The Effect of Collaborative Learning on Enhancing Student Achievement

A Meta-Analysis

Ravinder Kumar

A Thesis

in

The Department

of

Education

Presented in Partial Fulfilment of the Requirements

for the Degree of Master of Arts (Educational Technology) at

Concordia University

Montreal, Quebec, Canada

April 2017

© Ravinder Kumar, 2017

 CONCORDIA UNIVERSITY
School of Graduate Studies

This is to certify that the thesis prepared

By: Ravinder Kumar

Entitled: The Effect of Collaborative Learning on Enhancing Student Achievement

A Meta- Analysis

and submitted in partial fulfillment of the requirements for the degree of

Master of Arts (Educational Technology)
complies with the regulations of the University and meets the accepted standards with respect to
originality and quality.

Signed by the final Examining Committee:

_______________________________ Chair
Chair’s name/ Dr. Saul Carliner

_______________________________ Examiner
Examiner’s name / Dr. Robert M. Bernard

_______________________________ Examiner
Examiner’s name / Dr. Claude Martel

_______________________________ Supervisor
Supervisor’s name / Dr. Richard F. Schmid

Approved by ___________________________________________________
Chair of Department or Graduate Program Director

_______________ _______________________________________________
Date Dean of Faculty
 Abstract

The Effect of Collaborative Learning on Enhancing Student Achievement: A Meta-Analysis

Ravinder Kumar, M.A.

Concordia University, 2017

The purpose of this meta-analysis was to investigate the effect of collaborative learning

on student achievement. The sample of the study consisted of 20 representative studies involving

2434 participants selected from an extensive literature search based on the use of collaborative

activities in a formal education setting cutting across multiple grade levels and subject domains.

Analysis of representative studies (k = 28) yielded a moderately weighted average effect size of

0.26. A mixed effects model was used for the analysis of the moderators of effect size. The

research design was not significant across true and quasi- designs. Two groups – high

collaboration (Experimental condition) and no or less collaboration (control condition) – were

compared to measure the effect of collaborative learning on the dependent variable of student

achievement. The analyses of moderator variables were not significant or suffered from a lack of

statistical power (i.e., grade levels). Implications for the use of the collaborative learning are

discussed, and recommendations for future researchers are suggested along with the limitation

and conclusions.

Keywords: collaborative learning, meta-analysis, student achievement

iii
 Acknowledgements

As a graduate student, I experienced many memorable learning moments during the entire

program. The thesis option provided me with a real task to synthesize my experiences as a

capstone of all the learning. The journey was rigorous and challenging. However, the continuous

support and guidance provided by some of the wonderful individuals in the department helped

me complete this task successfully. I want to extend my immeasurable appreciation and deepest

gratitude for the support offered by the following persons :

Foremost, Dr. Richard Schmid, my supervisor, chair, and professor, for his full support from

the day one. He showed confidence in my abilities and helped me through his continuous

advices, suggestions, feedback, comments, and support during the wrting process of the entire

thesis. In addition, I am indebted to him for his official support to complete the departmental

formalities for my thesis defense and submission. His commitment, patience, and sincerity were

manifested during every meeting and/or communication with him which boosted my morale to

stay in tune at every stage of this work. In addition, I thank him again for serving as my referee

for various research grant projects and mentorship.

Dr. Robert Bernard, my professor and research committee member, and a specialist in meta-

analysis, for his guidance during every stage of this project for all his advices on technical

matters. I sincerely thank him for allowing to use the dataset for my study sample. In addition, he

always showed his willingness to help me out whenever it was needed. His expertise has

contributed immensely to keep the quality of the current work intact.

iv
Dr. Claude Martel, my second committee member and professor, for his emotional support and

time-to-time help during the program. His expertise in applied research and educational

consultancy helped me devise better strategies for a successful planning of this project. I always

found him a great person with his empathetic touch to uplift my morale and motivation.

Evgueni Borokhovski, another faculty member from Concordia University, for his regular

support and motivation during the entire journey of this work. He has been exceptional to

support me through all the phases of this work including coding, data analysis and reporting of

the results. He always rendered his services to polish this work. He is such an appealing person

that I befriended with him during my interactions with him.

Saul Carliner, another professor from the faculty of education technology, for offering his

service to chair my defense committee. He has been a great influence on my learning, as his

quality books on instructional design and resourceful peer-reviewed publications throughout his

academic career.

Lastly, I would extend my thanks to all the other teaching community members for their support

and teaching. In addition, I express my gratitude to all the administrative staff of the Department

of Educational technology for their continuous support and care during the entire program. In

particular, I would like to thank Mr. Sean Gordon, my program coordinator, for his amazing

interactions and timely information during the my entire journay in this program.

v
 Dedicated to

With the incessant grace and blessings of the Almighty I

dedicate this work with all my love and gratitude to Mrs.

Sunita, a departed soul and integral part of my family,

for her continuous motivation and support throughout my

academic career,

to my parents and brother who always supported me,

to Anushaka and Aryan, my sweet buds for their affection,

and most importantly

to Anju, my beloved and everything, for her love, trust,

and support throughout my life.

vi
 Table of Contents

Chapter One : Introduction ....................................................................................................................... 1

Background of the Project ........................................................................................................................ 1
Statement of the Problem .......................................................................................................................... 2
Research Questions ................................................................................................................................... 2
Significance of the Study .......................................................................................................................... 3
Student interaction and collaboration ................................................................................................... 5
Technology and pedagogy. ................................................................................................................... 6
Learning and achievement. ................................................................................................................... 8
Chapter Two : Literature Review ............................................................................................................. 9
Terms and Definitions............................................................................................................................. 15
Ethical Considerations and Access to the Studies .................................................................................. 16
Procedures ............................................................................................................................................... 17
Formulating the problem/ research questions. .................................................................................... 17
Meta-analysis as a methodology ......................................................................................................... 17
Study design ........................................................................................................................................ 18
Variables. ............................................................................................................................................ 19
Searching the literature/ data sources.................................................................................................. 20
Criteria for inclusion and exclusion and review procedure. ............................................................... 20
Coding Study Features. ....................................................................................................................... 21
Calculating Effect Sizes. ..................................................................................................................... 21
Synthesis of the Effect Size (ES). ....................................................................................................... 21
Chapter Four : Analysis and Results ...................................................................................................... 23
Descriptors .............................................................................................................................................. 23
General study information. ................................................................................................................. 23
Substantive Characteristics. ................................................................................................................ 25
Effect Size Extraction and Synthesis ...................................................................................................... 29
Procedure for calculating ES............................................................................................................... 29
Synthesis of ES. .................................................................................................................................. 31
Results ..................................................................................................................................................... 32
Sample size. ........................................................................................................................................ 32

vii
 Inter-rater reliability ............................................................................................................................ 34
Achievement Outcomes. ..................................................................................................................... 34
Research design and methodological quality. ..................................................................................... 35
Publication Bias Analysis. .................................................................................................................. 36
Sensitivity Analysis. ........................................................................................................................... 39
Chapter Five : Discussion ....................................................................................................................... 46
Limitations of the Study.......................................................................................................................... 50
Implications and Future Directions ......................................................................................................... 51
Chapter Five : Conclusion........................................................................................................................ 54
Appendix A ................................................................................................................................................ 70
Appendix B ................................................................................................................................................ 74

viii
List of Figures

Figure 1. Funnel plot with effect size for using random model .................................................... 38
Figure 2. Forest plot of 28 effect size from the distribution (k =20) and study-level statistics .... 40

ix
List of Tables

Table 1. Meta-analyses addressing the effects of cooperative/collaborative learning.................. 14

Table 2. Frequency distribution of the type of publication ........................................................... 23
Table 3. Frequency distribution of year of publication................................................................. 24
Table 4. Frequency distribution of time frame of publication ...................................................... 24
Table 5. Shows the collaboration, use of technology and its principal purpose, and the link
between collaboration & technology ............................................................................................ 26
Table 6. Frequency distribution of grade level addressed in the studies ...................................... 28
Table 7. Frequency distribution of subject matter addressed in the studies ................................. 29
Table 8. Study- level statistics used in meta-analysis and explanations (adapted from Bernard et
al. 2014) ........................................................................................................................................ 30
Table 9. Studies included in the meta-analysis with titles ............................................................ 33
Table 10. Overall weighted average random effects and fixed effects sizes & heterogeneity ..... 35
Table 11. Mixed-effects analysis of collaboration strength .......................................................... 41
Table 12. Mixed effects analysis by the instrumental value of technology .................................. 42
Table 13. Mixed-effects analysis of cognitive support for Distributed Cognition & Deep
Learning ........................................................................................................................................ 43
Table 14. Mixed effects analysis using technology for the cognitive support for deep learning . 43
Table 15. Mixed effects analysis by moderator variable subject matter ....................................... 44
Table 16. Mixed effects analysis by treatment duration ............................................................... 45

x
 Chapter One : Introduction

Background of the Project

Teaching and learning in a modern classroom is no longer an act of transferring knowledge. The
act of teaching has become a multidiciplinary enterprize to develop critical thinking, interaction,
and collaboration among learners (Nelson ,1994). Given these multidisciplinary changes in
curriculum and its relative learning objectives, the need to collaborate in order to create learning
environments has gained momnetum in this decade or so. Instead of teacher-centred approaches,
the focus has shifted to learner-centred and learning-centred strategies. In the current educational
landscape, learners are no more the empty vessels to be filled in, rather they need to be the co-
creators of knowledge; they should be willing to take ownership of their learning and contribute
to the development of knowledge.

Collaborative learning, according to Dillenbourg (1999), is “a situation in which two or

more people learn or attempt to learn something together,” and more specifically as joint
problem solving (p. 1). The use of collaborative approaches in pedagogies have gained currency
in the last few decades. According to Wilczenski, Bontrager, Ventrone, & Correia, (2001),
collaborative learning entails "students working together without immediate teacher supervision
in groups small enough that all students can participate collectively in a task" (p. 270). Further,
Roschelle and Teasley evaluate collaboration as the “mutual engagement of participants in a
coordinated effort to solve a problem together” (as cited in Dillenbourg et al., 1996, p. 2). Thus
collaboration can be termed as “coordinated, synchronous activity that is the result of a
continued attempt to construct and maintain a shared conception of a problem” (Roschelle &
Teasley, 1995, p. 70). A collaboratively rich environment has been strongly recommended to
improve students’ cognitive performance, social relationships, and metacognition (Paris &
Winograd, 1990). Keeler & Anson (1995) also asserted that cooperative learning significantly
improved learning performance and reduced attrition among students.

1
 According to Lai (2011), research on collaboration has stemmed from three distinct
strands: research comparing group performance to individual performance, identifying the
conditions which favor or challenge collaboration as more or less effective, and research
exploring the characteristics of interactions that evaluate the impact of collaboration on learning
and achievement including the moderators such as use of new technologies that facilitate
numerous interactions. For example, research have been conducted in this domain including
designing interactive learning environments (Borokhovski, Tamim, Bernard, Abrami, &
Sokolovskaya, 2012), technology integration in postsecondary education (Schmid, Bernard,
Borokhovski, Tamim, Abrami, Surkes, Wade & Woods, 2014), and collaboration and its impact
on student learning (Uribe, Klein, & Sullivan, 2003; Beldarrain, 2006; Williams, 2009; Tomcho,
& Foels, 2012).

Statement of the Problem

In light of these issues, the purpose of this research is to undertake a meta-analysis to

investigate the effect of collaborative learning on student achievement and the role of other
associated moderators such as technology, subject domains, grade levels and duration to
facilitate and/or support collaborative learning impacting student achievement.

Research Questions

The current study attempts to investigate the effect of collaborative learning on student
achievement while comparing the two conditions – high collaboration (experimental condition)
and less or no collaboration (control condition). In addition, the study is interested in exploring
the moderator variables which affect student collaboration and achievement outcomes. The study
also examines if the degree of collaboration has a varying effect at different grade levels, subject
domains, and duration which in turn might influence student achievement outcomes. The
following three research questions guide this meta-analysis:

2
 1) Does collaborative learning have any statistically significant effect on student
achievement outcomes when compared learning without (or under lesser degree of)
collaboration?
2) Do different types of technology have varying effects on student achievement when used
to enhance/support /promote collaboration?
3) Do grade levels and subject domains have any moderating effects on student`s
achievement?

Significance of the Study

The study is significant in that interaction and collaboration help develop and improve
student performance and their achievement outcomes. Accroding to Van Boxtel, Van der Linden,
and Kanselaar (2000), collaborative learning activities help learners to find explanations of their
understanding that assists them elaborate and reorganize their knowledge. Various studies have
found that collaboration among students has considerable impact on their achievement outcomes
(Borokhovski, Bernard, Tamim, Schmid, & Sokolovskaya, 2016; Fjermestad, 2004; Schmid et
al., 2014). Some meta-analyses which focussed on the use of technology in distance and/or
online learning, have found a positive impact of technology on student collaboration and on
learning outcomes (Abrami, Bernard, Bures, Borokhovski, & Tamim, 2011; Beldarrain, 2006).
Another group of studies explored design-based cooperative /collaborative learning, and reported
a significant impact of collaborative activities on student learning outcomes (Lou, Abrami, &
Appolonia, 2001; Chou & Min, 2009; Lee, 2007; Puzio & Colby, 2013; Wright, Kandel-Cisco,
Hodges, Metoyer, Boriack, Franco-Fuenmayor & Waxman, 2013). Other researchers who
investigated the pedagogical and theoretical aspects of technology integration (Arts, Gijselaers &
Segers, 2002; Jehng, 1997; Kanuka & Anderson, 1999; Pedró, 2005; Lou, Bernard, & Abrami,
2006; Mantri, Dutt, Gupta & Chitkara, 2008), found mixed effects of technology intergration on
student learning outcomes.

3
 Furthermore, Bernard, Abrami, Borokhovski, Wade, Tamim, Surkes, and Bethel (2009)
were interested in the relative effectiveness of the designed and contextual interaction treatments
in distance education . They found a strong association between the strength of interaction
treatments and achievement for asynchronous DE courses compared to courses containing
mediated synchronous or face-to-face inter (p. 1243). Borokhovski et al. (2012) found that
higher levels of collaboration and cooperation among students could be achived by employing
technology to enable , support and facilitate discussions (p. 321). Schmid et al. (2014) explored
the impact of technology-enhanced instruction in postsecondary education and reported that
learning is best supported when students are engaged in active and meaningful exercises via
technological tools that provide cognitive support. (p. 285). According to Baepler, Walker, and
Driessen (2014), in an active learning classroom, student faculty contact could be reduced by
two-thirds and students achieved learning outcomes that were at least as good, and in one
comparison significantly better than, those in a traditional classroom (p. 227).

