You are on page 1of 26

Educational Psychologist

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/hedp20

Individual preparation for collaborative learning:


Systematic review and synthesis

Stephan Mende , Antje Proske & Susanne Narciss

To cite this article: Stephan Mende , Antje Proske & Susanne Narciss (2021) Individual
preparation for collaborative learning: Systematic review and synthesis, Educational Psychologist,
56:1, 29-53, DOI: 10.1080/00461520.2020.1828086

To link to this article: https://doi.org/10.1080/00461520.2020.1828086

© 2020 The Author(s). Published with


license by Taylor & Francis Group, LLC.

Published online: 22 Oct 2020.

Submit your article to this journal

Article views: 1551

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=hedp20
EDUCATIONAL PSYCHOLOGIST
2021, VOL. 56, NO. 1, 29–53
https://doi.org/10.1080/00461520.2020.1828086

Individual preparation for collaborative learning: Systematic review


and synthesis
Stephan Mende, Antje Proske, and Susanne Narciss
Psychology of Learning and Instruction, Technische Universit€at Dresden, Germany

ABSTRACT
Collaboration provides learners with opportunities to develop an understanding beyond what
they could achieve alone. To this end, learners need to build on each other’s knowledge to draw
new conclusions. This requires successful retrieval, inferencing, and mutual referencing during col-
laboration. Although individual preparation is considered as effective means to foster these proc-
esses it has not been systematically investigated whether, why, and under what conditions it does
so. We revisit research on collaborative learning, collaborative memory, and group brainstorming
to develop hypotheses about the cognitive advantages and disadvantages of individual prepar-
ation for collaboration and how these might be influenced by the design of the individual prepar-
ation phase. Subsequently, we test these hypotheses by systematically reviewing experimental
studies. Results indicate that (a) individual preparation affects retrieval, inferencing, and referenc-
ing differently, and (b) generative preparation tasks and supporting learners’ cognitive group
awareness can enhance the advantages and mitigate the disadvantages of individual preparation
for collaboration.

Introduction contradictory findings attract attention to the issue of how


existing research findings can help to explain whether, why
Being able to effectively collaborate is considered a core
and under which conditions collaborative learning supports
21st-century skill because modern work environments
the acquisition of in-depth domain knowledge.
require performing complex teamwork (Graesser et al.,
Engaging in productive social interactions is considered a
2018). In order to promote collaboration skills and the
core condition for effective collaborative learning (Andrews &
acquisition of in-depth knowledge, collaborative learning is
Rapp, 2015; Arvaja et al., 2008; Deiglmayr et al., 2015;
widely used across schools and university settings (Kirschner
& Erkens, 2013; Nokes-Malach et al., 2015). In collaborative Dillenbourg et al., 2009). Many researchers agree that pro-
learning settings, the members of pairs or small groups ductive social interactions involve a specific kind of activity
interact with each other to reach a common learning goal sometimes referred to as knowledge co-construction
(Dillenbourg, 1999; Roschelle & Teasley, 1995). This can (Deiglmayr & Spada, 2010b; Hogan et al., 1999; Scardamalia
take various forms, such as asking pairs of learners to & Bereiter, 2006), or interactive activity (Chi, 2009). Learners
explain an expository text to each other, asking four stu- are said to perform an interactive activity when they (a) build
dents to create a joint learning product via a shared elec- on the information contributed by a co-learner to (b) generate
tronic workspace, or asking groups of five to reach a new knowledge not already stated in the instructional mater-
consensus on a controversial content-related issue in an ial. If collaborating learners perform interactive activities, they
online discussion. Even though research on collaborative can achieve a deeper understanding of the subject matter
learning has received much attention in the last decades, compared to when learning individually (Chi et al., 2018;
designing and implementing collaborative learning settings Fischer, Kollar, Stegmann, Wecker, et al., 2013).
systematically based on scientific evidence is challenging Hence, collaborative learning settings offer the potential
because research findings are rather mixed. Regarding the benefit of using co-learners as additional learning resources.
issue of how collaborative learning affects learners’ informa- However, these co-learners also impose cognitive costs in
tion processing in terms of in-depth knowledge acquisition, terms of complex coordination demands (Janssen &
studies show that collaborative learning can be superior but Kirschner, 2020; Nokes-Malach et al., 2015). Accordingly,
also inferior to individual learning (Andrews & Rapp, 2015; learners are often not spontaneously engaged in interactive
Kirschner et al., 2009; Nokes-Malach et al., 2015). These activities when they collaborate without support (Bromme

