks, Slavin (1990) found a median effect size of only 0.30.
Although positive effects for cooperation continue to be reported (e.g., Nastasi and Clements, 1991; O'Donnell and Dansereau, 1992), there has been a substantial number of reports of no differences (e.g., Slavin, 1990). Unfortunately, the huge number of practitioner-oriented articles about cooperative learning (see bibliography in Totten et al., 1991) tend to ignore these findings of no differences. In order to clarify this situation, many researchers have attempted to determine what elements are necessary and sufficient for effective cooperative learning.
Social-Behavioral Perspectives and Findings
Social-behavioral perspectives, which emphasize the establishment of appropriate cooperative conditions (phase Al in Figure 5-1), have dominated the cooperative learning literature. Within these perspectives, there has been controversy over which conditions or combinations of conditions are critical for productive learning. Social interdependence theory, which is derived from the early work of Lewin (1935, 1948) and Deutsch (1949, 1962), is based on the premise that the type of social interdependence, created by goal specification, determines how individuals act and interact within the situation, which, in turn, determines outcomes (D.W. Johnson and R.T. Johnson, 1989). According to this theory, positive goal interdependence results in enhanced motivation to engage in promotive interaction: participants focus on both increasing their own achievement and increasing the achievements of their groupmates. Conversely, negative goal interdependence results in oppositional interaction: participants focus on increasing their own achievement but also on discouraging and obstructing others' efforts to achieve. In the absence of goal interdependence, there is no interaction: participants work independently, focusing on increasing their own achievement while ignoring as irrelevant the efforts of others. In the context of this chapter, social interdependence theory assumes that positive cooperative efforts are based on motivation that is generated by joint aspirations to achieve a significant goal, and by interpersonal factors such as being part of a mutual effort, a joint sense of purpose and meaning, social support, and positive relationships among group members.
This focus on goal interdependence conflicts somewhat with a behaviorally oriented, incentive viewpoint, which emphasizes the importance of providing group incentives or rewards to promote cooperation. The research findings are mixed. Some earlier work (Slavin, 1983) concluded that
both cooperative task structure (i.e., goal or resource interdependence) and a cooperative (group) incentive structure (e.g., certificates to be given to groups whose members achieve a prescribed level of individual achievement) were necessary for significant gains in achievement in comparison to individual learning. There is also evidence that positive goal interdependence alone is sufficient to increase individual achievement (D.W. Johnson and R.T. Johnson, 1983, 1987; Mesch et al., 1988). More recently, Slavin (1992) found that, under some circumstances (e.g., intrinsically interesting learning tasks), group incentives may not be necessary. Other work (e.g., Mesch et al., 1986) concludes that, while positive goal interdependence is sufficient to produce higher achievement, the combination of goal and reward interdependence is more effective.
Both Slavin and the Johnsons believe that the motivation that results from positive interdependence is the primary cause of the promotive interactions that eventually lead to positive outcomes. Except for a few studies examining social skills pretraining, postcooperative group processing, and the effects of controversy (e.g., D.W. Johnson et al., 1985), the socialbehavioral theorists have not attempted to systematically examine or directly manipulate the information processing within and between cooperating group members. The general assumption apparently has been that if the cooperative conditions (e.g., goals, incentives, resources) are appropriately established, the participants will naturally engage in effective information processing. This view may be overly optimistic under some circumstances. Research with individual learners suggest that they typically do not use optimum cognitive (e.g., Dansereau, 1985) or metacognitive (Palinscar and Brown, 1989) strategies and that their learning outcomes can be improved substantially by strategy training (see O'Neil, 1978; Segal et al., 1985; and Weinstein et al., 1988, for reviews of this work).
Slavin (1990) reviewed 68 studies of cooperative learning in elementary and secondary schools; each involved durations of at least 4 weeks and compared individual achievement gains due to cooperative learning with those of control groups. Slavin reports that of 43 studies of cooperative learning methods that provided group rewards based on the sum of group members' individual learning outcomes, nearly all found positive effects on achievement. Studies of methods that used group goals based on a single group product or provided no group reward found few positive effects. This pattern of results is partially congruent with D.W. Johnson and R.T. Johnson's (1989:60) observations that studies that do not include positive interdependence through group rewards tend not to find cooperative effects on individual outcomes. However, other researchers within the social-behavioral perspective have found that positive cooperative effects can occur without group incentives. Under some circumstances, clearly specifying group goals and appropriately dividing the learning tasks to create social
cohesion appears to be sufficient to generate positive outcomes (e.g., Sharan and Shachar, 1988; Yager et al., 1985). As an attempt at resolution of differences with regard to critical elements, Slavin (1992) suggests that group rewards may be necessary for less interesting and lower-level tasks, such as fact memorization, while other methods of establishing interdependence may be sufficient with more interesting, and somewhat more complex, conceptual tasks.
Cognitive-Developmental Perspectives and Findings
In contrast to the social-behavioral researchers, cognitive and developmentally oriented researchers have made the task-oriented cooperative interactions the focus of their inquiries. For example, Webb (1980, 1982, 1985, 1989, 1992) has conducted extensive analyses on the internal dynamics of small-group learning with particular emphasis on the activities of asking for, receiving, and giving help. She has outlined the complex relationships between the success and failure of these activities and learning outcomes (for details, see Webb, 1992). Recent research by Meloth and Deering (1992) and Ross and Raphael (1990) have validated and extended Webb's findings to more structured interactions.