Borokhovski et al. (2016) tried to map out the added value of planned collaborative
activities versus unplanned grouping of students in the context of postsecondary education. They
reported that the designed-based treatments outperformed contextual treatments on measures of
achievement and they strongly suggested the value of planning and instructional designing in
technology integration in post-secondary education (p. 15). These studies have contributed to the
literature on collaborative learning. Though Kyndt, Raes, Lismont, Timmers, Cascallar, and
Dochy (2013) have explored the effects of face-to-face cooperative learning in this regard, no
study has attempted to investigate the effect (s) of collaborative learning on student achievement
in both online and and face-to -face modes collectively across varied levels of formal education
and subject domains. Therefore, it would be interesting to investigate how the effect of
collaborative learning on student achievement varys, especially, amidst the different grade levels
and subject domains in both online and face-to-face modes of instruction.The current meta-
analysis will fill this gap by exploring the effect of collaborative learning on student achievement
in the context of formal education settings cutting across all grades and subject domains. Indeed,

4
the findings of this study will contribute to the pool of existing body of knowledge on
collaboration. Furthermore, while the outcomes of this meta-analysis may have limited
generalizability, this study would serve as a springboard for the prospective studies in the
domain.

Student interaction and collaboration. Interactions among student-student, student-

teacher and student-content are vital to engage learners for collaboration (Schmid et al., 2014).
Since interaction is a primary condition to collaborate, three types of interactions, as mentioned
above, have been effective, particularly in various theoretical frameworks on distance and online
education (Anderson, 2003; Beldarrain, 2006; Moore, 1989). Interactions among students trigger
the learning process. In this connection, the understanding of collaborative and cooperative
learning is a must (Decuyper, Dochy, & Van den Bossche, 2010). As collaboration is an
umbrella term, it encompasses interactions between and/or among student-student, teacher-
students and student-content (Anderson, 2003; Beldarrain, 2006; Moore, 1989). Roschelle and
Teasley define collaboration as a “mutual engagement of participants in a coordinated effort to
solve a problem together” (Dillenbourg et al., 1996, p. 2). The central assumption of
collaboration lies in the interaction between individuals or groups. It may take many forms –
cooperative, collective, peers, group and/or team learning. According to some authors (Bruffee
1995; Eastmond, 1995; Webb & Palincsar, 1996), small-group learning strategies reside along a
continuum from loosely structured (i.e. collaborative) to highly structured (i.e. cooperative).

Presumably, the aim of collaborative learning is the co-construction of knowledge

through interaction. So collaborative learning refers to a variety of educational approaches that
encourage students to work together, including cooperative learning; problem-based instruction;
guided design; writing groups; peer teaching; workshops; discussion groups; and learning
communities (Smith & McGregor, 1992). Vygotsky (1962) stressed the need for collaborative
learning either between teacher-students or among students to assist students in advancing
through their zone of proximal development (the gap between what can a student achieve alone
and in collaboration with the group). Collaboration can be manifested in the face-to-face form

5
(in-class group project) and online technology-mediated formats (synchronous and asynchronous
media–wikis, blogs, forums and any other forms of online communication). However, Kirschner
and Erkens (2013) relate:

It has become clear that simply placing learners in a group and assigning them a task does
not guarantee that they will work together … coordinate their activities … engage in
effective collaborative learning processes… lead to positive learning outcomes (p. 1).

Regardless of the environments in which collaborative learning is implemented, there is an

agreement that simply grouping students does not promote higher achievement or more positive
relationships among students (Johnson & Johnson, 1996). Collaborative learning means specific
forms of interaction that have a purposive intent and a general set of structural elements to
facilitate it (Harasim, Hiltz, Teles, & Turoff, 1995).

Concerning collaborative learning, the allied concept of cooperation is also important. In

cooperative learning individuals work together to achieve shared goals; they collaborate to
maximize their own and each other's learning (Johnson & Johnson, 1989). According to Johnson
and Johnson (2009), five elements must be present in cooperative/collaborative learning: 1)
positive interdependence, 2) individual accountability, 3) face-to-face promotive interaction, 4)
social and interpersonal skills, and 5) group processing. However, in the current project, the
focus is on the purposive cooperative/collaborative activities which enhance students’
achievement. Collaboration is understood as interaction between and/or among student-student
(Bernard et al., 2009 ; Borokhovski et al., 2012), teacher-students and student-content
(Anderson, 2003; Beldarrain, 2006; Moore, 1989) which enhance student learning outcomes,
excluding collaboration meant for non-academic intentions.

Technology and pedagogy. Technology has been revolutionizing the learning sciences

significantly since the 1990s when the internet access with personal computers became widely
available for educational purposes (Kozma, 1994), and computers provided more capacity for
information processing and other advanced functions. The historical debate between Clark and

6
Kozma over the use of technology as a medium or more than a medium opened the floodgate for
experimentation on and initiation of multiple instructional strategies in and out of the classroom,
promising the significance of technology to support student learning and attitudes (Dede, 1996,
2004; Kozma, 1991, 1994; Mayer, 2008).

Since then, technology has become a buzz word in the parlance of the current educational
practices. In the last fifteen years, the influx of technological innovations has bombarded the
field of education immensely. Technology is not merely a set of hardware or software
paraphernalia, but also is, as Ross, Morrison, and Lowther (2010) define, “a broad variety of
modalities, tools, and strategies for learning, [whose] effectiveness, therefore, depends on how
well [they] help teachers and students achieve the desired instructional goals” (p. 19).
Technology-mediated instruction has complimented education in enhancing teaching and
learning both inside and outside the classroom. Uses of computer-assisted technology and cloud-
based authoring technology have facilitated multiple forms of pedagogical and instructional
strategies such as online learning/distance education (Bernard, Rojo de Rubalcava, & St-Pierre,
2000 ; Bernard, Abrami, Lou, Borokhovski, Wade, Wozney, 2004 ; Bernard et al., 2009),
blended learning (Henrie, Halverson, & Graham, 2015 ; Schmid et al., 2014), and MOOCs
(Margaryan, Bianco, & Littlejohn, 2015). In addition, technology has widened the access and
scope of learning and offered options to collaborate worldwide.

The use and impact of technology on student learning, however, are two different
things. This is so because the impact is directly decided by the manner and the purpose of the use
of technology. Technology as a moderator variable in collaborative learning is interesting for
many reasons. Given recent technologies such as cognitive tools, communication methods,
search and retrieval strategies, and other presentational tools, it would be interesting to
investigate the use of technology in collaborative activities that impacts student achievement. A
meta-analysis by Susman (1998) found that participants in collaborative, computer-based
conditions showed a greater increase in elaboration, higher-order thinking, metacognitive
processes, and divergent thinking than participants in individual computer-based instruction.

7
Given this perspective, it would be interesting to know how the use of technology in its varied
manifestations facilitate collaboration which might affect student achievement. In other words,
this study explores whether and how technology-supported collaborative activities help learners
interact and collaborate leading to enhanced achievement outcomes.

Learning and achievement. Dillenbourg (1999) defines “learning” as a “biological

and/or cultural process” that takes place over several years; a “joint problem solving in which
learning is assumed to occur as a by-product of interactions” (p.4). Learning can be more
effective when students discuss their ideas, experiences, and perceptions with peers (Jonassen &
Kwon, 2001; Kanuka & Anderson, 1998). In a face-to-face classroom setting, collaborative
learning can affect learning outcomes, social skill development, and self-esteem positively
(Johnson & Johnson, 1996; Slavin, 1990). Similarly, the use of collaborative learning strategies
in computer-based instruction (CBI) is also supported by research (Cavalier & Klein, 1998;
Dalton, Hannafin, & Hooper, 1989 ; Hooper 1992; Hooper & Hannafin, 1991; Klein & Doran,
1999; Kulik & Kulik, 1991; Sherman & Klein, 1995). In addition, theorists also have
propounded that computer-mediated collaboration can positively enhance learning, problem-
solving, and other higher-order thinking (Adelskold, Alklett, Axelsson, & Blomgren, 1999 ;
Johnson & Johnson, 1996 ; Jonassen, Prevish, Christy, & Stavulaki, 1999). In the context of the
current study, the analysis is restricted to the collaborative activities which affect student
achievement and how different moderator variables, including technology, grade levels, subject
domains, and duration influence student collaboration.

Achievement, as a dependent variable, informs the academic performance by means of

standardized and/or validated measures (Schmid et al. 2014). However, in the current study, the
term achievement is confined to students’ accomplishment, participation, success, and attainment
in the parlance of collaborative activities measured by exams, tests, and grades.

8
 Chapter Two : Literature Review

Historically, according to Dillenbourg, Baker, Blaye, and O’Malley (1996), a great

amount of research on collaborative and cooperative learning stemmed from the works of Piaget
and Vygotsky. Piaget’s system of developmental stages describing children’s cognitive progress
contributed to the development of socio-constructivists paradigm. Similarly, his concept of
cognitive conflict provided a cognitive framework to understand learner’s experiences. In
addition, social interactions mediate cognitive conflict that allow learners to interact with peers at
more advanced developmental levels.

On the contrary, regarding cognitive conflict Vygotsky stressed the value of social
interaction itself for causing individual cognitive change, as opposed to being merely stimulated
by it (Dillenbourg et al., 1996). Internalized social interaction causes conceptual changes in
participants that help them negotiate meaning. A similar concept, the zone of proximal
development, according to Vygotsky, is the distance between what a student can accomplish
individually and what he/she can accomplish with the help of a more capable “other.” While
Piaget suggests pairing children based on different developmental stages to facilitate cognitive
conflict, Vygotsky, on the other hand, recommends pairing children with adults. Unlike Piaget
and Vygotsky who maintain that cognitive conflict causes conceptual change, socio-culturalists
privilege collaborative learning that takes place within the zone of proximal development
(Dillenbourg et al., 1996).

According to Kreijns, Kirschner, and Jochems (2003), a new strand of research regarding
collaborative learning emerged in the late 1990s that focused on new technologies for mediating,
observing, and recording interactions during. On the whole, four strands came into existence out
of the seminal works of Piaget, Vygotsky and their shared concept of cognition and research
built on them – the “effect” paradigm, the “conditions” paradigm, the “interactions” paradigm,
and “computer-supported” paradigm respectively (Dillenbourg et al., 1996, pp. 8 -17 ). In the
next paragraphs, the author discusses these paradigms.

9
 The “effect” paradigm investigates outcomes of collaboration rather than the
collaborative process itself, and compares group performance with individual performance. This
paradigm maintains that a collaborative classroom culture can have powerful effects on student
learning and performance. Webb (1993) found that the students who worked in groups on
computational math problems scored significantly higher than equivalent-ability students who
worked individually.

The “conditions” paradigm tries to determine the conditions that moderate the
effectiveness of collaboration on learning, for instance, individual characteristics of group
members, group heterogeneity and size, and task features. Webb’s (1991) study reported
significant differences in the collaborative learning experiences of boys and girls. Boys were
more likely than girls to give and receive elaborated explanations, and their explanations were
more likely to be accepted by group mates than girls’ explanations (Dillenbourg et al., 1996).

To overcome the complexities of former paradigms, the “interactions” paradigm emerged

to identify the intersecting systems between collaboration and learning outcomes. In a way, this
paradigm tried to explain the characteristics and processes of interactions which measure the
effect of collaboration on learning (Dillenbourg et al., 1996). In this regard, Webb (1991)
reported that the effect of collaborative learning on student achievement depends on the quality
of the interactions among them.

The fourth paradigm, the contemporary one, was developed to explore whether the
theoretical benefits of collaborative learning as harvested in face-to-face settings can be repeated
in a computer-mediated or computer-assisted interactions given the asynchronous, text-based
interactions. For example, Curtis and Lawson (2001) found that in online media there were fewer
exchanges among student during collaboration given their unfamiliarity prior to online
interactions. Further, the “online medium was found effective only in planning activities and
coordinating work than challenging ideas” (pp. 29-30). For the purpose of the current study, the
author will rely upon the collective contributions of these approaches to investigate his research
questions.

10
 Over the past three and half decades numerous meta-analyses have been conducted to
investigate the effects of collaborative and small-group instruction on student learning and
achievement outcomes. Twelve meta-analyses spanning from 1981 to 2016 are discussed in the
next section (See Table 1). Johnson, Maruyama, Johnson, Nelson, and Skon (1981) reported that
cooperation both with and without intergroup competition is more effective than the
interpersonal competition and individual efforts. Similarly, Newmann and Thompson’s (1987)
study which was conducted in the context of secondary education found 68% of the studies
yielded positive effects in favor of the cooperative condition. Qin, Johnson, and Johnson (1995)
investigated cooperative versus competitive efforts and problem solving. They found that studies
with non-linguistic problems (for example the study domain of mathematics or exact sciences)
showed slightly more positive effects than studies with linguistic problems. Lou, Abrami,
Spence, Poulson, Chambers, and d’Apollonia, (1996) who explored the differences in
achievement and attitudes at all grade levels of education, concluded that “on average, students
learning in small groups within classrooms achieved significantly more than students not
learning in small groups” (p. 439).

On the other hand, Johnson, Johnson, and Smith (1998) focused their research on higher
education settings and adults. They reported that cooperative learning results in positive effects
on achievement in comparison with competitive or individualistic learning. Springer, Stanne, and
Donovan’s meta-analysis (1999) investigated the effects of cooperative learning on achievement,
attitudes and persistence in the context of undergraduate STEM courses. They reported that
students who were learning in cooperative groups showed better achievement than students who
were not learning in cooperative groups. Bowen’s (2000) second meta-analysis, which focused
on high school and college level chemistry students, pointed out that “on average, using aspects
of cooperative learning can enhance chemistry achievement for high school and college
students” (p. 119). He found that cooperative learning had a significant positive effect on student
attitudes toward STEM courses.

11
 Another meta-analysis conducted by Lou, Abrami, and D’Appolonia (2001) reported
that, on average, small group learning had significantly more positive effects than individual
learning on student individual achievement (mean ES = + 0.15) and on group task performance
(mean ES = +0.31). Similarly, Lou et al. (2006) reported a significant correlation between
student-student interaction and greater achievement success (g+ = 0.11, k = 30, p < .05) in the
context of the undergraduate distance education courses. In the same vein, Bernard et al. (2009)
were interested in three types of interaction treatments (i.e. student-student, student-teacher, and
student-content). They found an explicit link between interaction and academic performance in
distance education that improved student learning. The student-student interaction emerged as
the most important group among the three ( g+ = 0.49, k = 10, p < .05). In addition, they found
higher achievement effect size regarding the presence of technology which appeared to have
facilitated or at least improved the effectiveness of interaction among students, as reflected in
achievement learning outcomes.