CONTACT Stephan Mende stephan.mende@tu-dresden.de Faculty of Psychology, Technische Universit€at Dresden, Zellescher Weg 17, Dresden 01062, Germany.
This article was accepted under the editorship of Kathryn Wentzel.
ß 2020 The Author(s). Published with license by Taylor & Francis Group, LLC.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.
0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in
any way.
30 S. MENDE ET AL.

et al., 2005; Jeong & Hmelo-Silver, 2016; Kirschner & with unrestricted capacity where information is stored, and
Erkens, 2013; Menekse & Chi, 2019). This raises the ques- a working memory of limited capacity where information
tion of helpful strategies that teachers can apply in the class- from external sources can be processed to encode and store
room and/or teach their students. it in long term memory (e.g., Atkinson & Shiffrin, 1968;
Prior research has mainly addressed different support Cowan, 2014). In collaborative learning, the external infor-
strategies and tools guiding or structuring the collaboration mation sources comprise the instructional material and—as
process itself (e.g., Jeong & Hmelo-Silver, 2016). Another far as they externalize it (e.g., in the form of spoken or writ-
support strategy which takes the approach of initially avoid- ten language)—the knowledge of co-learners.
ing the costs to subsequently raise the benefits of collabor- To be classified as interactive, a learning activity has to
ation is individual preparation for collaboration: providing simultaneously fulfill two criteria: the learner must produce
learners with time to perform activities directed at process- an externalization which (a) contains topic-relevant informa-
ing the instructional material on their own before the collab- tion beyond what is given in the instructional material and
oration (e.g., Lam & Kapur, 2017; Tsovaltzi et al., 2015). (b) incorporates a co-learner’s externalized knowledge (Chi
Since it is a widely recommended strategy (e.g., Andrews & et al., 2018).
Rapp, 2015; cf. Tsovaltzi et al., 2015), individual preparation Unfortunately, studies measuring interactive activities
has been included in the study protocols of several lines of according to this definition are scarce in collaborative learn-
collaborative learning research such as research on collabor- ing research (cf. Deiglmayr et al., 2015; Mende et al., 2017)
ation scripts (e.g., Fischer, Kollar, Stegmann, Wecker, et al., and—to our knowledge—not existent in research on individ-
2013) or group awareness support (e.g., Janssen & Bodemer, ual preparation for collaborative learning. However, prior
2013). However, within these fields, individual preparation research has captured measures that can be considered to
has often been applied implicitly (e.g., Engelmann, Tergan reflect the execution of cognitive subtasks which are them-
et al., 2009) or manipulated together with other strategies selves necessary but not sufficient for an interactive activity
(e.g., Noroozi et al., 2013) not allowing disentangling the to occur. These subtasks include (Chi et al., 2018; Cress &
pure effect of individual preparation. Hence, it has not often Kimmerle, 2008; Deiglmayr & Spada, 2010a, 2010b): (a) acti-
vating content-relevant knowledge stored in long-term
been the subject of systematic investigation.
memory (retrieval), (b) generating new pieces of knowledge
By contrast, the recently emerging preparation for future
not already contained in the instructional material (inferenc-
collaboration paradigm (Lam & Kapur, 2017; Lam & Muldner,
ing), and (c) paying attention to and taking up content-rele-
2017) has focused on the questions of whether and how learn-
vant information externalized by co-learners, that is, using
ers should cognitively prepare on their own prior to collabor-
co-learners’ externalizations as a reference (referencing; cf.
ation. Thus, first attempts have been made to explore the
Teasley, 1997; Weinberger et al., 2007).
cognitive mechanisms by which individual preparation can
Please note that these subtasks are not unrelated to each