From a developmental perspective, the theories of Vygotsky (1978) and Piaget (1926) both have implications for the processing mechanisms underlying successful cooperative learning. Vygotsky (1978:86) suggests that effective instruction falls within a zone of proximal development: "the distance between the actual developmental level as determined by independent problem-solving and the level of potential development as determined through problem-solving under adult guidance or in collaboration with more capable peers." Since cooperating peers are likely to have some compensatory strengths and weaknesses, they would be likely to help each other, moving within their respective zones by modeling and feedback processes.
Piaget has strongly implicated confrontation and conflict as major catalysts for change and cognitive growth. Many modern Piagetians (e.g., Damon, 1984; Dimant and Bearison, 1991; Murray, 1982) have argued that cooperative interactions among students on learning tasks will lead directly to improved outcomes because, in their discussions, cognitive conflicts between students occur and are resolved and inadequate reasoning is exposed and modified.
Although there clearly needs to be some incentive for individuals to engage in cooperation, the cognitive-developmental perspective would suggest that, once engaged, the cooperative activities themselves are motivating and, to some extent, self-sustaining. Consequently, group rewards based on subsequent individual achievements are considered to be largely unnecessary.
Researchers viewing cooperative learning from cognitive-developmental perspectives focus on manipulating and measuring the actual interactions
occurring within the cooperative episode (see Figure 5-2). In general, they appear to believe that the direct manipulation of cooperative activities (e.g., by altering the make-up of the group to ensure heterogeneity of cognitive styles and by promoting intellectual controversies) should supersede the manipulation of conditions surrounding cooperation (i.e., the development of positive goal interdependence, as advocated by social-behavioral theorists).
Researchers operating within cognitive-developmental perspectives have found the following manipulations to be effective in enhancing individual achievement in cooperative learning scenarios:
• the use of explicit, instructor-provided, interaction scripts to orchestrate the cooperative activities (e.g., Fantuzzo et al., 1990; Fantuzzo et al., 1989a; D.W. Johnson and R.T. Johnson, 1979; O'Donnell and Dansereau, 1992; Smith et al., 1984; Tjosvold, 1991);
• the use of an instructor to monitor and guide the peer interactions (e.g., Brown and Palinscar, 1989; Rosenshine and Meister, 1991);
• the use of computer software support to direct and enhance the cooperative activities (e.g., Hythecker et al., 1985; Swallow et al., 1988);
• the use of pretraining and post-cooperation group processing to improve critical interaction skills and patterns (e.g., Mesch et al., 1988).
These direct methods of scripting, teacher and computer guidance, and training, as well as in-depth examinations of process (e.g., Webb, 1992), have received substantially less attention than have the indirect methods of social structure and reward.
Given that methods developed from both the social-behavioral and cognitive-developmental perspectives appear to have the potential for promoting cooperative effects under some conditions, it is tempting to conclude that combining elements of all of these approaches would produce the ultimate cooperative scenario. Obviously, however, the development of any combinatorial approach must proceed with caution. Not only may some of the components be in conflict with one another, but even a set of synergistic components can quickly overload participants during cooperative learning (see Dansereau, 1988). Rather than building the ultimate cooperative scenario, a more reasonable goal, following the suggestions of Slavin (1992), would be to examine which elements are necessary and sufficient in particular learning contexts.
Overall, the research on cooperative learning provides sufficient evidence to establish the efficacy of this approach for enhancing individual achievement under some conditions. This literature also provides some initial ideas about
the mechanisms underlying successful cooperation. What is now needed is more systematic examinations of the processes by which cooperation leads to achievement, development of theories to reflect these processes and mechanisms as they operate in specific educational environments, and development and evaluation of new cooperative learning methods that incorporate these theoretical advances (see Knight and Bohlmeyer, 1990).
Limitations of the Research
Although some limitations of the research and development efforts are implicit in the previous sections, in this section we make them more explicit. The identified limitations are divided into three categories: quality and precision of the research, scope and neglected issues, and context dependency.
Quality and Precision
As with many domains, the quality of cooperative learning research is highly variable. The laboratory studies in this area are typically in the tradition of experimental social psychology, and, consequently, have tended to use artificial tasks. Although these types of experiments can be useful for theory construction, they have often been inappropriately used to establish the validity of cooperative learning. But the field studies of cooperative learning have frequently not been well controlled (e.g., nonrandom assignments to treatments, uncontrolled "teacher" and treatment effects). With few exceptions (e.g., Webb, 1989; D.W. Johnson and R.T. Johnson, 1989; King, 1990; Meloth and Deering, 1992; Ross and Raphael, 1990), there has been virtually no examination of the relationship between the characteristics of cooperative processing and subsequent outcomes, either through direct observation or diagnostic questionnaires and tests. Generally, cooperative manipulations are not carefully controlled, making the interpretation of findings difficult.
Most of the dependent measures used in cooperative experiments are omnibus, nondiagnostic measures of learning, rather than the more specialized sets of measures typically used in individual learning and cognition experiments (e.g., recognition versus recall). Consequently, it is not clear what kinds of information are being learned or not learned during cooperation. Furthermore, the lack of focus on individual differences in cooperative learning has limited the use of more powerful statistical methods for finding between-group differences and has precluded the examination of interaction effects. In general, as cooperative learning research moves beyond the effect demonstration stage, it behooves researchers to incorporate more of the precise methodologies used in research on individual learning and cognition in their experiments.