Furthermore, Bernard et al. (2004) found a synergistic relationship between technology

and pedagogy in distance education and speculated the same relationships in the context of other
learning environments where the use of technology in designing environment might promote
student collaboration and knowledge construction.Tamim, Bernard, Borokhovski, Abrami, and
Schmid, (2011) also provided a useful insight about the use of technology in regular formal
educational contexts. The authors (Tamim et al., 2011) found that technology had a positive
impact on student achievement. In the same coin, Schmid and colleagues (2014) examined the
effects of technology integration on students in postsecondary classroom settings. The degree of
technology use was the defining characteristic for the effect size. The authors concluded that
learning was best supported when students were engaged in active and meaningful exercises
using technological tools that provided them cognitive support.Technological tools themselves
do not make learning happen. In this connection, Abrami and colleagues (2011) rightly remarked
:

12
 [J]ust because opportunities for interactions were offered to students does not mean that
students availed themselves of them, or if they did interact, that they did so effectively.
The latter case is the more likely event, so the achievement effects resulting from well-
implemented interaction conditions may be underestimated in our review (p. 86).

Borokhovski et al. (2012) observed that interactively designed activities are more
conducive to increase student learning than do the contextual instructional settings which are not
intentionally designed to create collaborative learning environments. Recently, Borokhovski and
colleagues (2016) found that designed treatments outperformed contextual treatments ( g+ =
0.52, k = 25 vs. g+ = 0.11, k = 20, Q-Between = 7.91, p < .02) on measures of achievement and
emphasized the importance of planning and instructional design in the integration of technology
at postsecondary levels. Similarly, Kyndt, Raes, Lismont, Timmers, Cascallar, and Dochy (2013)
reported a positive effect of cooperative learning on student achievement and attitudes. In
addition, they reported that the study domain, the age group, and the students’ cultures also
produce variations in effect size.

Having discussed these meta-analsyes, the cooperative / collaborative learning

showed a positive effect on student achivevment. Though these studies have contributed to the
literature on collaborative learning, however, there is a discernible gap informed by these studies.
The gap is : How do the moderators such as technology, subject domain, and grade levels
collectively affect the student collaboration and in turn student achivement? The current meta-
analysis will fill this gap.

13
Table 1. Meta-analyses addressing the effects of cooperative/collaborative learning

Review Study (Year) k Dependent Conditions of the Independent Mean Effect Size
variable variable
Johnson et al. (1981) 122 Achievement Coop./Competitive 0.78

Qin et al. (1995) 46 Achievement Coop./Competitive 0.55

Lou et al. (1996) 51 Achievement Within class grouping / no 0.17

Johnson et al. (1998) 168 Achievement - Coop./Competitive 0.49

Springer et al. (1999) 37 Achievement - Coop./Competitive 0.51

Bowen (2000) 37 Achievement Coop./Traditional 0.51

Lou et al. (2001) 122 Achievement Small group/ Individual 0.15

Bernard et al. (2009) 74 Achievement Interaction treatments (ITs) with 0.38

other DE instructional treatments
Schmid, R. et al. (2009) 310 Achievement High Technology/ some or no 0.28
technology
Tamim, R. M. (2009). 37 Achievement Face-to-face/Traditional 0.30

Kyndt, Eva et al. (2013) 65 Achievement Coop./Traditional 0.54

Borokhovski, Eugene (2016) 45 Student learning Designed treatments /contextual 0.52

outcomes treatments groups

14
 Chapter Three : Methodology

In this section, firstly, the author defines the major terms as the working definitions
associated with the research problem and the research questions to be investigated in the current
study. Secondly, he informs about the proceedures used for the ethical considerations meant for
this meta-analysis. Thirdly, the author unpacks the methodology which includes : 1) the study
design, variables, methods and instruments ; 2) literature search ; 3) the coding study features ;
and 4) process of the calculation and synthesis of the effect sizes.

Terms and Definitions

Collaboration: According to Panitz (1996), is a philosophy of interaction and personal lifestyle

whereas cooperation is a structure of interaction, a classroom technique, designed to facilitate the
accomplishment of an end product or goal. Cooper, Prescott, Cook, Smith, and Mueck (1990)
have defined, cooperative/collaborative learning as “an instructional technique that requires
students to work together in small, fixed groups on a structured learning task”. The current study
focuses on collaborative learning as “a situation in which two or more people learn or attempt to
learn something together,” and more specifically as joint problem solving (Dillenbourg, 1999, p.
1).

Formal educational settings: According to Coombs and Ahmed (1974), a formal education
setting is structured in a hierarchical manner that spans from primary school to university levels
including general academic studies and a varied of specialized programs for full-time technical
and professional training.

Pedagogical uses of technology : Based on the meta-analysis of Schmid et al. (2014), the
current meta-analysis distinguishes among the following pedagogical uses (major purposes) of
educational technology :

15
1) To promote communication and/or facilitate the exchange of information that includes
technology which enables a higher level of interaction between individuals (i.e., two-way
communications among learners and between learners and the teacher);

2) To provide cognitive support for learners which encompasses various technologies that
enable, facilitate, and support learning by providing cognitive tools (e.g., concept maps,
simulations, wikis, different forms of elaborate feedback, spreadsheets, word processing);

3) To facilitate information search and retrieval using technology to access to additional

information (e.g., web links, search engines, electronic databases); and

4) To enable or enhance content presentation where technology is primarily used by teachers

and/or students to present, illustrate and otherwise enrich the content of instruction for example
PowerPoint presentations, graphical visualizations, computer tutorials with limited interactive
features.

Achievement: According to Ollendick, & Schroeder (2003), ‘‘academic achievement is the

knowledge and skills that an individual learns through direct instruction’’ (p. 1). For the current
meta-analysis, the author uses any objective measure of academic performance (e.g., exams or
test scores) including but not limited to standardized and validated measures to estimate student
achievement.

Ethical Considerations and Access to the Studies

The current meta-analysis is a part of an on-going large meta-analysis conducted by

Bernard et al. (2014). The selected studies included in the current meta-analysis is a part of
Bernard et al. (2014) collection of studies. Consent has been granted from Bernard and
colleagues to access and use these studies before the start of this meta-analysis.

16
Procedures

Formulating the problem/ research questions. The current project primarily attempts to
explore the differential effects of high versus low/no level of collaboration on student
achievement. Of the additional interest is to decipher further the moderators which promote and
/or hinder the effect of collaborative on student achievement. The author formulated three
research questions to investigate the effect of collaborative learning on achievement:

1) Does collaborative learning have any statistically significant effect on student

achievement outcomes when compared learning without (or under lesser degree of)
collaboration?
2) Do different types of technology have varying effects on student achievement when used
to enhance/support /promote collaboration?
3) Do grade levels and subject domains have any moderating effects on student`s
achievement?

To answer these questions, a quantitative approach was used. More specifically, the
author utilized a meta-analysis approach. Collaboration, as the independent treatment variable,
used to measure its effect on student achievement. In addition, an analysis of moderator variables
was conducted to examine whether moderators such as specific forms (purposes) of technology
use, subject demographics, and grade levels could help or hinder collaboration that would affect
student achievement outcomes.

Meta-analysis as a methodology. In general, a meta-analysis (quantitative synthesis) is a

statistical analysis of the results from individual studies that address a common research
question, examine systematic sources of differences in results among these studies, and leading
to a quantitative summary of the results (Porta, 2008). There are two widely used approaches to a
meta-analysis in the research literature. First, the Hunter and Schmidt’s (1990) psychometric
meta-analysis and the second is the classic or Glass’ meta-analysis. Accordind to Bangert-
Drowns and Rudner (1991), Hunter and Schmidt's approach is a mix-bag of some of the best

17
characteristics of other approaches. In this approach, all studies related to a given problem are
collected without any consideration to the qualities of studies. Then, the distribution of effect
sizes is corrected for any variability among studies such as sampling error, measurement error,
range restriction, and other systematic artifacts. Even if any variability affect the distribution of
overall effect size, then, “the effect sizes are grouped into subsets according to preselected study
features, and each subset is meta-analyzed separately” (p.3). Unfortunately, this technique is not
very feasible for my project, for this requires substantial information from individual studies for
accurate correction of effect sizes. In reality, however, such information are not always available
in all the studies.

On the other hand, Glass' approach to meta-analysis is much more of conventional. This
approach starts with defining questions to be examined, then, collecting studies, coding study
features and outcomes, and finally, analyzing the relations between study features and outcomes.
In addition, firstly, Glassian meta-analysis applies liberal inclusion criteria and stresses not to

disregard studies based on study quality a priori; a meta-analysis itself can determine if
study quality is related to variance in reported treatment effect. Secondly, the unit of
analysis is the study finding. A single study can report many comparisons between
groups and subgroups on different criteria. Effect sizes are calculated for each
comparison. Thirdly, meta-analysts using this approach may average effects from
different dependent variables, even when these measure different constructs (Bangert-
Drowns, & Rudner, 1991, p. 3).

For the current meta-analysis, I will use Glassian approach because it is quite robust for the
critical re-analysis, its use of conventional statistical tests, and systematic in design.

Study design. The sample for this analysis was selected from an existing pool of studies
which belonged to Bernard and colleagues’ (2014) onging larger meta-analysis. The current

18
meta-analysis included 20 studies representative of the main research question resulted for a
review process involving abstracts and full study analysis of 78 potentially relevant studies. The
study compared two groups – higher degree of collaboration (as the experimental condition) with
low/no collaboration (the control condition). Additionally, the moderator analyses of the other
associative variables such as type and major purpose of technology use, subject domains,
duration, and grade levels also conducted to measure the relative influences of these factors on
collaboration and for that matter on student achievement.

First, two reviewers coded five studies independently to decide whether the experimental
condition of each study satisfy the inclusion criteria of been higher in collaboration than the
control condition and featuring educational technology in experimental condition. In addition,
twenty-four study features (methodological, substantive, and demographic) were coded for
further use in the moderator variable analysis. The average pairwise agreement rate on the initial
coding was 84.17% (Cohen’s kappa = 0.68). Disagreements were resolved either through
discussion between reviewers or by the involvement of a third party. Two studies (Terwel, Oers,
Dijk, & Eden, 2009; Mastropieri, Scruggs, Norland, Berkeley, Mcduffie, Tornquist, & Connors,
(2006) were excluded from the sample given the absence of technology in those studies. After
establishing sufficient reliability, the author reviewed the rest of the original sample. A final
sample of 20 included studies yielded 28 independent effect sizes with a total of 2434
participants.

Variables. The current study investigates the effect of collaborative learning on student
achievement. As stated previously, a treatment variable collaboration is used (with high in
experimental and low or no collaboration in control conditions) to measure its effect on the
dependent variable of student achievement. Specifically, the meta-analysis aimed to estimate the
weighted average effect size (i.e. how much better – positive effect, or worth – negative effect
Experimental group compared to Control group in terms of their respective achievement
outcomes). Among other variables, included in analyses as moderators, were technology type
and major purpose of use, subject demographics, grade level, and treatment duration.

19
 Searching the literature/ data sources. As mentioned earlier, the current representative
sample is a part of the study collection authored by Bernard et al. (2014), comprised of numerous
primary research studies identified through extensive systematic literature searches designed and
conducted to identify and retrieve studies relevant to the research questions. The literature search
involved more than ten electronic databases (e.g. ERIC, EdLib, Education Abstracts, Medline,
ProQuest Digital Dissertations & Theses, PsycINFO, British Education Index) branching from
previous relevant reviews and tables of content for major educational journals. In addition, the
manual Google Internet searches and searches for various conference proceedings to form a pool
of relevant studies were performed.

Criteria for inclusion and exclusion and review procedure. The representative sample
was selected using several qualifying criteria before including the studies into the current meta-
analysis. The criteria entailed, firstly, the studies should be conducted in formal educational
settings with varying grade levels from elementary and secondary to higher education. Only
empirical studies with collaboration /cooperation either in face-to-face and/or virtual
learning/online such as computer-based collaborative learning were included. A set of inclusion
criteria, as discussed below, guided the study characteristics required to retain the studies for
inclusion. Studies that did not meet the following criteria were excluded from the current meta-
analysis.The inclusion of studies needed to have:

 To be published no earlier than 1996;

 To be publically available (or archived);
 Two main factors – collaboration and technology;
 Address students’ achievements, as measured for example by final exams, cumulative
posttest scores;
 Contain at least one between-group comparison where one group is considered the
experimental condition (i.e., high level of collaboration) and the other group the control
condition (i.e., lower/ no collaboration);
 Contain sufficient statistical information for effect size extraction.

20
 Coding Study Features. Coded moderator variables (i.e., study features) used to explore
between-study variability in effect sizes. The study features used were mainly based on those
employed by Bernard et al., 2009 (in studying distance education) and Schmid et al., 2014
(technology integration in postsecondary education). Major study features, in addition to effect-
size defining difference in degree of collaboration, include type and purpose of technology uses,
presence of technology, subject demographics, grade levels, treatment duration, etc. (See
Appendix A for the codebook).

Calculating Effect Sizes. The estimation of effectiveness of collaborative learning was

based on extracting relevant effect sizes from included primary studies. In particular, d-type
standardized mean difference effect size used as the common metric (i.e., Cohen’s d). The
𝑆𝐷
equation for its extraction is expressed as d = 𝑋𝑒 − 𝑋𝑐 ൗ . In case of non-availability of
𝑝𝑜𝑜𝑙𝑒𝑑
the descriptive statistics, effect sizes were extracted from inferential statistics, such as t-tests, F-
tests, or exact p-values, using conversion equations from Glass, McGaw, & Smith (1981) and
Hedges, Shymansky, & Woodworth (1989). To correct for small sample bias, ds were converted
to the unbiased estimates of g (Hedges & Olkin, 1985).

Synthesis of the Effect Size (ES). The synthesis of the data conducted using the random
effects model. Model selection is justified by relative non-uniformity of treatment conditions,
rather limited, i.e., non-exhaustive nature of the collection of studies, and, thus, likely
heterogeneity of the distribution of effect sizes, confirmed in actual analyses (Borenstein,
Hedges, Higgins, & Rothstein, 2009; Hedges & Olkin, 1985). The random effects model is used
to aggregate and report average effect sizes (g +), standard errors (SE), confidence intervals
(lower 95th and upper 95th) and z values with associated p-values, when systematic variation in
the distribution of effect sizes is not assumed and error term (tau-squared) is randomly added to
the weights of individual effects.