Another major problem with research and development in cooperative learning has been the lack of communication among researchers. There have been numerous conceptually overlapping cooperative techniques and approaches developed independently by individuals from diverse backgrounds. Part of the difficulty has stemmed from the variety of journals in which researchers in this field publish. Only very recently have integrated reviews (e.g., D.W. Johnson and R.T. Johnson, 1989; Nastasi and Clements, 1991; Slavin, 1992) begun to bring this field together.
Scope and Neglected Issues
As Slavin (1992) indicates, a large number of the studies often included under the rubric of cooperative learning compare team or group performance directly to individual performance. Not surprisingly, there is a clearcut advantage for groups in these cases (see D.W. Johnson and R.T. Johnson, 1989). However, most researchers would agree that one of the primary criteria for effective cooperative learning is the facilitation of individual achievement and performance. In this regard, there are far fewer studies that have successfully demonstrated advantages for cooperative versus individual learning. It is clear that many additional, well-controlled studies are needed to understand the parameters and boundary conditions relevant to cooperative enhancement of individual outcomes.
We believe four issues deserve particular attention. First, researchers should examine methods for directly promoting and supporting effective interactions between cooperating individuals. Such methods include pretraining; guidance during cooperation with scripts, instructors, and computers; the use of visual aids and worksheets; and the use of postcooperation group processing. These approaches stem from the cognitive-developmental perspective, which has been underrepresented in the cooperative learning literature.
Second, assessment and diagnosis of cooperative failures and potentially negative cooperative outcomes (e.g., overdependence on social support for task performance) are needed. Although a number of detrimental effects arising from cooperation have been identifiedthe "free rider," the "sucker," the ''status differential," and "ganging up" effects (see e.g., Salomon and Globerson, 1989)there has been little formal experimentation on what promotes these effects and how they can be ameliorated. Furthermore, there appears to be no research on potential long-term negative effects of cooperation, such as dependency on social support for effective learning.
Third, researchers need to examine skill transfer. In addition to learners' direct acquisition of content-related knowledge and skills, there is some limited evidence that, under certain conditions, cooperative experiences can indirectly foster learning, thinking, and communication skills that can be transferred to other group and individual tasks (see O'Donnell and Dansereau,
1992). Further research is needed to identify principles governing this type of transfer.
Fourth, there has been very little work on the relationship between individual difference variables and cooperative performance indices and subsequent outcomes. The research that has been conducted (e.g., Hall et al., 1988; Wiegmann et al., 1992) is likely to be of limited value due to the idiosyncratic nature of the cooperative techniques and dependent measures used.
Although some cooperative learning techniques have been specialized for specific content domains (e.g., small-group learning and teaching in mathematics, Davidson, 1990; team-assisted individualization, Slavin, 1985), most of the highly publicized techniques (circles of learning, Johnson and Johnson, 1975; student teams-achievement divisions, Slavin, 1980) are considered to be general purpose ones, useful in a large number of content areas, with a variety of different types of learners. Furthermore, the underlying principles guiding the development and implementation of cooperative scenarios (e.g., goal, task, and reward interdependence) are often described as being context-independent. As a result, cooperative learning approaches are being implemented in a wide range of instructional settings with only minimal evaluation and tailoring. Given the limited arenas in which they have been developed, primarily grades 2-9, it seems likely that these approaches and the principles that they are built on will not always match specific contextual constraints and learner characteristics. Unfortunately, at present these mismatches may be generally undetected because of "placebo" and "bandwagon" effects.
More specifically, techniques developed primarily for young children are typically applied with little modification to college and adult instructional contexts without systematic evaluation and experimentation (see Cooper and Mueck, 1990; Cooper et al., 1990). If nothing else, the dramatic differences between children and adults in cognitive and social developmental stages warrant a close look at the general applicability of cooperative principles across these age groups.
ADULT COOPERATIVE LEARNING
Although the research and development work with children should inform the study of adult cooperative learning, there are three strong ageand situation-related differences that need to be considered. First is context. In comparison with grade school and middle school, adult courses in colleges and technical settings are usually shorter duration and faster paced, leaving much less instructional time for cooperative activities. Conse-
quently, adult cooperative episodes typically need to be highly focused and intense.
The second set of differences concern materials and tasks. Adult materials (particularly in college courses) are usually at a high level of complexity, and typically require a greater emphasis on comprehension than memorization. In technical training and laboratory courses, the focus is often on learning to perform difficult procedures. Because of the difficulty and complexity of the tasks and materials, adult cooperation may require joint effort and outside support for all learning phases, rather than divided efforts on portions of the material as occurs with some of the techniques used with children.
The third difference is learner characteristics. Adults are typically at higher cognitive developmental stages than children and are more socially sophisticated and skilled. These characteristics may reduce the need for token group rewards (especially token rewards) to promote productive cooperation among adults. Most adults can see the intrinsic cognitive and social value of engaging in cooperative learning.
In addition, adults are often defensive about their learning, thinking, and communication skills, and frequently have well-established, though often inadequate, strategies for dealing with complex materials and tasks (Dansereau, 1985). These cognitive and communication characteristics and deficits suggest that adult cooperative episodes may need to be carefully scripted in order to encourage participants to engage in the type of cognitive activities necessary for enhanced learning.
In general, this analysis of differences between adults and children suggests that the cognitive-developmental and social interdependence perspectives may be more potent than the social-behavioral, incentive-structuring perspective in examining and structuring adult learning. In this next section we examine the research on adult cooperative learning with these issues and perspectives in mind. We consider general findings, nature of adult cooperative techniques, role of communication aids and cooperative support, transfer to new learning tasks, and individual differences.
The research on adult cooperative learning can be conveniently divided into two categories: implementation studies in ecologically valid settings and controlled laboratory studies. In general, the effect sizes tend to be in approximately the same range as found in the research with children (0.30 to 0.50; see, e.g., D.W. Johnson and R.T. Johnson, 1989).
Implementation studies in adult courses have usually compared cooperative learning with individual learning on typical achievement measures. The reported results indicate positive effects for cooperation over a variety
of topics and courses, including: abnormal psychology (Fantuzzo et al., 1989a, 1989b), engineering (Smith et al., 1984, 1986), social psychology (Fraser et al., 1977), multicultural education (Jacobs and Icola, 1990), chemistry (Marks, 1991), statistics (Bansangue, 1991), physical education (R.T. Johnson et al., 1983), introductory physics (Heller et al., 1992), methods of teaching music (Hwong et al., 1993), military history in college reserve officers training corps (ROTC) (D.W. Johnson et al., 1991), educational methods (King, 1990), preparation for nursing boards (Frierson, 1986), the charts and publication unit from an air-traffic control course (Holubec et al., 1993; Vasquez et al., 1993), Army communications electronics operating instructions (Shlechter, 1988), Army equipment records and parts specialist training (Brooks et al., 1987; Hagman and Hayes, 1986), and an Air Force communication center specialist course (Hungerland et al., 1976).
But a number of adult implementation studies have failed to find differences on achievement measures between cooperative learning and control groups (e.g., Carpenter, 1986; Lewis, 1991; Palmer and Johnson, 1989; Sherman, 1986). Since most of the significant and nonsignificant implementation evaluations are published in journals that do not require detailed reporting of methods and results, it is very difficult to assess the validity of these studies and to determine a pattern of differences between successful and unsuccessful implementations. Although the weight of the evidence suggests that cooperative learning can be effective in adult settings, the few published findings of no differences and the fact that such findings are generally underreported suggest that positive conclusions be drawn cautiously.
There have been relatively few controlled laboratory studies of the effects of cooperative learning on individual achievement with adult subjects. The most systematic program of laboratory research in this area has focused on the examination of scripted cooperation with pairs of college students (e.g., Hall et al., 1988; Hythecker et al., 1988; Larson et al. 1985b, 1986; Larson and Dansereau, 1986; O'Donnell and Dansereau, 1992). Typically, these student dyads are instructed to follow a cognitively based interaction script in the studying of excerpts from introductory science textbooks or in the learning of concrete procedures (e.g., administration of intravenous therapy). Figure 5-3 shows versions of two sample scripts. The stated learning goal is for the students to help each other acquire the presented information; no group rewards are given. Performances on tests taken individually over the material are compared with individual study control groups and with other scripted and unscripted cooperative groups. On some occasions, transfer to subsequent individual study tasks is assessed. The specific objective of this research has been to determine the critical elements in cognitively based scripts. In this regard, the results of these laboratory studies have indicated that certain types of explicit cooperative scripts lead to better performance than unscripted cooperation and individual study (see below).
As noted above, the primary goal of cooperative techniques is to increase task motivation or promotive interactions between group members in order to enhance learning. From the social-behavioral perspective, the goal is typically accomplished by promoting outcome interdependence through extrinsic group rewards contingent on the performance of the individual members. From the cognitive-developmental perspective, this enhanced learning is produced by promoting process interdependence through the provision of a cognitively based script for the members to follow. Of these methods, the use of group rewards has generated the greatest amount of controversy. With children involved in relatively low-level learning (as defined by Bloom's 1956 taxonomy), group rewards appear to be important in maximizing the effects of cooperative learning (Slavin, 1992). Even in
this case, however, Kohn (1991) has warned against the potential long-term negative effects on intrinsic motivation of such an incentive system. However, there is some evidence that, with higher-order learning (e.g., Sharan, 1990) and with scripted and supported cooperation (e.g., Brown and Palinscar, 1989), group reward may be unnecessary even for children.
In the adult literature, the importance of explicit outcome interdependence promoted by group incentives is diminished even further. Numerous implementation studies (Fantuzzo et al., 1989a, 1989b; King, 1990; Riggio et al., 1991) and laboratory studies (Dansereau, 1985, 1987, 1988) have shown cooperative effects without the use of group rewards. These findings are echoed strongly by James Cooper (editor of the Cooperative Learning and College Teaching newsletter) and his colleagues (Cooper et al., 1990:13):
At the collegiate level, participation in [the] cooperative learning groups appears to be sufficiently self-motivating that extrinsic reinforcers such as grades may not be critical to motivate students in groups. The group work is intrinsically reinforcing as long as criterion-referenced grading and individual accountability are components of the cooperative learning system and the tasks performed in the groups are perceived as meaningful to the students, not just busy work ....
This conclusion concerning the relative unimportance of group reward presumably would be true of adult learning in military and industrial settings as well. However, there is a notable exception: two well-controlled experiments by Hagman and Hayes (1986) with Army trainees showed that cooperative learning promoted significant gains in individual achievement only when a group reward was used and that the effect of the reward was to increase within-group communication. One difference between these two experiments and those that showed cooperative effects without rewards is in the degree of interactive scripting. The studies without rewards provided strong guidance as to the roles and activities of the participants; in the Hagman and Hayes experiments, no guidance was provided. Hagman and Hayes (1986:20) acknowledged and elaborated on this issue:
Although the present experiment used group rewards to encourage withingroup communication, any cooperative procedure that ensures meaningful communication among group members should also promote individual achievement. Thus, group reward may not be necessary when communication is brought about by other means. Recent research supports this notion (Dansereau, 1983; Yager et al., 1985). Dansereau (1983) for example, has shown that structuring interaction within cooperative groups by giving members specific assignments to orally summarize and elaborate upon to-be-learned materials can effectively promote individual achievement in the absence of group reward. The present research suggests that if group interaction is left unstructured, then group reward can be used to encourage the interaction among group members necessary for promoting individual achievement gains when trainees work cooperatively.