Furthermore, the analyses of the moderator variables was conducted to ascertain the
relative effectiveness of the moderators on the dependent variable according to the so-called

21
mixed model. In this model, the average effect sizes for categories of the moderators were
calculated using the random effects model. The variance component Q-Between calculated
across categories using a fixed effect model (Borenstein et al., 2009). All analyses, including
sensitivity and publication bias analysis, were performed in Comprehensive Meta-AnalysisTM
2.2.048 (Borenstein, Hedges, Higgins, & Rothstein, 2005). Eventually, a posthoc test
(Bonferroni correction) was conducted at the selected levels of the moderator variables.

22
 Chapter Four : Analysis and Results

This chapter entails various stages of the analyses and reporting of the results including
an overview of the selected studies, analyses aiming at the publication bias and sensitivity for
any outlier effect, and explaining heterogeneity across the included studies using moderator
variable analysis of methodological, substantive, and demographic study features.

Descriptors

This section consists of the descriptive data regarding general study information,
explanation of the Effect Size (ES) extraction procedures, as well as substantive, methodological
and contextual/demographic features of the included studies. Microsoft Word and Excel software
were used to classifiy these study features.

General study information. The general study information includes the type and the
year of publication. The studies included in this meta-analysis were selected from journal
publications, dissertations, and reports. The most frequent type was the journal publication
consisting of 16 of the included studies representing 80% of the included sample .

Table 2. Frequency distribution of the type of publication

Type of Publication Frequency Relative %

Journal 16 80.00
Dissertation 2 9.09
Report 2 9.09
Total 20 100

23
 As for the year of publication, the included studies were published between the years 1997
and 2010. The years of 2006 and 2008 offered the larger number of studies comprised four and
six published studies respectively. The frequency distribution is presented in Table 3.

Table 3. Frequency distribution of year of publication

Publication year Frequency Relative %

1997 1 5.00
1999 2 10.00
2002 1 5.00
2004 1 5.00
2006 4 20.00
2007 2 10.00
2008 6 30.00
2009 1 5.00
2010 2 10.00
Total 20 100

Then the publication dates were grouped into four units for a broader look. In the years between
2006 and 2010, fifteen studies (75%) published. The frequency distribution within respective
time frames is presented in Table 4.

Table 4. Frequency distribution of time frame of publication

Publication year Frequency Relative %

1997-1999 3 15.00
2000-2004 2 10.00
2006-2007 6 30.00
2008-2010 9 45.00

Total 20 100

24
 Substantive Characteristics. The primary premise established for this meta-analysis was
the investigation of the effect of collaborative learning on student achievement. The operational
definition for the variable of collaboration is presented in the codebook. A higher level of
collaboration was the necessary and sufficient criterion to distinguish the experimental
conditions from control ones. Among the moderator variables, technology was the most
important factor. The technology was further classified either as instrumental in
enabling/promoting/supporting collaboration among students or as just contextually present,
without any apparent influence on student collaborative work. Given the criterion of high
collaboration in the experimental condition, the moderator analysis was conducted to know the
role of technology in supporting, scaffolding and/or enhancing collaboration and to determine
what effect it has on student achievement. In the experimental condition, technology was
investigated via its types, purposes, and instrumental values when it was intentionally integrated
into the collaborative activities.

Table 5 in the next section maps out the types of technology included and their purposes
of use. In eight studies (40%), technology was used in a mixed manner, i.e. used for more than
one purpose. In five studies representing 25% of the total, technology was used for cognitive
support II (i.e., deep learning, e.g., simulations) and in three studies, technology was utilized for
the cognitive support I (i.e., distributed cognition, e.g., Excel) representing 15% of the total.
Interestingly, out of the final 20 included studies for analysis, 16 studies found a link between
collaboration and technology forming 80% of the collection. Only four studies were found with
no connection. Among the major types of technology utilized were web-based computer-driven
technology including ICT and other software applications followed by multimedia and videos.

25
Table 5. Shows the collaboration, use of technology and its principal purpose, and the link between collaboration & technology

Reference Collaboration Technology type Technology Link between

Purpose collaboration &
(Major) technology
Arts, J. A. et al. (2006) Low Web-based CMC & Multimedia/ Internet 1 YES
technology
Barnes, L. J. (2008) Audience Response Systems-Qwizdom & 6: 1+5 YES
Videos
Cavalier, J. & Klein, J. D. CBI: Hypercard (interactive) tutorial 6: 1&5 YES
(1998) Computers
DePriter, T. N. (2008) Computers & web-based discussion forum 6: 1&4 NO
Faro, S., & Swan, K. (2006) Low Web-based (on-line) study materials and 6: 1&4 YES
_1 chat &Studio/ CBI/ Online technology
Faro, S., & Swan, K. (2006) Low N/A N/A YES
_2
Freeman, S. et al. (2007) _1 Clickers 6: 3&4 YES
Freeman, S. et al. (2007) _2 Clickers 6: 3&4 YES
Freeman, S. et al. (2007) _3 Cards 6: 3&4 YES
Freeman, S. et al. (2007) _4 Cards 6: 3&4 YES
Gersten, R. et al. (2006) _1 Video & videos 4 YES
Gersten, R. et al. (2006) _2 Video & videos 4 YES
Hernández-Ramos, P. et al. Low mPower software 3 YES
(2009)
Hodges, T. L. (2008) Web CT for administer the exam &Board 2 YES
game
Hoon, T. S. et al. (2010) Low Computer/courseware 3 YES
Hosal‐Akman & Simga- High PowerPoint/slides 5 YES
Mugan (2010) _1
Hosal‐Akman & Simga- High N/A N/A YES
Mugan (2010) _2

26
 Hummel, H. G. et al. (2006) "Plea checker" computer program & Virtual 2 YES
Program/ emails
Lin, J. M., Wu, C., & Liu, H. SimCPU software package for computer 3 NO
(1999) _1 labSimCPU
(Computer software)
Lin, J. M., Wu, C., & Liu, H. SimCPU software package for computer 3 NO
(1999) _2

Nugent, G. et al. (2008) Mobile library, digital cameras & digital 6: 3&4 YES
cameras
Olgun, O. S., & Adali, B. Low Internet sites & Internet search tools 4 YES
(2008)
Pariseau, S. E. et al. (2007) Computer applications (e.g., Excel) & 2 YES
Laptops
Priebe, R. (1997) Burton Comprehension Instrument (BCI) & 3 NO
Propositional Logic Test (PLT)
Tsai, M. (2002) _1 Computers & electronic Bulletin Board 6: 1&2 NO
System (BBS)
Tsai, M. (2002) _2 Computers & (BBS) 6: 1&2 NO
Tsai, M. (2002) _3 Computers & (BBS) 6: 1&2 NO
Tsai, M. (2002) _4 Computers & (BBS) 6: 1&2 NO
Wenk, M. et al. (2008) Mannequin (Simulator) 3 YES
Zumbach, J. et al. (2004) Interactive MS PowerPoint & PPT 6: 3&5 YES

Legend: Technology use: 1 = Communication/interaction, 2 = Cognitive support (distributed cognition, e.g. Excel, Word, SPSS),
3 = Cognitive support (deep understanding - e.g., simulations, knowledge creation), 4 = Informational resources
5 = Presentation, 6 = A mixture of max two (should be really two major purposes where one cannot be singled out - e.g., 6: 2&5)
N/A = 999 Missing information.
Link between collaboration and technology: Yes, and NO

27
 The moderator analyses of the grade level, subject domain, and treatment duration were

conducted to investigate their impact on student achievement resulting from instructional

interventions in question. As for grade level, the included studies targeted all the grade levels

(from kindergarten to post-secondary). The undergraduate level was the most frequent with ten

studies forming 50% of the total collection. The second highest was high school representing

four studies (20%). It is worth noting that while calculating the individual grade levels, the entire

collection of 20 studies was considered. Table 6 shows the grade level distribution.

Table 6. Frequency distribution of grade level addressed in the studies

Grade level (s) Frequency Relative %

Kindergarten 0 0.00
Elementary (2) 3 15.00
Middle school (3) 3 15.00
High School (4) 4 20.00
Undergraduate (5) 10 50.00
Graduate (6) 0 0.00
Total 20 100

Legend: Elementary = 2, Middle school = 3, High school = 4, Undergraduate = 5, Graduate = 6

Regarding the subject matter (Table 7), two categories were formd – STEM and Non-
STEM, encompassing a large variability of individual disciplines. Twelve studies related to
STEM represented 60% of the population. The Non-STEM category included eight studies
forming 40% of the total collection. STEM included the subjects in the domains of science,
math, technology, and engineering, while Non-STEM comprised subject categories related to
humanities, social sciences, languages, and arts.

28
Table 7. Frequency distribution of subject matter addressed in the studies

Subject domain (s) Frequency Relative %

STEM 12 60.00
Non-STEM 8 40.00
Total 20 100

STEM = Science, Technology, Engineering and Math,

Non-STEM = Social sciences & Humanities etc.

Effect Size Extraction and Synthesis

Procedure for calculating ES

For the ES calculation, the author used the Cohen`s d metric (1988), based on the
division of the mean differences by the pooled standard deviations of both groups. The equations
and formulas used for the calculation of ES can be found in Table 8. The information for ES
calculation was gathered from means, precisely reported standard deviations, and sample sizes
for the experimental and control conditions (Shymansky & Woodworth, 1989). To correct for
small sample bias, d was converted to the unbiased estimator g (Hedges & Olkin, 1985). In the
case of non-availability of the descriptive statistics, the ES was extracted from inferential
statistics, such as t-tests, F-tests, or exact p-values, using conversion equations from Glass,
McGaw & Smith (1981), and Hedges, Shymansky & Woodworth (1989).

Additionally, Cohen`s (1988) benchmark was used for the qualitative assessment of the
magnitude of an ES. It states three types of magnitude of an ES: (a) d ≥ 0.20 ≤ 0.50 = small
effect; (b) d > 0.50 ≤ .080 = medium effect and (c) d > 0.80 is called a large effect. However,
Valentine & Cooper (2003) warned against this type of fallacy saying that in the domain like
education even smaller ES can be considered effective

29
Table 8. Study- level statistics used in meta-analysis and explanations (adapted from Bernard et al. 2014)

30
 Synthesis of ES. An analytical approach of the random effects model (Borenstein et al.,
2009; Hedges & Olkin, 1985) was chosen for this meta-analysis. In the random effects model,
effect sizes are weighted by the inverse of the sum of their within-study variance (Vi) and
average between-study variance (tau-squared). This resulted in no between-study variance left
unaccounted for after the analysis is performed. The random effects model was used to interpret
and report average effect sizes (g+), standard errors (SE), confidence intervals (lower 95th and
upper 95th) and z values with relative p-values. In addition, a fixed effect model was used to
estimate total between-study variability (Q-Total) and test for heterogeneity. I2 (i.e., the
percentage of heterogeneity in effect sizes exceeding chance sampling expectations, e.g. Higgins,
Thompson, Deeks, & Altman, 2003). A total of 28 effect sizes were extracted from the twenty
studies. Four studies, namely, Faro, & Swan (2006) ; Freeman, O'connor, Parks, Cunningham,
Hurley & Haak et al. (2007) ; Gersten, Baker, Smith-Johnson, Dimino, and Peterson (2006) ;
Lin, Wu & Liu (1999), and Tsai (2002) produced more than one independent effect size.

31
Results

Sample size. The current meta-analysis is a part of the ongoing larger meta-analysis
conducting by Bernard et al. (2016). A collection of 78 studies was considered for the current
project. Only empirical studies that address collaboration/cooperation either in face-to-face, real-
life classroom and/or virtual learning/online such as computer-based collaborative learning are
included. A set of inclusion criteria below presents the study characteristics used to retain the
studies for the meta-analysis. The included studies needed to be published no earlier than 1996
and be publically available (or archived), and necessarily feature some form of student
collaborative work as the major instructional variable. Moreover, they must have contained at
least one between-group comparison where one group is considered the experimental condition
(i.e., higher level of collaboration) and the other group the control condition (i.e., lower/ no
collaboration) and contains sufficient statistical information for effect size extraction.

The primary abstract reviews resulted in twenty-two eligible studies according to the set
criterion (see criterion for inclusion/exclusion in the methods section). After this, reviews of the
full-text was conducted of the selected studies to ensure compliance with all of the project
inclusion criteria. A close analysis of the studies reported two studies (Mastropieri et al., (2006)
and Terwel et al., (2009) as ineligible given missing data, and they were therefore removed from
the collection of the final analysis. A total of twenty-eight effect sizes were extracted from
twenty studies involving 2434 participants. Two groups – collaborative learning environment
with high collaboration versus traditional instruction setting with less or no collaboration – were
compared for the relative effect on student achievement. The twenty studies entailed a variety of
collaborative activities to measure their impact on student learning outcomes. Table 9 presents
the list of included studies in this meta-analysis with the title for each.

32
Table 9. Studies included in the meta-analysis with titles

Reference Title (s)

Arts, J. A. et al. Enhancing problem-solving expertise by means of an authentic,
(2006) collaborative, computer supported and problem-based course
Barnes, L. J. (2008) Lecture-Free High School Biology Using an AUDIENCE RESPONSE
SYSTEM
Cavalier, J. & Klein, Using Cooperative Learning and Objectives with Computer-Based
J. D. (1998) Instruction
DePriter, T. N. Individual or collaborative learning: An investigation of teaching
(2008) strategy in the distance learning mathematics classroom
Faro, S., & Swan, An investigation into the efficacy of the studio model at the high school
K. (2006) level
Freeman, S. et al. Prescribed Active Learning Increases Performance in Introductory
(2007) Biology
Gersten, R. et al. Eyes on the Prize: Teaching Complex Historical Content to Middle
(2006) School Students with Learning Disabilities
Hernández-Ramos, Learning History in Middle School by Designing Multimedia in a
P. et al. (2009) Project-Based Learning Experience
Hodges, T. L. Examination of gaming in nursing education and the effects on learning
(2008) and retention
Hoon, T. S. et al. Effect of an Interactive Courseware in the Learning of Matrices
(2010)
Hosal‐Akman & An assessment of the effects of teaching methods on academic
Simga-Mugan performance of students’ in
(2010) accounting courses
Hummel, H. G. et Effects of cueing and collaboration on the acquisition of complex legal
al. (2006) skills
Lin, J. M., Wu, C., Using SimCPU in cooperative learning laboratories
& Liu, H. (1999)
Nugent, G. et al. The Impact of a Field-Based, Inquiry-Focused Model of Instruction on
(2008) Preservice Teachers’ Science Learning and Attitudes
Olgun, O. S., & Teaching Grade 5 life Science with a Case Study Approach
Adali, B. (2008)

33
 Pariseau, S. E. et al. The Effect of Using Case Studies in Business Statistics
(2007)
Priebe, R. (1997) The Effects of Cooperative Learning in a Second-Semester University
Computer Science Course.
Tsai, M. (2002) Do male students often perform better than female students when
learning computers? A study of Taiwanese eighth graders’ computer
education through strategic and cooperative learning
Wenk, M. et al. Simulation-based medical education is no better than problem-based
(2008) discussions and induces misjudgment in self-assessment
Zumbach, J. et al. Using multimedia to enhance problem-based learning in elementary
(2004) school

Inter-rater reliability. Two trained raters were involved in the reviewing and coding of
studies throughout the entire process of this meta-analysis. Following is the agreement rates
regarding each stage:

 Study features coding – 81% (k = 0.62)

 Effect size calculation (for accuracy of data extraction, selection and application of
equations) – 96.06%, (k = 0.92).