We note, however, that underlying this conclusion is the assumption that unscripted group members know beforehand how to best learn the material and how to help one another and that the major issue is motivating the participants to communicate. In the Hagman and Hayes (1986) experiments, the learning material was highly structured information on maintaining a prescribed load list (a portion of the Army's equipment records and parts specialist training sequence). This material was relatively straightforward and primarily required memorization; consequently, it is likely that participants knew effective methods for learning it. With more complicated material that requires substantial comprehension and organization skills, however, it is less likely that participants will have optimal learning and communication strategies in their existing repertoire (Dansereau, 1985, 1988). Under these conditions, merely increasing motivation by providing group incentives may not be sufficient to enhance performance; the provision of a coherent cooperative script is likely to be much more effective.
Although it may be tempting to combine cooperative group rewards (based on achievement scores) with cooperative scripts, recent research by O'Donnell (1992) suggests these two components may not be synergistic. Explicit incentives appear to cause the participants to deviate from the script and to engage in their typical, often nonproductive, learning behaviors, such as rote rehearsal (see also King, 1990; and Pressley et al., 1988). One promising combination that does not seem to have been formally examined is the use of process rewards in conjunction with explicit scripts: that is, group members are rewarded on the basis of effective implementation of the cooperative activities rather than their outcome scores.
The interactions among cooperative group members can be scripted indirectly by dividing up the task or materials and by pretraining on social skills or, it can be done directly by providing explicit scripts that prescribe specific cognitive or social roles and activities. Task division approaches to influencing interactions and interdependence require students to master different parts of the material and teach it to the other members of their groups. These approaches are likely to be effective with material that is easy to understand and organize, but they appear to be relatively ineffective with larger bodies of complex information for which collaboration over the entire set of material may be important (see Lambiotte et al., 1987; and Palmer and Johnson, 1989).
Some of the most promising research with adult learners encourages all members of the group to focus on all aspects of the material with explicit scripting, including structured controversy (D.W. Johnson et al., 1986); reciprocal peer tutoring (Fantuzzo et al., 1989a); reciprocal questioning (King, 1990); and scripted cooperation (O'Donnell and Dansereau, 1992). These methods presumably enhance the possibility that cooperating individuals will share perspectives, insights, and elaborations and will challenge and
correct each others' misconceptions. Engaging in cooperative information processing is also believed to increase an individual's motivation and activity level. Two of the most thoroughly researched approaches to scripting are scripted cooperation and structured controversy.
O'Donnell and Dansereau (1992) have articulated a general conceptual framework for examining and structuring cooperative interactions. This framework is based loosely on Webb's analyses of children's cooperative interactions (e.g., Webb, 1989, 1992), King's analysis of college students' interactions (e.g., King, 1990), and a series of experiments on manipulating cognitive scripts (e.g., Lambiotte et al., 1987; Larson et al., 1985b; McDonald et al., 1985; Wiegmann et al., 1992). In this framework, group interactions are considered to involve complex combinations of learners' cognitive/motor, affective, metacognitive, and social activities, referred to as CAMS. Cognitive/motor activities include comprehension, elaboration, organization, retrieval, and skilled performance. Affective activities include motivation, anxiety, and concentration. Metacognitive activities include monitoring of comprehension and performance, detection and correction of errors, and awareness of performance levels. Social activities include awareness of and effective communication with others in the cooperative situation. In this framework the effectiveness of a cooperative learning situation is believed to depend on the combination of CAMS activities of participating members: an overemphasis on any one activity may disrupt the synergistic balance and inhibit performance. If, for example, a participant places too much emphasis on the metacognitive system, the participant's ability to generate information may be impaired.
The CAMS framework has been used to guide the design of dyadic cooperative scripts. One example is a simple text-processing script, cooperative murder (Dansereau, 1985); mobilize, understand, recall, detect, elaborate and review. Participants first mobilize their resources for learning by establishing an appropriate mood and by surveying the text to establish cooperative action points (asterisks in the margin to indicate where they will stop reading and engage in cooperative information processing). Both partners then read for understanding until they reach the first action point. One partner then (recalls or recites) what has been learned to that point while the other partner detects and corrects errors and omissions. Both partners then collaboratively elaborate on the material by forming images, analogies, and direct connections to other information. They then continue reading for understanding until they reach the next action point where they reverse roles and repeat the recall, detect, and elaboration steps. The partners proceed through the material, alternating roles until they have completed the assignment. They then coop-
eratively review and organize the entire body of information, once again alternating presentation and monitoring roles.
This script, which has been generalized to writing and concrete performance tasks, was designed to facilitate a number of potentially effective activities that have emerged from research findings in cognitive and educational psychology, such as oral summarization during the recall and recite stage (Ross and DiVesta, 1976; Yager et al., 1985), metacognitive activities during the detect and correct stage (Baker and Brown, 1984), elaborative activities during the elaboration stage (Reder, 1980), cross-modeling and imitation of personal strategies and perspectives throughout the stages (Bandura, 1971), and the use of multiple passes through the material (Dansereau, 1985).
Research with college students enacting versions of the murder script without group rewards have shown that it leads to better individual outcomes than unscripted cooperative scenarios in which the participants determine their own cooperative activities. This outcome is not surprising given the adult learner characteristics described above. Since adults often have not developed optimum strategies for handling complex information and many are defensive about their own thinking, learning, and communication skills, they are likely to gravitate toward familiar and comfortable roles and activities that minimize productive interactions with their learning partners.
A natural occurrence in cooperative learning groups is for members to disagree and argue with each other. Such intellectual conflicts can have powerful influences on learning when they are managed constructively. D.W. Johnson and R.T. Johnson (1979) described such conflicts as controversies and proposed both a theoretical explanation of why they are so powerful and an operational procedure for instructors to use in structuring or scripting controversies among students.
Controversy exists when one student's ideas, information, conclusions, theories, and opinions are incompatible with those of another, and the two seek to reach an agreement. Controversies are an inherent aspect of virtually all decision making, problem solving, reasoned judgment, and critical thinking. If students get intellectually and emotionally involved in cooperative efforts, controversies will inevitably occur. In a structured controversy, students make an initial judgment, present their conclusions to other group members, are challenged with opposing views, become uncertain about the correctness of their views, actively search for new information and understanding, incorporate others' perspective and reasoning into their thinking, and reach a new set of conclusions.
The process through which controversy works can be described in more detail. When individuals are presented with a problem or decision, they
have an initial conclusion based on categorizing and organizing incomplete information, their limited experiences, and their specific perspective. When individuals present their conclusions and rationales to others, they engage in cognitive rehearsal, deepen their understanding of their position, and use higher-level reasoning strategies. The presenters are then confronted by other people with different conclusions based on other people's information, experiences, and perspectives. They become uncertain as to the correctness of their views. A state of conceptual conflict or disequilibrium is aroused.
Uncertainty, conceptual conflict, and disequilibrium motivate an active search for more information, new experiences, and a more adequate cognitive perspective and reasoning process in hopes of resolving the uncertainty. This active search is called epistemic curiosity (Berlyne, 1960). Divergent attention and thought are stimulated. By adapting their cognitive perspective and reasoning through understanding and accommodating the perspective and reasoning of others, individuals develop a new, reconceptualized, and reorganized conclusion. Novel solutions and decisions are detected. Individuals working alone in competitive and individualistic situations obviously do not have the opportunity for such a process.
Using this process, D.W. Johnson and R.T. Johnson (1992a) have developed a procedure for structuring experimental controversies in cooperative learning:
1. Students are assigned to a cooperative learning group of four, divided into two pairs. Each pair is assigned the task of researching and preparing the best case possible for either the pro or con position on the issue being considered. In an engineering class, for example, pairs may be assigned two different ways of disposing of hazardous waste or the need to be concerned about acid rain in their design of power plants.
2. Each side presents its position to the other. The pair's task is to present the best case possible for the position they have been assigned with the goal of persuading the opposing pair to agree with them. This task requires students to rehearse orally the relevant information, advocate a position and point of view, and teach their knowledge to peers.
3. Both sides participate in an open discussion characterized by refutation and rebuttal. Students are given two major tasks. The first is to analyze, critically evaluate, and then refute the rationale underlying the opposing position. To do so, students must reason both deductively and inductively. Intensive criticism of ideas and logic is encouraged. The second task is to rebut the attacks made on their position and its rationale by the opposing pair.
4. Each side engages in a perspective-reversal in which they present the perspective, position, and rationale of the opposing pair. Students are
to ensure that they completely understand the opposing position and its rationale.
5. All advocacy is ended, students view the issue from both perspectives simultaneously, and a common position is agreed on through synthesis and integration. To arrive at their best reasoned judgment about the issue, students synthesize and integrate information into factual and judgmental conclusions that are summarized in a joint position that both sides support.
Research on structured controversy with adults has been conducted primarily in engineering schools (Smith et al., 1984, 1986) and in business settings (Tjosvold, 1991). The results of this research indicate that controversy, compared with concurrence-seeking, debate, and individualistic efforts results in greater achievement and retention (Smith et al., 1981), higher quality problem-solving (Maier and Hoffman, 1964; Nemeth and Wachtler, 1983; Tjosvold, in press), greater creativity (Hall and Williams, 1966, 1970; Torrance, 1970, 1971, 1973), and more accurate exchange of expertise (Lowry and Johnson, 1981).
Other Factors: Communication Supports, Transfer, and Individual Differences
Many cooperative learning implementers have recommended and anecdotally supported the use of computer software (Broome and Chen, 1992; Hythecker et al., 1985; Rocklin et al., 1985), and expert instructors (McDonnell, 1990) to guide and monitor adult cooperative learning episodes. Although these supports may have initial value in getting the cooperative interactions on track, it is not clear that they offer any advantages over a detailed script for most adult tasks. Furthermore, since they usually require additional resources, extensive use of such supports may not be cost-effective in many settings. However, in the case of cooperation between individuals who are not in physical proximity, computers may provide a useful communication vehicle. Computer-mediated approaches have been used effectively in collaborative writing (see Duin, 1991 for a review), and, more generally, in education (e.g., O'Malley and Scanlon, 1990).