 Effect size comparison decisions (for defining the treatment and control conditions and
number of ES and data sources to use) – 91.66% (k = 0.83)

Achievement Outcomes. In response to the research question, “Does collaborative

learning have any statistically significant effect on student achievements outcomes?” a

statistically significant random weighted effect size of g+ = 0.266 was found. It is a low-to-

moderate positive effect size according to Cohen`s standards (For detail see Cohen, 1988).
Comprehensive Meta-AnalysisTM software (Borenstein et al. 2005) was used to carry out
analyses and derive the outcomes. Main results are presented in Table 10. The summary statistics

are based on k = 28. It shows an average effect size with both fixed and random effects models.

The table depicts the lower limit, and upper limits of the CI (0.09 and 0.44 respectively) and the
z-value along with the two-tailed probability is 2.93. The fixed weighted average effect of g+ =

34
0.281 which is also low to moderately low average effect size according to Cohen`s standards.
Overall the weighted average effect sizes for both random and fixed models are significant with
p < 0.001. The heterogeneity statistics as shown for the fixed effect model, Q-Total = 99.67, p <
.001 and I2 = 72.91% which reports that almost 73% variability is due to real differences in the
effect size and only 27% can be attributed to the sampling error. The Q- value statistics validates
that the distribution is significantly heterogeneous. This magnitude of between-study variability,
according to Higgins, Thompson, Deeks, and Altman (2003) is moderately high.

Table 10. Overall weighted average random effects and fixed effects sizes & heterogeneity

Analytical Models 95th Confidence interval

K g+ SE Lower limit Upper limit
Random Effects 28 0.266* 0.09 0.09 0.44
Model
Fixed Effects Model 28 0.281** 0.05 0.19 0.37

Heterogeneity Q-total = 99.67, I2 = 72.91% Ʈ2 = 0.16

df = 27, p < .001
Z* =2.93, p < .001; Z** = 6.17, p < .001

Research design and methodological quality. It was necessary to ensure that the
methodological quality of the included studies did not substantially affect the outcomes of this
meta-analysis. The first thing was to ascertain if there was any difference in primary studies’
research design that might have favored one category of study methodological quality over
others. With this intention, research design of each study was reviewed. The study designs of the
included studies in this meta-analysis were of a true experimental (randomized control trials)
nature representing 25% of total effect sizes (the gold standard) and quasi-experimental (adjusted
and refined by various means of statistical control) representing 75% of the total effect sizes. For
the quality methodological checks, the author used Valentine and Cooper`s (2008) instrument
called The Study Design and Implementation Assessment Device (Study DIAD). The moderator

35
variable analysis comparing two types of research design (RCT vs. QE) was not statistically
significant (Q-value = 3.47, p = .063). Though the p value (p = 0.63) was quite close to affect the
design, the study research design did not differentially bias the findings of the meta-analysis.
The second concern was to verify if the psychometric quality and representativeness of
the assessment tool had not affected the outcomes of the meta-analysis. For the analysis, two
major measure types – single cumulative and calculated average of several complementary
measures – were used. In this regard, the mixed effects analysis by measure source was
conducted. The results showed no statistical significance (p = .152 with one study removed)
regarding any effect on overall effect size. One selected single measure which did not belong to
either category of measure type was removed. This did not affect any significant change in the
outcomes (Q-value = 2.05). Similarly, the effect size extraction procedure (p = .562 with one
study removed) proved non-significant. The ES extraction involved two procedures – ES
precisely calculated from reported descriptive or inferential statistics and ES estimated with
various reasonable assumptions.

Furthermore, there might have been other factors that could have affected the outcomes
given the methodological quality of the analysis (e.g., instructor’s equivalence, the equivalence
of content materials). Considering these factors, the author collapsed the several aspects of the
methodological qualities (including previously described individually tested qualities of research
design, assessment instruments, and extraction procedures) of the studies to enable a composite
analysis. The mixed effects analysis of this composite also reported a non-significant effect Q-
value = 0.24 (p = .624).

Publication Bias Analysis. According to Rothstein, Sutton and Borenstein (2005), an

analysis of publication bias helps to ensure if a sizeable number of studies might have been
missed which could have otherwise been included in the analysis. Given the inclusion of these
missing studies and considering the fact of accommodating them with more positive findings by
journals (Polanin, Tanner-Smith, & Hennessy, 2015), some statistical procedures became
warranted to examine if there is any reason to question the robustness of the average effect size.

36
In this case, the author used the funnel plot for the visual inspection and two statistical
procedures – classic fail-safe analysis (for nullifying the average effect) and Orwin`s fail-safe
(for the trivial effect of magnitude) – to verify the publication bias for the current meta-analysis.
The visual inspection of the funnel plot depicted in Fig. 1, the Funnel plot, showed the
symmetrical dispersion of effect sizes around the mean of the distribution (g+ = 0.266). The
following analytical statement about publication bias analysis appears in Comprehensive Meta-
AnalysisTM.

This meta-analysis incorporates data from 28 studies, which yield a z-value of 5.72 and

corresponding 2-tailed p-value of 0.000. The fail-safe N is 212. This means that we would

need to locate and include 212 'null' studies for the combined 2-tailed p-value to exceed

0.050. The Orwin fail-safe N is 51. This means that we would need to locate 51 studies with

mean Hedges' g of 0 to bring the combined Hedges' g under 0.1. The Trim and Fill (Duval

and Tweedie, 2004) analysis also reported the same pattern of inclusiveness. Using these

parameters, the method suggests that no studies are missing. Under the fixed effect model the

point estimate and 95% confidence interval for the combined studies is 0.28123 (Lower 95th

= 0.19184, Upper 95th = 0.37061). Under the random effects model the point estimate and

95% confidence interval for the combined studies is 0.26566 (Lower 95th = 0.08815, Upper

95th = 0.44317). Using Trim and Fill these values are unchanged (p = 0.000).

The author judged and concluded that there was no publication bias present which might impact
the effect size adversely resulting in skewness or any negotiation with the results.

37
Figure 1. Funnel plot with effect size for using random model

38
 Sensitivity Analysis. The sensitivity analysis aimed to eschew the distorting effects (both
overall mean and variability) due to the presence of any outlier in the included studies. The
author encountered one study (Hernández-Ramos, & Paz, 2009) with a comparatively higher
positive ES (g = 2.534). The reason for this outlier was not known given the missing information
in the study. Therefore, the author reduced the magnitude of this aberrant effect size to the
second highest effect size, within the range of other large ES by using Comprehensive Meta-
AnalysisTM. This adjustment of ES resulted in g = 1.409 which was in the range of other effect
sizes in the distribution. After the outlier adjustment, the newly calculated averages fall within
the confidence interval of the total collection g+ = 0.266 (k = 28, SE = 0.09, Lower 95th = 0.09
and Upper 95th 0.44). After this, the data were found quite stable and unaffected by any outliers
for the further analysis.

Figure 2. depicts the Forest plot with individual and overall ESs for the included studies.
On the left side of the figure are the study identifications, in this case, the author names. In the
center are the study-level statistics for the twenty-eight ES : Hedges g, the standard error, the
variance, the upper and lower boundaries of the 95th confidence interval, the z value, and its
relative probability (p-value). A visual representation called a Forest plot is on the right side of
the figure. The ES for each study are depicted in the shape of squares. The lines around squares
show the width of the 95th confidence interval for each study. The z-test of these effect sizes was
significant (p > .05). Furthermore, the smaller dots represent, the lower leverage effect sizes (i.e.,
smaller contributors to the weighted averages ES ), while larger dots demonstrate the higher
leverage effects characterized by larger sample size.

39
Figure 2. Forest plot of 28 effect size from the distribution (k =20) and study-level statistics

40
 Table 11. shows the mixed effects analysis by treatment (i.e., Collaboration) strength. It
relates the degree of difference between two groups in collaboration. It clearly depicts that
average effect size with a high degree of difference in collaboration (k = 9) with g = 0.35 which
is significantly different from zero and average effect size g = 0.29 (k = 12) that is moderately
different while an average effect size g = 0.08 (k = 7) which is at a low level of difference in
collaborative work. Q-Between = 2.46 was not statistically significant (p = .293). This analysis
shows an increasing trend in effect size magnitude with the difference in collaboration in two
conditions across groups.

Table 11. Mixed-effects analysis of collaboration strength

Level of Effect size 95th Confidence interval

Collaboration
k g+ SE Lower limit Upper limit
Low 7 0.082 0.10 -0.12 0.28

Moderate 12 0.299 0.15 0.00 0.60

High 9 0.357 0.18 0.00 0.71

Q-Between = 2.46 (df = 2, p = 0.293)

To further explain the detected variability in g+, four moderators – technology (both
regarding its relevance to collaborative work and its functionality), subject matter, grade levels,
and treatment duration – were analyzed. The moderator analysis of technology types was
conducted using the mixed model analysis (Borenstein et al., 2009). Two qualitative categories
about the added instrumental value were created – Yes and No. Yes (k = 22) favored the
instrumental value of technology in collaboration (i.e., technology was an important factor in
enabling and supporting collaborative learning activities), while No category (k = 6) (merely
reflected some contextual presence of technology). However, an upward trend for the
instrumental value of technology was discernible in the analysis (see Table 12). The moderator

41
analysis of the instrumental value of technology in collaboration was not statistically significant
(Q-Between = 0.34, p = .558). This trend may inform that though technology had added value in
collaboration, it was not statistically significant to affect the overall effect.

Table 12. Mixed effects analysis by the instrumental value of technology

95th Confidence Interval

Groups Number g+ SE Lower limit Upper limit Q-Between
Studies
No 6 0.17 0.17 -0.16 0.51

Yes 22 0.29 0.10 0.08 0.50

Overall 28 0.26 0.09 0.08 0.43

Between groups df = 1 0.34
z = 2.87 p = 0.558
Legend: No = No use of technology in collaboration,
Yes = Use of technology in collaboration

To dig deeper into the matter, another analysis for the types of cognitive support provided
by technology was conducted (see table 13). By cognitive support, the author means the use of
technology for two types of supports – cognitive support for distributed cognition (e.g. Word,
Excel, and SPSS) and cognitive support for deep learning/understanding ( e.g. simulations and
knowledge creation). The Q-Between = 0.17 with p = 0.681 was not statistically significant for
the type of cognitive support. When tested for the presence of technology for cognitive support
(distributed and deep learning), the trend refers that k = 17 did not favor both distributed and
deep learning and only k =10 favored the value of technology for both types of cognitive support.

42
Table 13. Mixed-effects analysis of cognitive support for Distributed Cognition & Deep
Learning

95th Confidence Interval

Groups Number g+ SE Lower Upper Q-Between

Studies limit limit
No 10 0.31 0.16 0.00 0.63
Yes 17 0.23 0.12 0.00 0.47
Overall 27* 0.26 0.10 0.08 0.45
Between df = 1 0.17
groups, z = 2.76 p = 0.681

* ES 27 since one study was excluded from this analysis given the missing information (999)
Legend: No = No cognitive support of the use of technology in collaboration
Yes = Cognitive support offered by the use of technology in collaboration

Next, the author decided to explore the use of technology in collaboration for cognitive
support for deep learning only. To perform the analysis, the frequency were calculated. This
analysis was also not statistically significant (Q-Between = 0.02, p = 0.893). Table 14
demonstrates the use of technology for the cognitive support for deep learning only.

Table 14. Mixed effects analysis using technology for the cognitive support for deep learning

95th Confidence Interval

Groups Number Studies g+ SE Lower Upper Q-Between
limit limit
No 15 0.25 0.11 0.03 0.47
Yes 12 0.27 0.16 -0.03 0.58
Overall 27* 0.26 0.09 0.08 0.44
Between df = 1 0.02
groups, z = 2.81 p = 0.89

* ES 27 since one study was excluded from this analysis given the missing information
Legend: No = No cognitive support, Yes = Cognitive support offered for deep learning

43
 Furthermore, the author investigated if the subject domains had any moderating effects
on the overall effect of collaboration. Table 15 shows that there was no difference in learning
across STEM and Non-STEM subjects. Collaborative learning is equally effective across all
subject domains. The Q-between = 0.00 with p = 0.992 inferred a statistically non-significant
effect of subject domains. This result informed that collaboration is not limited only to STEM
domains. Table 15 portrays an equal effect of collaboration across Non-STEM (g = 0.26 with k =
10) and STEM ( g = 0.26 with k = 18) subject domains. Notwithstanding this outcome, it is
important to note the the classification of a course being STEM versus non-STEM is
problematic, so any interpretation of this result should be qualified.

Table 15. Mixed effects analysis by moderator variable subject matter

95th Confidence Interval

Category Number g+ SE Lower Upper Q-Between

Studies limit limit
Non STEM 10 0.26 0.18 -0.09 0.62
STEM 18 0.26 0.10 0.06 0.46
Overall 28 0.26 0.09 0.08 0.44
Between df = 1 0.00
groups, z = 2.92 p = 0.992

Non-STEM = Social Sciences, Humanities and STEM: Science, Technology, Engineering and Math

Furthermore, the author tried to explore the effect of duration on collaboration. Three

groups of duration were formulated to measure the impact of duration on overall effect size. As

Table 16 depicts the effect of duration on collaborative learning was not statistically significant

44
(Q-between = 3.27, p = 0.19). The In Between group ( k = 14) produced showed the highest

impact of collaboration (g = 0.43) in comparison to short (k = 3) and semester length (k = 11).