With regard to communication aids, a few studies have indicated that providing cooperative groups with spatial-verbal displays of the to-be-learned informationsuch as diagrams and flowchartsrather than purely verbal textual presentations leads to better individual outcomes (e.g., Patterson et al., 1992, 1993; Rewey et al., 1992). Presumably the spatial and graphic signaling provides for easier reference during group discussion. On the basis of these findings, it would be expected that any visual aid that enhances the availability and accessibility of information during cooperative interactions could lead to more productive cooperative information processing and thus more positive individual outcomes.
There is evidence that adults given scripted cooperative experiences perform better than control subjects on subsequent individual learning tasks (see Dansereau, 1987; and O'Donnell and Dansereau, 1992). Apparently, individuals acquire content-independent knowledge and skills from a cooperative episode that can be used to improve individual learning of new information. This type of content-independent transfer seems to occur only when the cooperative interactions are well scripted (McDonald et al., 1985), and especially when the scripting emphasizes cognitive elaboration activities (Larson et al., 1985b). In the studies cited above, content-independent transfer was an incidental outcome from the cooperative experience. During cooperative learning, the participants in these studies were directed to learn a particular body of information; they were not informed about the subsequent individual transfer task. In essence, they appeared to acquire transferrable knowledge and skills as a by-product of engaging in a taskoriented cooperative episode. The implementation of new learning strategies with a detailed script and the opportunity for cross-modeling between partners may be responsible for this incidental acquisition.
There has been very little research examining the effects of individual differences on adult cooperative learning outcomes. In one study, it was found that college students with relatively high scores on a measure of social orientation gain more from cooperative interactions than from individual study sessions, but the reverse is true for students with low scores on that measure (Hall et al., 1988). In terms of matching cooperative partners, Larson et al. (1984, 1985a) looked at the effects of verbal skills and field dependence/independence (the skill of separating target information from a complex background) on gains achieved in adult cooperative learning dyads and found that students with lower verbal skills performed best when paired with students with higher verbal skills. Importantly, the students with higher verbal skills were not adversely affected by these pairings. Similar patterns of results were found for the field dependence/ independence variable. These results are congruent with the recommendations of heterogeneous groupings from various cooperative learning researchers (e.g., D.W. Johnson et al., 1991b).
Conclusions and Future Directions
Although there has been substantially less cooperative learning research conducted with adults than with children, there is sufficient information to draw a few tentative conclusions:
• Adult cooperative learning can be more effective than individual learning across a variety of topics and tasks. However, because of the complexity of adult learning tasks, explicit cooperative scripts that encourage promotive interactions may be necessary to reap the benefits of cooperation,
at least with participants who lack previous experience with cooperative learning.
• Because most adults are able to see the potential value of collaboration, group rewards appear to be generally unnecessary and may even hinder following a script.
• Cooperative supports, such as worksheets, computers, and instructors, may be useful in the early stages of cooperation and with potentially difficult materials.
• Information displays (e.g., maps, charts, diagrams, and pictures) that increase the availability and accessibility of material during cooperative processing appear to enhance individual outcomes.
• Transferrable individual learning skills can be acquired as a byproduct of well-scripted cooperative episodes.
These conclusions about adult cooperation are attenuated by many of the same factors discussed above as limitations of the general field of cooperative learning. Certain components of cooperative scenarios (see Figure 5-1) have been underresearched. In particular, very little is known about the nature of adult cooperative interactions. Analyses of the type conducted by Webb on children (e.g., Webb, 1992) are needed at the adult level in order to provide a foundation for further theoretical developments. At the same time, the more precise experimental methodologies used in the areas of individual cognition and learning need to be modified and implemented in the study of cooperative learning and cognition. This approach may, in return, also facilitate the understanding of individual processing. Cooperative learning episodes tend to make public the participants' thinking in a relatively naturalistic way, thus providing processing data that is less artificial than that typically collected through asking individuals to describe their thinking. It is also possible that certain cognitive processes may be more highly activated during socially driven cognition than during individual cognition: if so, these processes may be more available for examination during cooperation.
Group processing to improve subsequent cooperative episodes does not seem to have received experimental attention in the adult literature (see phase C in Figure 5-1). Such research would be useful not only in guiding further implementation, but in allowing participants to elaborate on their thinking during the cooperative episode. This research would provide converging information on expert, Webb-like analyses of the interactions. Expert and participant analyses of cooperative episodes could also be used as the basis for assigning group rewards. Such work would permit a comparison of the effectiveness of process-based and outcome-based rewards.
There is also a need for additional studies on the role of individual
differences in adult cooperative learning and on the potential of cooperative learning as a training ground for individual learning and thinking skills.
Approaches to implementing cooperative learning can be placed on a continuum from conceptual applications to direct applications. In conceptual applications, instructors are taught a general conceptual model of cooperative learning, which they use to tailor cooperative learning specifically for their circumstances and trainees. Direct applications are packaged lessons, curricula, and scripts that are used in a prescribed manner.
There are a number of advantages of both approaches. The direct approach requires very little instructor training and, therefore, is quick and cheap. Initial instructor success rates in using cooperative learning are high. In a way, direct approaches tend to train instructors to be technicians who use the cooperative learning curriculum or strategy without a theoretical understanding of how it works. In contrast, the conceptual approach trains instructors to be engineers who adapt cooperative techniques to their specific circumstances, students, and needs. The specific planning and adaptation required of instructors by the conceptual approach may promote more personal commitment to and ownership of cooperative learning than the direct approach (D.W. Johnson and R.T. Johnson, 1985b, 1992b).