Table 16. Mixed effects analysis by treatment duration

95th Confidence Interval

Category Number Studies g+ SE Lower Upper Q-Between

limit limit
In Between 14 0.43 0.17 0.10 0.76
Semester 11 0.09 0.07 -0.04 0.23
Short 3 0.13 0.12 -0.09 0.36

Overall 28 0.14 0.06 0.03 0.25 3.27

Between df = 2 p = 0.195
groups, z = 2.55

Legend: More than three days but less than eight weeks = In Between
Nine weeks or more = semester, Three days or less = short

Next, the author had committed to conducting a moderator analysis of grade level to
investigate its effect on collaboration. The analysis included grades from Elementary school to
Undergraduate levels. There were no eligible studies found from Kindergarten and Graduate
levels in the collection according to the inclusion and exclusion criteria. Even though the
variation in grade level of education significantly differentiated student achievement outcomes
(Q-Between = 11.18, p = 0.011), the small k sizes (high school - k = 6; elementary k = 3; and
middle school - k = 5) were such that any further interpretation of this outcomes was abamdoned
due to lack of statistical power.

45
 Chapter Five : Discussion

This section involves the interpretations of the evidence in the form of discussion about
the results obtained. The author here discusses the results to inform the three research questions
which guided this meta-analysis and to underline any possible conceptual, theoretical and/or
practical implications of the findings. This discussion is orchestrated in the backdrop of the
literature of the domain and in the light of previous studies.

The purpose of the current meta-analysis was to measure the effect of collaborative
learning on student achievement in the context of formal education across multiple subject
domains and grade levels. As mentioned earlier, it included a total of 28 effect sizes from a
collection of 20 studies. The main research question was: Does collaborative learning have any
statistically significant effect on student achievement outcomes when compared learning without
(or under lesser degree of) collaboration? The most explicit statement drawn from the analysis is
that the effect of collaborative learning on student achievement is positive but low, though
significantly greater than zero. Regarding percentile difference (i.e. U3 minus 50%), 60% of
students yielded an increase in scores, or a person with an average (50th percentile) could expect
to move to the 60th percentile after participating in a collaborative learning group. The average
effect size of g+ = 0.266 (k = 28) is a little higher than the low category in terms of Cohen’s
(1988) qualitative effect size magnitude (i.e. g+ > 0.20 < 0.50). However, given the considerable
degree of heterogeneity among studies, it was difficult to fix the exact location of the population
mean aside from the probability that it is located between g+ = 0.09 and 0.44 (i.e. lower and
upper levels of 95th confidence interval respectively).

The next few paragraphs discuss the findings in the light of previous studies undertaken
to measure the effects of collaborative/cooperative learning on student achievement . The
findings of the current study are consistent with Johnson et al. (1988, 1998) who found high
positive effects g+ = 0.78 and g+ = 0.49 ( k = 122 and k = 168 respectively) of
cooperative/competitive conditions on student achievement. In comparison to the Bowen’s study

46
(2000) that found a statistically significant effect (g+ = 0.51, k = 37) of cooperative learning on
student achievement in high school and college chemistry courses, the current study relates the
same positive effect i.e. findings are consistent. Springer’s (1999) findings suggested that various
forms of small-group learning are effective in promoting greater academic achievement in case
of STEM courses and reported a statistically significant effect of small-group learning on student
achievement (g+ = 0.51, k = 37). Here, again the results of the current study are in line with
Springer’s findings. Further, the findings of the current analysis are consistent with Lou et al.
(2001) who found that small group learning had significantly more positive effects than
individual learning on student individual achievement (g+ = 0.15, k =122). Therefore, this
consistent positive trend indicates that collaborative learning helps enhance student achievement
considerably.

Comparatively, the average effect size of the current study (g+ = 0.26 ) is in the middle
of achievement effect of g+ = 0.17 for Lou et al. (1996) and Bernard et al. (2009) which is g+ =
0.38. This difference may be accounted for by the number of studies and the variables
incorporated in those meta-analyses. For example, Bernard et al. (2009) were interested in the
three types of interactional conditions in distance education. However, the current study aimed at
the investigation of the effect of collaborative learning in all forms of instruction in multiple
subject levels across all grade levels. Also worth noting are the findings of Borokhovski et al.
(2016) on the use of collaborative activities to support student-student interaction in a technology
rich environment, i.e., g+ = 0.52 (k = 25) is much higher in magnitude. The possible explanation
for this difference may be again be the number of studies included with multiple collaborative
conditions, use and purpose of technology in instructional design, grade levels, and the primary
research questions asked.

The second research question was: Do different types of technology have varying effects
in collaborative activities when used to enhance/support/promote collaboration? A study by
Schmid et al. (2014) reported a statistically significant effect of technology for cognitive support.
However, the findings of the current study are not consistent with Schmid et al. (2014). The use

47
of technology for collaborative activities did not favor cognitive support in distributed cognition
and deep learning, collectively. This difference in the findings could be attributed to the
difference in the use and degree of technology in classroom. For example, in the current study,
technology as moderator was analyzed to measure its effect on collaborative learning rather than
the effect of technology on student achievement. In addition, small k-sizes and varied treatment
conditions might have resulted in different outcomes in both of the studies.

Interestingly, the use of technology for deep learning was also not statistically significant
in the current study. On the contrary, Tamim, Bernard, Borokhovski, Abrami, & Schmid, (2011),
found an average effect size g+ = 0.28 (k = 25) while investigating the effect of technology on
student achievement. They tried to map out the exclusive effect of technology on student
achievement. However, unlike both Tamim et al., (2011), and Schmid et al., (2009, 2014), the
current study aimed at the exploration of technology as a moderator to enhance/support/promote
collaborative learning affecting student achievement . The possible explanation for the different
findings in these studies could be attributed to the insufficient sample size, lack of training for
both teachers and students, and use of technology for secondary purposes other than the
collaborative activities.

In some studies, technology was used for cognitive support in the treatment condition.
For example, Wenk, Waurick, Schotes, Gerdes, Aken, & Pöpping, (2008) used an electronic
mannequin ( a life size simulator ) to enhance deep learning on the processes of medical
treatment. Similarly, in other studies ( Hoon et al., 2010 ; Hernández-Ramos et al., 2009; Tsai,
2002; Priebe, 1997; Freeman et al., 2007; Faro et al., 2006) technology was used to support the
deep learning and for information resources.

In the context of the current findings, it would be interesting to reflect on the ongoing
great debate between Clark (1983, 1994) and Kozma (1994). What Clark (1983, 1994, 2009)
claims is that role of technology in learning is minimal (or negligible ). Instead, it is the nature of
pedagogy (for example deep learning) and learning design (learning environment) that matter in
teaching and learning process irrespective of any mode of technology use. Kozma claims that

48
technology is helpful for learners to remember, seek information, and to collaborate. However,
these claims are too big to generalize from the findings of current studies given the lack the
statistical power.

The third research question was: Do grade levels, and subject domains have any
moderating effects on student achievement? As noted above, the lack of statistical power for this
analysis resulted in abandoning any further discussion that might misinform the literature.
Hence, there is a need for further exploration to understand what factors influence collaboration
at various grade levels.
Regarding the subject domain, there was no significant difference found between STEM
and Non-STEM comparisons groups. Students achieved almost equally in both domains.
Therefore, the findings relate that collaborative activities can impact student learning outcomes
across all subjects as opposed to the findings of Lou et al., (1996) and Qin et al., (1995), who
reported that STEM influenced more significantly student achievement. The rationale for the
outcome of current study is that collaborative activities are employed almost equally to enhance
interactions among learners across STEM and Non-STEM subject domains. Therefore,
collaborative activities serve as a means to create learning environment rather than as an end to
maximize any subject specific content.

Further, the author decided to investigate if the duration of treatment had any impact on
the determination of the degree of collaboration. The findings of the current analysis indicated
that duration as moderator was not statistically significant which differed from the findings of
other studies. In the between-study comparison group, however, there was an indication that
students liked moderate duration (In Between) for collaborative activities. This outcome is
consistent with what Fisher (1981) says that students’ interest and choice of the content may
determine their inclination toward specific duration. In this regard, the construct of academic
learning time is important to predict student achievement. It is so because, for example, allocated
time, engagement rate, and success rate of school activities are all associated with student

49
achievement. This signifies that more academic learning time can be interpreted as helpful to an
ongoing measure of student learning.

Limitations of the Study

This study has encountered some general and specific limitations. First, only a small
number of the studies qualified for the final sample meant for the analysis. Here is the biggest
limitation of this study because the number of samples included (k) were very low.
Consequently, the low k affects the power of the study to find significant effects, especially in
moderator variable analysis where the total number of samples is split by the number of levels of
the moderator variable. Therefore, the generalizability of the findings of this study cannot be
established given this small sample.

Second, the common critique on meta-analyse is the heterogeneity of the included

studies; each study is often conducted in different conditions (for example different treatment,
different variables, and a diverse population of participants). However, the results of a meta-
analysis are combined as if they were similar (Eysenck, 1994). In this situation, it is not always
easy to compare the results included in a reliable manner and the interpretation may be biased.
As Higgins (2008) says, the heterogeneity between studies is to be expected and is acceptable
‘‘providing both that the predefined eligibility criteria for the meta-analysis are sound and that
the data are correct’’ (p. 1158). The heterogeneity can be attributed to the variability found
between studies regarding the participants, interventions effects and outcomes studied (diversity
and statistical heterogeneity). This heterogeneity affects the overall average effect size of an
analysis. In the current analysis , the author encountered one study (Hernández-Ramos, & Paz,
2009) with a comparatively higher positive ES (g = 2.534) which was adjusted to make the
distribution more representative.

Third, the study might have publication bias regarding the exclusion of studies published
in the languages other than English. In addition, the accommodation of the studies with more

50
positive findings in journals could have affected the representativeness of the sample (Polanin,
Tanner-Smith & Hennessy, 2015).

Implications and Future Directions

The primary purpose of the current meta-analysis was to explore the effect of
collaborative learning on student achievement. In general, the collaborative learning was found
favourable to enhance student achievement. The analysis reported some implications as to how
collaborative activities with what combination of conditions yield better learning outcomes for
students. In collaboratively designed instruction, students outperformed their control
counterparts. Collaborative learning activities are beneficial in that these help enhance student
achievement and persistence, (e.g. Bowen, 2000) change attitude and self-concept, and support
those students who feel fearful while participating in classroom activities (e.g. Kyndt et al.
2013).

Next ,the differential use of technology in collaborative activities was not found to
matter. Though technology was used in both groups for various purposes, the impact of different
technology on student achievement was not significant. This outcome may provide opportunities
for future researchers to explore the questions of what types of technology and what contexts
help teachers design collaborative activities which can enhance student achievement.

Among the other future implications include, firstly, the cultural differences among
learners impact their degree of collaboration considerably (Kyndt et al., 2013). Culture, as a
moderator which has not been explored in this analysis, may be added for future exploration. It is
so because the exploration of culture as moderator will add more insights into the factors which
are conducive/detrimental for collaborative learning. The cultural differences between
Western/individualistic and Eastern/collectivistic cultures may have significant influences on the
ways students cooperate in the learning activities. Studies have found that ‘‘cross-cultural studies
have shown that Northern and Western Europeans and North Americans tend to be
individualistic and that Chinese people, other Asians, Latin-Americans, and most East and West

51
Africans tend to be collectivists’’ (Cox, Lobel, & McLeod , 1991, p. 828). This means that the
Western approach of cooperative learning embraces critiquing opinions by challenging each
other’s reasoning and dealing with conflicts which may be culturally inappropriate for
collectivistic cultures. For example, Thanh et al. (2008) found that the incongruity between
cooperative learning principles and Asian culture accounted partially for the failure of
cooperative learning.

Secondly, the aspects such as the forms, contexts and purposes of any selected
collaborative activities and the roles of a teacher ( e.g. either facilitator, partner or observer)
during the process of collaboration are other major areas to guide how cooperative learning can
be used to enhance student achievement.

Thirdly, the investigation of the use of technology for secondary purposes will be helpful,
as these secondary purposes trigger students’ interests to participate in classroom activities. This
means that how students’ previous exposures to various forms of social technology such as
Facebook, Twitter, Instagram, or Snapchat may help them collaborate in their learning activities
(Phua, Jin, & Kim, 2017; Kim & Kim, 2017).

A fourth direction for future researchers may be to verify whether collaborative activities
are favourable among lower grade levels in comparison to higher educational levels. It would be
interesting to unpack the favourable conditions and types of tools which enhance participation
among post-secondary students for collaborative activities. An extensive study is warranted to
investigate the factors that influence collaboration at all levels.

Lastly, Bowen (2000) collected data on persistence from nine studies and found that
cooperative learning has a significant and positive effect on student attitudes towards STEM
courses. There is no clear cut definition of the construct of persistence. In the educational
context, a student persistence entails the capacity which allows students to continue his efforts
through self-regulation, motivation, and positive affirmation to the achieve the set goals. The
degree of persistence dictates the student achievement. Bowen’s (2000) meta-analysis reported

52
that persistence for continued study in STEM courses taught with cooperative learning
approaches was 22% greater than persistence of students taught by traditional approaches.
Students in cooperative learning classes also had more positive attitudes toward their classes
(p.11). Therefore, it would be interesting to investigate other personality characteristics such as
“persistence or fear” (e.g. Bowen, 2000) which affect student collaboration.

53
 Chapter Five : Conclusion

The overall effect of collaborative learning on student achievement was positive and
significant despite some limitations of the study. However, there are some moderators which
impact student achievement. The analysis found that collaborative activities organized at
different school levels may affect student achievement. Technology in its various forms are used
in classroom, however, these forms are integrated purposefully in instructional and curricular
designs (Borokhovski et al., 2016). Embeding technology in pedagogy may help improve student
collaboration and thereby their social, and academic success. Instead of incorporating technology
as a mere extension or ancillary in instructional designs, technology might be added in curricula
to improve student collaboration.

Furthermore, the subject domains and treatment duration are important to understand the
impact of collaborative learning on student achievement. STEM and Non-STEM subjects can be
taught equally successfully when one uses collaborative methodology. While duration dictates
the level of collaboration among learners, however, the reasons for favouring small duration over
longer duration depend on learners’ interests, teacher, available time, and nature of content.
Again, an understanding of learners’ needs and interests may help them choose their best options
in this regard.

On the whole, the findings of this meta-analysis are valuable for teacher-educators and
curriculum designers to take informed decisions on the conditions and forms of collaborative
activities to be included while planning, designing, and implementing effective instructional
strategies. For example, a teacher may employ collaborative activities based on arts, culture and
local issues related to science, environment and health to develop group projects. These exercises
will serve two purposes - knowledge creation and development of critical skills.

Further, these findings may guide teachers in making choices of types and purposes of
technology use in their classroom. For instance, the use of augmented and virtual reality can
help create an environment to simulate scientific and natural phenomena such as study of stars

54
and galaxies, earthquakes and earth. Findings on the other contextual factors such as grade levels
and subject domains along with duration of treatment will help inform subject experts and
researchers to improve student learning. Finally, these findings have added to the knowledge of
the domains and have replicated the results of the previous studies.

55
 References

Note: A leading asterisk (*) indicates the primary studies included for the meta-analysis.

Abrami, P. C., Bernard, R. M., Bures, E. M., Borokhovski, E., & Tamim, R. (2011). Interaction
in distance education and online learning: Using evidence and theory to improve practice.
Journal of Computing in Higher Education, 23(2/3), 82-103.

Adelskold, G., Alklett, K., Axelsson, R., & Blomgren, G. (1999). Problem-based distance
learning of energy issues via computer network. Distance Education, 20(1), 129-143.

Anderson, T. (2003). Getting the mix right again: an updated and theoretical rationale for
interaction. International Review of Research in Open and Distance Learning, 4(2), 9-14.

*Arts, J. A., Gijselaers, W. H., & Segers, M. S. (2006). Enhancing problem-solving expertise by
means of an authentic, collaborative, computer supported and problem-based course.
European Journal of Psychology of Education, 21(1), 71-90.

Arts, J. A. R., Gijselaers, W. H., & Segers, M. S. R. (2002). Cognitive effects of an authentic
computer-supported, problem-based learning environment. Instructional Science, 30,
465-495.

Baeplar, P., Walker, J.D., & Driessen, M. (2014). It’s not about seat time: Blanding, flipping,
and efficiency in active learning classrooms. Computers & Education, 78, 227-236.

Bangert-Drowns, R. L., & Rudner, L. M. (1991). Meta-Analysis in Educational Research. ERIC

Digest. Retrieved from ERIC database. (ED339748)

*Barnes, L. J. (2008). Lecture-free high school biology using an audience response system. The
American Biology Teacher, 70(9), 531-536. doi: 10.2307/27669338

Beldarrain, Y. (2006). Distance education trends: Integrating new technologies to foster student
interaction and collaboration. Distance Education, 27(2), 139-153.

56
Bernard, R. M., Abrami, P. C., Borokhovski, E., Wade, A., Tamim, R., Surkes, M., et al. (2009).
A meta-analysis of three interaction treatments in distance education. Review of
Educational Research, 79(3), 1243-1289.

Bernard, R. M., Abrami, P. C., Lou, Y., Borokhovski, E., Wade, A., Wozney, L., et al. (2004).
How does distance education compare to classroom instruction? A Meta-analysis of the
empirical literature. Review of Educational Research, 74(3), 379-439.

Bernard, R. M., Rojo de Rubalcava, B., & St-Pierre, D. (2000). Collaborative online distance
education: Issues for future practice and research. Distance Education, 21(2), 260-277.

Bernard, R. M., Borokhovski, E., Schmid, R. F., Tamim, R. M., & Abrami, P. C. (2014). A meta-
analysis of blended learning and technology use in higher education: From the general to
the applied. Journal of Computing in Higher Education, 26(1), 87-122.

Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. (2005). Comprehensive meta-
analysis. Englewood, NJ: Biostat.

Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). A basic introduction
to fixed-effect and random-effects models for meta-analysis. Research Synthesis
Methods, 1(2), 97-111.

Borokhovski, E., Bernard, R. M., Tamim, R. M., Schmid, R. F., & Sokolovskaya, A. (2016).
Technology-supported student interaction in post-secondary education: A meta-analysis
of designed versus contextual treatments. Computers & Education, 96, 15-28.

Borokhovski, E., Tamim, R. M., Bernard, R. M., Abrami, P. C., & Sokolovskaya, A. (2012). Are
contextual and design student-student interaction treatments equally effective in distance
education? A follow-up meta-analysis of comparative empirical studies. Distance
Education, 33(3), 311-329.

57
Bowen, C. W. (2000). A quantitative literature review of cooperative learning effects on high
school and college chemistry achievement. Journal of Chemical Education, 77(1), 116-
119.

Bruffee, K. (1995). Sharing our toys. Change, 27(1), 12-18.

*Cavalier, J. C., & Klein, J. D. (1998). Effects of cooperative versus individual learning and
orienting activities during computer-based instruction. Educational Technology Research
and Development, 33(3), 52-72.
Chou, S., & Min, H. (2009). The impact of media on collaborative learning in virtual settings:
The perspective of social construction. Computers & Education, 52(2), 417-431.

Clark, R. E. (1983). Reconsidering research on learning from media. Review of Educational

Research, 53, 445–449.

Clark, R. E. (1994). Media will never influence learning. Educational Technology Research and
Development, 42(2), 21–29.

Cox, T., Lobel, S., & McLeod, P. (1991). Effects of ethnic group cultural differences on
cooperative and competitive behaviour on a group task. Academy of Management
Journal, 34, 827-847. http://dx.doi.org/10.2307/256391.

Cobb, T. (1997). Cognitive efficiency: toward a revised theory of media. Educational

Technology Research and Development, 45(4), 21-35.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum.

Coombs, P. with Ahmed, M. (1974). Attacking Rural Poverty, Baltimore: The John Hopkins
University Press.

Cooper, J., Prescott, S., Cook, L., Smith, L., & Mueck, R. (1990). Cooperative learning and
college instruction. California: California State University.

58
Curtis, D. D., & Lawson, M. J. (2001). Exploring collaborative online learning. JALN, 5(1), 21-
34.

Dalton, D. W., Hannafin, M. J., & Hooper, S. (1989). The effects of individual versus
cooperative computer-assisted instruction on student performance and attitudes.
Educational Technology Research and Development, 37(2), 15-24.

Decuyper, S., Dochy, F., & Van den Bossche, P. (2010). Grasping the dynamic complexity of
team learning. An integrative systemic model for effective team learning. Educational
Research Review, 5, 111-133.

Dede, C. (1996). Emerging technologies and distributed learning. American Journal of Distance
Education, 10(2), 4-36.

Dede, C. (2004). Enabling distributed learning communities via emerging technologies. The
Journal, 32(2), 12-22.

*DePriter, T. N. (2008). Individual or Collaborative Learning: An Investigation of Teaching

Strategy in the Distance Learning Mathematics Classroom. ProQuest.

Dillenbourg, P. (1999). What do you mean by ‘collaborative learning?’ In P. Dillenbourg (Ed.),

Collaborative-learning: Cognitive and Computational Approaches (pp.1–19). Oxford:
Elsevier.

Dillenbourg, P., Baker, M., Blaye, A., & O’Malley, C. (1996). The evolution of research on
collaborative learning. In E. Spada & P. Reiman (Eds.), Learning in humans and
machine: Towards an interdisciplinary learning science (pp. 189-211). Oxford: Elsevier.

Dochy, F., Segers, M., Van den Bossche, P., & Gijbels, D. (2002). Effects of problem based
learning: A meta-analysis. Learning and Instruction,13, 533-568.

Duval, S. & Tweedie, R. (2000b). A nonparametric ‘‘trim and fill’’ method of accounting for
publication bias in meta-analysis. Journal of the American Statistical Association, 95, 89-
98.

59
Duval, S., & Tweedie, R. (2004). Trim and fill: A simple funnel-plot–based method of testing
and adjusting for publication bias in meta-analysis. Biometrics, 56(2), 455-463.

Eastmond, D. (1995). Alone but together: Adult distance study through computer conferencing.
Cresskill, NJ: Hampton Press.

Eysenck, H. J. (1994). Systematic reviews: Meta-analysis and its problems. British Medical
Journal, 309, 789-792.

*Faro, S., & Swan, K. (2006). An investigation into the efficacy of the studio model at the high
school level. Journal of Educational Computing Research, 35(1), 45-59.

Fisher, C. W. (1981). Teaching behaviors, academic learning time, and student achievement: An
overview. Journal of Classroom Interaction, 17(1), 2-15.

Fjermestad, J. (2004). An analysis of communication mode in-group support systems research.

Decision Support Systems, 37(2), 239-263.

*Freeman, S., O'connor, E., Parks, J. W., Cunningham, M., Hurley, D., Haak, D., . . . Wenderoth,
M. P. (2007). Prescribed active learning increases performance in introductory biology.
Cell Biology Education, 6(2), 132-139.

*Gersten, R., Baker, S. K., Smith-Johnson, J., Dimino, J., & Peterson, A. (2006). Eyes on the
prize: Teaching complex historical content to middle school students with learning
disabilities. Exceptional Children, 72(3), 264-280.

Glass, G.V, McGaw, B., & M.L. Smith (1981). Meta-analysis in social research. Beverly Hills,
CA: Sage.

Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. Orlando, FL: Academic
Press.

Hedges, L. V., Shymansky, J. A., & Woodworth, G. (1989). A practical guide to modern
methods of meta-analysis. Washington, DC: National Science Teachers Association.

60
*Hernández-Ramos, P., & Paz, S. D. (2009). Learning history in middle school by designing
multimedia in a project-based learning experience. Journal of Research on Technology in
Education, 42(2), 151-173.

Henrie, C. R., Halverson, L. R., & Graham, C. R. (2015). Measuring student engagement in
technology-mediated learning: A review. Computers & Education, 90, 36-53.

Higgins, J. P. (2008). Commentary: Heterogeneity in meta-analysis should be expected and

appropriately quantified. Journal of Epidemology, 37, 1158-1160.

Higgins, J. P. T., Thompson, S. G., Deeks, J., & Altman, D. G. (2003). Measuring inconsistency
in meta-analyses. British Medical Journal, 327, 557–560.

*Hodges, T. L. (2008). Examination of gaming in Nursing Education and the effects on learning
and retention. ProQuest.

Hooper, S. (1992). Cooperative learning and computer-based instruction. Educational

Technology Research and Development, 40(3), 21-38.

Hooper, S., & Hannafin, M. J. (1991). The effects of group composition on achievement,
interaction, and learning efficiency during computer-based cooperative instruction.
Educational Technology Research and Development, 39(3), 27-40.

*Hoon, T. S., Chong, T. S., & Ngah, N. A. B. (2010). Effect of an Interactive Courseware in the
Learning of Matrices. Educational Technology & Society, 13(1), 121-132.

* Hosal‐Akman, N., & Simga‐Mugan, C. (2010). An assessment of the effects of teaching

methods on academic performance of students in accounting courses. Innovations in
Education and Teaching International, 47(3), 251-260.
doi:10.1080/14703297.2010.498176

Hoskins, S. L., & van Hoof, J. C. (2005). Motivation and ability: Which students use online
learning and what influence does it have on their achievement. British Journal of
Educational Technology, 36(2), 177-192.

61
*Hummel, H. G., Paas, F., & Koper, R. (2006). Effects of cueing and collaboration on the
acquisition of complex legal skills. British Journal of Educational Psychology, 76(3),
613-631.

Hunter, J.E., & F.L. Schmidt (1990). Methods of meta-analysis. Newbury Park, CA: Sage.

Jehng, J. C. J. (1997). The psycho-social processes and cognitive effects of peer-based

collaborative interactions with computers. Journal of Educational Computing Research, 17,
19-46.
Johnson, D. W., & Johnson, R. (1989). Cooperation and competition: Theory and research.
Edina, MN: Interaction Book Company.

Johnson, D. W., & Johnson, R. T. (1996). Cooperative learning and traditional American values:
An appreciation. NASSP Bulletin, 80(579), 63-66.

Johnson, D. W., & Johnson, R. T. (2009). An educational psychology success story: Social
interdependence theory and cooperative learning. Educational Researcher, 38, 365-379.

Johnson, D. W., Johnson, R. T., & Smith, K. A. (1998). Cooperative learning returns to college:
What evidence is there that is works? Change, 30, 26–35.

Jonassen, D. H., & Kwon, H. I. (2001). Communication patterns in computer-mediated and face-
to-face group problem-solving. Educational Technology Research and Development, 49,
35-51.

Johnson, D. W., Maruyama, G., Johnson, R. T., Nelson, D., & Skon, L. (1981). Effects of
cooperative, competitive and individualistic goal structures on achievement: A meta-
analysis. Psychological Bulletin, 89, 47–62.

Jonassen, D., Prevish, T., Christy, D., & Stavulaki, E. (1999). Learning to solve problems on the
Web: Aggregate planning in a business management course. Distance Education, 20(1),
49-63.

62
Kanuka, H., & Anderson, T. (1998). On-line interchange, discord, and knowledge construction.
Journal of Distance Education, 13(1), 57-74.

Kanuka, H. & Anderson, T. (1999). Using constructivism in technology mediated learning:

Constructing order out of the chaos in the literature. Radical Pedagogy, 1(2).

Keeler, C. M., & Anson, R. (1995). An assessment of cooperative learning used for basic
computer skills instruction in the college classroom. Journal of Educational Computing
Research, 12(4), 379-393.
Kim, M.-S., & Kim, H.-M. (2017). The effect of online fan community attributes on the loyalty
and cooperation of fan community members: The moderating role of connect hours.
Computers in Human Behavior, 68, 232-243.
Kirschner, P. A., & Erkens, G. (2013). Toward a framework for CSCL research. Educational
Psychologist, 48(1), 1-8.

Klein, J. D., & Doran, M. S. (1999). Implementing individual and small group learning structures
with a computer simulation. Educational Technology Research and Development, 47(1),
97-110.

Kozma, R. (1991). Learning with media. Review of Educational Research, 61, 179-221.

Kozma, R. (1994). Will media influence learning: reframing the debate? Educational Technology
Research and Development, 42(2), 7-19.

Kreijns, K., Kirschner, P. A., & Jochems, W. (2003). Identifying the pitfalls for social interaction
in computer-supported collaborative learning environments: a review of the research.
Computers in Human Behavior, 19(3), 335-353. doi:http://dx.doi.org/10.1016/S0747-
5632(02)00057-2

Kulik, C. & Kulik, J. A. (1991). Effectiveness of computer-based instruction: An updated

analysis. Computers in Human Behavior, 7, 75-74.

63
Kyndt, E., Raes, E., Lismont, B., Timmers, F., Cascallar, E., & Dochy, F. (2013). A meta-
analysis of the effects of face-to-face cooperative learning. Do recent studies falsify or
verify earlier findings? Educational Research Review, 10, 133-149.