Training may proceed best when both the conceptual and direct approaches are used. Simply presenting a theoretical framework makes it too difficult for most instructors to create their own cooperative learning lessons. Simply teaching instructors how to use a lock-step cooperative procedure that they do not understand, however, allows them to use it immediately but leaves them without the conceptual tools required to adapt it to their students and situation and without the skills to solve implementation problems. A carefully crafted training program may require both a clear conceptual understanding of the nature of cooperative learning, its essential elements, and the instructor's role, and concrete examples of scripts, structures, and lessons.
There are two recent guidebooks for implementing cooperative learning in college settings. A 51-page book by Cooper et al. (1990) discusses principles that can be used to guide implementation and provides examples of a few cooperative learning exercises. A comprehensive report by D.W. Johnson et al. (1991b) describes underlying principles, supporting research, and an extensive collection of specific guidelines for implementing cooperative learning in college courses.
One of the most thoroughly developed approaches presented by D.W. Johnson et al. (1991b) is the use of formal, informal, and base cooperative groups in the context of an ongoing course. In formal cooperative learning
groups, participants work together, for one class period to several weeks, to achieve shared learning goals and complete specific tasks and assignments. Informal cooperative learning consists of having students work together to achieve a joint learning goal in temporary, ad hoc groups that last from a few minutes to one class period. Cooperative base groups are long-term, heterogeneous cooperative learning groups with stable membership that last for a semester, a year, or even longer.
According to these authors, formal cooperative learning groups can be used for almost any learning assignment. In formal cooperative learning groups, teachers (1) specify the objectives for the lesson; (2) make a number of preinstructional decisions (e.g., size of groups, assignment to groups, materials needed, room arrangement); (3) explain the task and the positive interdependence; (4) monitor the interaction among learners and intervene to provide task assistance, assistance in using an assigned script, or assistance in using social skills appropriately; and (5) evaluate trainees' learning and ensure that group members process how effectively they are working together.
However, there are times when instructors have to make an extended presentation of factual information in an organized and logically sequenced way. In such situations, informal cooperative learning groups can be used to ensure that learners actively process the information being presented. In such informal cooperative learning groups, learners engage in (1) an introductory focused discussion aimed at promoting advance organizing of what the trainees know about the topic to be presented and establishing expectations about what the lecture will cover; (2) a discussion in pairs every 10-15 minutes to summarize what has just been presented; and (3) a focused discussion at the end of the lecture to provide an overall summary of its content and intellectual closure to the lecture session. There are many important advantages to having students discuss what they are learning with a partner before, during, and after a lecture, film, or demonstration. In the traditional lecture and whole-class discussion, most students stay uninvolved (Barnes, 1980), a very small minority of students tends to dominate (Karp and Yoels, 1987), and many students are inhibited from participating (Stones, 1970). Informal cooperative learning groups can positively alter this pattern.
Cooperative base groups are long-term settings that provide a permanent set of relationships with other trainees. The purposes of the base group are to give the support, help, encouragement, and assistance each learner needs to progress successfully through the training program. Base groups may meet at the beginning and ending of each day (or at least twice a week) to discuss current assignments and the progress of each member and to plan how members can give each other assistance and encouragement. Base groups build relationships among learners to motivate them academically and ensure successful completion of the training program. In
this regard, it is important to note that the major reason given for dropping out of training programs and colleges is the failure to establish a social network of friends and classmates. Tinto (1975, 1987) concluded that the social networking processessocial involvement, integration, and bonding with classmatesare strongly related to higher rates of student retention. Astin (1985), on the basis of 10 years of research on colleges, concluded that involvement with peers and instructors was the ''cornerstone" of persistence and achievement, especially for "withdrawal-prone" participants (such as disadvantaged minorities) who are generally passive in academic settings. Cooperative learning experiences have been found to lead to lower rates of attrition and higher achievement (Wales and Stager, 1978), especially for black students majoring in math and science (Treisman, 1985).
The formal, informal, and base-group approach developed by D.W. Johnson et al. (1991b) has been used primarily in college settings. The needs and structures of technical training courses may require a somewhat different approach. Along this line, Brooks (1987) has developed an instructor's guide for implementing cooperative learning in an Army Quartermaster course. This document is based on the reward-driven cooperative strategy investigated by Hagman and Hayes (1986; discussed above). Although often specialized, some aspects of this guide are generally applicable to implementation of cooperative learning in technical training settings.
Although the various implementation guides provide useful general approaches, it is important to keep in mind that they are based on very little adult research. Most of the recommendations are derived directly from techniques developed for children; development of more precise guidance for cooperative learning awaits further research.
Cooperative learning typically does not require more instructional resources than those provided in most academic and technical courses. In fact, except for cooperative training materials (which are typically only one or two pages in length), cooperative approaches may actually require less resources than individual learning approaches since cooperative group members can often share computers and laboratory materials. However, greater classroom space may be needed to accommodate increased noise levels during cooperative activity.
Start-up costs include instructor and student training. This training typically requires no more than a one-day workshop for instructors and approximately two class periods for students. Of course, these times are dependent on the cooperative technique being taught and on the nature of the implementation setting.
In comparison with other emerging instructional technologiessuch as computer tutoring systemscooperative learning is extremely low cost. This low cost, in conjunction with its apparent effectiveness, makes it very attractive as a potential instructional alternative. The fact that it can be used
in conjunction with a variety of educational approaches further enhances its attractiveness.
1 It is important to note that one could ask the symmetric question: What factors control the effectiveness of individual learning that make it superior to some central condition of cooperative learning? This is not the focus of this chapter, but it is surely a topic that deserves attention.