Lai, E. R. (2011). Collaboration: A Literature Review (Rep.). Retrieved from Pearson website:
http://www.pearsonassessments.com/research

Lee, K. (2007). Online collaborative case study learning. Journal of College Reading and
Learning, 37(2), 82-100.

*Lin, J. M., Wu, C., & Liu, H. (1999). Using SimCPU in cooperative Learning laboratories.
Journal of Educational Computing Research, 20(3), 259-277.

Lou, Y., Abrami, P. C., & D’Appolonia, S. (2001). Small group and individual learning with
technology: A meta-analysis. Review of Educational Research, 71(3), 449-521.

Lou, Y., Bernard, R. M., & Abrami, P. C. (2006). Media and pedagogy in undergraduate distance
education: A theory-based meta-analysis of empirical literature. Educational Technology
Research & Development, 54(2), 141-176.

Lou, Y., Abrami, P. C., Spence, J. C., Poulson, C., Chambers, B., & d’Apollonia, S. (1996).
Within-class grouping: A meta-analysis. Review of Educational Research, 66, 423-458.

Mantri, A., Dutt, S., Gupta, J. P., & Chitkara, M. (2008). Design and evaluation of a PBL-based
course in analog electronics. IEEE Transactions on Education, 51, 432-438.

Margaryan, A., Bianco, M., & Littlejohn, A. (2015). Instructional quality of massive open online
courses (MOOCs). Computers & Education, 80, 77-83.

*Mastropieri, M. A., Scruggs, T. E., Norland, J. J., Berkeley, S., Mcduffie, K., Tornquist, E. H.,
& Connors, N. (2006). Differentiated curriculum enhancement in inclusive middle school
science: effects on classroom and high-stakes tests. The Journal of Special Education,
40(3), 130-137.

64
Mayer, R. E. (2008). Learning and instruction (2nd ed.). Upper Saddle River, NJ: Merrill
Prentice-Hall.

Moore, M. G. (1989). Three types of interaction. American Journal of Distance Education, 3(2),
1-7.

Nelson, C. E. (1994). Critical thinking and collaborative learning. New Directions for Teaching

and Learning, 1994 (59), 45-58.

Newmann, F., & Thompson, J. A. (1987). Effects of cooperative learning on achievement in

secondary schools: a summary of research. Madison, WI: University of Wisconsin-
Madison, National Center on Effective Secondary Schools.

*Nugent, G., Kunz, G., Levy, R., Harwood, D., & Carlson, D. (2008). The impact of a field-
based, inquiry-focused model of instruction on preservice teachers’ science learning and
attitudes. Electronic Journal of Science Education, 12(2).

Ollendick, T., & Schroeder, C. S. (2003). Encyclopedia of clinical child and pediatric
psychology. NY: Kluwer Academic/Plenum Publishers.

*Olgun, O. S., & Adali, B. (2008). Teaching grade 5 life science with a case study approach.
Journal of Elementary Science Education, 20(1), 29-44.

Panitz, T. (1996). A definition of collaborative vs co-operative learning. Retrieved from

http//:www.londonmet.ac.uk/deliberations/collaborative-learning/panitz paper.cfm

Paris, S. G., & Winograd, P. (1990). How metacognition can promote academic learning and
instruction. In B. F. Jones & L. Idol (Eds.), Dimensions of thinking and cognitive
instruction (pp. 15-51). Hillsdale, NJ: Lawrence Erlbaum.
*Pariseau, S. E., & Kezim, B. (2007). The Effect of Using Case Studies in Business Statistics.
Journal of Education for Business, 83(1), 27-31.

65
Pedró, F. (2005). Comparing traditional and ICT-enriched university teaching methods:
Evidence from two empirical studies. Higher Education in Europe, 30, 399-411.

Phua, J., Jin, S. V., & Kim, J. (2017). Gratifications of using Facebook, Twitter, Instagram, or
Snapchat to follow brands: The moderating effect of social comparison, trust, tie strength,
and network homophily on brand identification, brand engagement, brand commitment,
and membership intention. Telematics and Informatics, 34(1), 412-424.

Polanin, J. R., Tanner-Smith, E. E., & Hennessy, E. A. (2015). Estimating the difference between
published and unpublished effect sizes: a meta-analysis. Review of Educational Research.
OnlineFirst, Available from http://rer.aera.net

Porta, M., (2008). (Eds). A Dictionary of Epidemiology. Oxford University Press: New York.

* Priebe, R. (1997). The Effects of Cooperative Learning in a Second-Semester University

Computer Science Course. Paper presented at the National Association for Research in
Science Teaching Annual Meeting 1997. https://www.learntechlib.org/p/82425

Puzio, K., & Colby, G. T. (2013). Cooperative learning and literacy. Journal of Research on
Educational Effectiveness, 6(4), 339-360.

Qin, Z., Johnson, D. W., & Johnson, R. T. (1995). Cooperative versus competitive efforts and
problem solving. Review of Educational Research, 65, 129-143.

Schmid, R. F., Bernard, R. M., Borokhovski, E., Tamim, R. M., Abrami, P. C., Surkes, M. A., et
al. (2014). The effects of technology use in postsecondary education: a meta-analysis of
classroom applications. Computers & Education, 72, 271-291.

Schmid, R. F., Bernard, R. M., Borokhovski, E., Tamim, R., Abrami, P. C., Wade, C. A., et al.
(2009). Technology’s effect on achievement in higher education: A stage I meta analysis
of classroom applications. Journal of Computing in Higher Education, 21(2), 95-109.

66
Sherman, G. P., & Klein, J. D. (1995). The effects of cued interaction and ability grouping
during cooperative computer-based science instruction. Educational Technology
Research and Development, 43(4), 5-24.

Slavin, R. E. (1990). Research on cooperative learning: Consensus and controversy. Educational

leadership, 47(4), 52-54.

Smith, B. L. & McGregor, J. (1992). What is cooperative learning? In A. S. Goodsell, M. R.

Maher,V. Tino, B. L. Smith, & J. McGregor (Eds.), Collaborative learning: A
sourcebook for higher education.Washington DC: ERIC.

Springer, L., Stanne, M. L., & Donovan, S. S. (1999). Effects of small-group learning on
undergraduates in science, mathematics, engineering, and technology. Review of
Educational Research, 69I1), 21-51.

Susman, E. B. (1998). Cooperative learning: A review of factors that increase the effectiveness
of cooperative computer-based instruction. Journal of Educational Computing Research,
18(4), 303-322.

Tamim, R. M., Bernard, R. M., Borokhovski, E., Abrami, P. C., & Schmid, R. F. (2011). What
forty years of research says about the impact of technology on learning. Review of
Educational Research, 81(1), 4-28. doi:10.3102/0034654310393361

Thanh, P. T., Gillies, R., & Renshaw, P. (2008). Cooperative learning (CL) and academic
achievement of Asian students: A true story. International Education Studies, 1(3), 83-
88. doi:10.5539/ies.v1n3p82

*Terwel, J., Oers, B. V., Dijk, I. V., & Eeden, P. V. (2009). Are representations to be provided or
generated in primary mathematics education? Effects on transfer. Educational Research
and Evaluation,15(1), 25-44.

Tomcho, T. J., & Foels, R. (2012). Meta-analysis of group learning activities: Empirically based
teaching recommendations. Teaching of Psychology, 39(3), 159-169.

67
*Tsai, M. (2002). Do male students often perform better than female students when learning
computers? A study of Taiwanese eighth graders' computer education through strategic
and cooperative learning. Journal of Educational Computing Research, 26(1), 67-85.

Uribe, D., Klein, J. D., & Sullivan, H. (2003). The effect of computer-mediated collaborative
learning on solving ill-defined problems. Educational Technology Research and
Development, 51, 5-19.

Valentine, J. C., & Cooper, H. (2003). Effect size substantive interpretation guidelines: Issues in
the interpretation of effect sizes. Washington, DC: What Works Clearinghouse.

Valentine, J. C., & Cooper, H. (2008). A systematic and transparent approach for assessing the
methodological quality of intervention effectiveness research: the study design and
implementation assessment device (Study DIAD). Psychological Methods, 13(2), 130-
149.

Van Boxtel, C., Van der Linden, J., & Kanselaar, G. (2000). Collaborative learning tasks and the
elaboration of conceptual knowledge. Learning and Instruction, 10(4), 311-330.

Vygotsky, L.S. (1962). Thought and language. Cambridge, MA: MIT Press.

Webb, N.M. (1991). Task-related verbal interaction and mathematical learning in small groups.
Research in Mathematics Education, 22(5), 366-389.

Webb, N. M. (1993). Collaborative group versus individual assessment in mathematics:

Processes and outcomes. Educational Assessment, 1(2), 131-152.

Webb, N. M., & Palincsar, A. S. (1996). Group processes in the classroom. In D. Berliner & R.
Calfee (Eds.), Handbook of educational psychology (pp. 841-873). New York:
Macmillan.

Wecker, C., & Fischer, F. (2014). Where is the evidence? A meta-analysis on the role of
argumentation for the acquisition of domain-specific knowledge in computer-supported
collaborative learning. Computers & Education, 75, 218-228.

68
*Wenk, M., Waurick, R., Schotes, D., Gerdes, C., Aken, H. K., & Pöpping, D. M. (2008).
Simulation-based medical education is no better than problem-based discussions and
induces misjudgment in self-assessment. Advances in Health Sciences Education,14(2),
159-171.

Wilczenski, F. L, Bontrager, T., Ventrone, P., & Correia, M. (2001). Observing collaborative
problem-solving processes and outcomes [Electronic version]. Psychology in the Schools,
38(3), 269 – 281.

Williams, S. M. (2009). The impact of collaborative, scaffolded learning in K-12 schools: A

meta-analysis. Retrieved from Cisco Systems website: https://www.cisco.com/web/about/
citizenship/socio-economic/docs/Metiri_Classroom_Collaboration_Research.pdf

Wright, K. B., Kandel-Cisco, B., Hodges, T. S., Metoyer, S., Boriack, A. W., Franco Fuenmayor,
S. E., & Waxman H. C. (2013). Developing and assessing students’ collaboration in the IB
programme. Retrieved from http://www.ibo.org/globalassets/publications/ib-
research/developingandassessingstudentcollaborationfinalreport.pdf

*Zumbach, J., Kumpf, D., & Koch, S. (2004). Using multimedia to enhance problem-based
learning in elementary school. Information technology in childhood education annual,
1(2004), 25-37.

69
 Appendix A
Codebook: based on the meta-analysis conducted by Schmid et al. (2014).

Categories and details of the codebook

Study ID number

Sequential number of the ES

Author(s)

Publication Data

Year of publication

Type of publication

1) Journal

2) Dissertation

3) Conference Proceedings

4) Report/Gray literature

Effect Size Extraction

n for the experimental group

n for the control group

Total N (the entire sample size)

Effect size (d)

Procedure of ES extraction

1) Calculated using reported descriptive statistics

2) Calculated using reported inferential statistics

70
 3) Estimated from partial inferential statistics, e.g. reported p-value

4) Estimated from hypothesis (p < a) or assumption of equal sample size when only N is
given.

5) ES reported by the authors (only used when no other information is available)

Outcome Information

Outcome type:

Achievements (Exam, GPA, Grades, and tests)

Form of outcome measure

1) Most representative (cumulative) one-time performance measure

2) Reported composite of several tests/evaluations

3) Calculated composite of several assessments reported in the paper

4) Individual measure/item selected to represent the corresponding outcome

Nature of comparison

Brief description of both, experimental and control, conditions and of the source of data for ES
extraction (open-ended entry)

Methodological Quality

Research design

1) Quasi-experimental design (QED, non-equivalent groups with control for selection bias, etc.)

2) True experimental design (RCT, random assignment of participants to groups, etc.)

Learner Demographics

Academic level of learners

71
 1 = Kindergarten (KG)

2 = Grades 1-5 (Elementary)

3 = Grades 6-8 (Middle school)

4 = Grades 9-12 (High school)

5 = Higher education: Undergraduate

6 = Higher Education: Graduate

7 = Combination: Specify (e.g., 7: 5+6)

8 = Other: Specify (e.g., 8: Military)

Subject matter

Open entry:

Specify subject matter as reported in the study

For comparison in analysis: STEM (1) and Non- STEM (2)

999 = Missing information

Nature of Treatment

Duration of the treatment

More than three days but less than eight weeks = In Between
Nine weeks or more = semester,
Three days or less = short
999 = Missing information
Delivery mode (Ge & Gc)

1 = F2F (Classroom Instruction)

2 = BL *a substantive mix of both modes are used simultaneously

3 = DE (Distance Education)

72
 4 = Fixed Computer automated program: in lab, class or on campus, with or without the
presence of a lab assistant

5 = Flexible Computer automated program with no instructor Lab-based (class or

campus)

999 = Missing information

Technology Presence (Ge & Gc)

1 = Yes

2 = No

999 = Missing information (???)

Technology type (Ge & Gc)

Open entry:

Please, name technological tool(s) as reported in the study OR N/A - when none

Major purpose of technology use (Ge & Gc)

1 = Communication/interaction

2 = Cognitive support (distributed cognition, e.g. Excel, Word, SPSS)

3 = Cognitive support (deep understanding - e.g., simulations, knowledge creation)

4 = Informational resources

5 = Presentation

6 = A mixture of max two (should be really two major purposes where one cannot be
singled out - e.g., 6: 2 & 5)

999 = Missing information (???)

73
 Appendix B

Mixed effects analyses of the methodological features of the studies

Mixed effects analysis by research design

Groups Number of Studies Point estimate (g) Standard error P-value

QE 21 0.34 0.107
RCT 7 0.04 0.116
Total within
Total between 0.062
Overall 28 0.20 0.079

Mixed effects analysis by Measure source

(k =1 removed)

Groups Number of Studies Point estimate (g) Standard error P-value

Single cumulative
measure 19 0.31 0.12
Calculated average 8 0.07 0.12
Total within
Total between 0.152
Overall 27 0.20 0.08

Legend: 1 = Single cumulative measure (Final score)

3 = Calculated average (e.g. Projects, assignments)
4 = Single selected measure which was removed)

74
Mixed analysis by ES extraction Procedure
(k =1 removed)

Groups Number of Studies Point estimate (g) Standard error P-value

Calculated from
Descriptives 21 0.29 0.12
Estimated with
assumptions 6 0.19 0.13
Total within
Total between 0.562
Overall 27 0.24 0.09

Mixed effects analysis by MQ composite

Groups Number of Studies Point estimate (g) Standard error P-value

High 21 0.28 0.12
Low 7 0.20 0.10
Total within
Total between 0.624
Overall 28 0.24 0.08

75