foster the execution of interactive activities during subsequent other: Retrieving associated knowledge to process external
collaboration (see for example J€arvel€a et al., 2015 or Miller & information in working memory is a prerequisite for inferenc-
Hadwin, 2015 for another perspective targeting the metacog- ing (Chi et al., 2018; Cress & Kimmerle, 2008; Jorczak, 2011).
nitive regulation processes necessary for the execution of Further, referencing can facilitate retrieval and inferencing
interactive activities). However, theoretical and empirical (Janssen & Kirschner, 2020; Wegner, 1987). In contrast to the
work in this regard has only just begun. subtasks of retrieval and inferencing, the subtask of referenc-
Hence, to date, it is not very clear whether, when, and why ing can only be performed in collaborative learning.
individual preparation may foster the execution of interactive
activities during subsequent collaboration. Consequently, by
adopting an information processing perspective, the present Benefits and costs of processing information
work aims to address this research gap by (a) specifying more collaboratively versus individually
precisely the cognitive subtasks involved in interactive activ- Individual preparation for collaboration means devoting a
ities; (b) developing hypotheses about why and how individual part of the learning time for learning individually prior to
preparation may affect the execution of these subtasks; and (c) learning collaboratively. To develop hypotheses on the
exploring the role of individual preparation design variables effects and conditions of individual preparation with respect
(i.e., preparation task type and awareness induction support) to the subtasks of interactive activity, we will now examine
for the impact of individual preparation. Subsequently, a sys- research addressing the cognitive benefits and costs of col-
tematic review of experimental studies concerned with these laboration as compared to individual learning regarding
issues is conducted. retrieval and inferencing. Furthermore, we will investigate
the role that referencing plays for these benefits and costs.
The benefits and costs of collaboration are investigated
Interactive activities viewed from an information
within different research fields including not only collabora-
processing perspective: Identifying subtasks
tive learning research but also more basic research in collab-
According to an information processing perspective, collab- orative memory and group brainstorming. The latter two
oration can be seen as the interaction between several cogni- examine the impact of collaboration with a high level of
tive systems (i.e., learners; Cress & Kimmerle, 2008; Jorczak, experimental control. Thus, the basic findings of these lines
2011). Each cognitive system involves a long-term memory of research can provide valuable insights into the cognitive
EDUCATIONAL PSYCHOLOGIST 31

mechanisms contributing to the benefits and costs of collab- their individual retrieval strategies. Thus, in a collaborative
oration in authentic learning processes. retrieval situation, they may disrupt each other. This risk is
attenuated during closed recall since the test itself deter-
mines the retrieval order (Marion & Thorley, 2016; Rajaram
Benefits and costs of collaboration for retrieval & Pereira-Pasarin, 2010). As a further mechanism, an item
Collaborative memory research investigates the impact of retrieved by a co-learner may induce another co-learner to
collaboration on subjects’ retrieval performance (e.g., retrieve the same item which inhibits competing responses
Rajaram & Pereira-Pasarin, 2010) based on theoretical elicited in the process of retrieval (retrieval inhibition; e.g.,
approaches such as the multi-store model (Atkinson & Cuc et al., 2007; Marion & Thorley, 2016). Moreover,
Shiffrin, 1968), the spread of activation theory of memory retrieval errors made by one group member may be trans-
(e.g., Anderson & Pirolli, 1984), or levels-of-processing the- mitted to other group members (social contagion; e.g.,
ory (Craik & Lockhart, 1972). It provides insights into why Basden et al., 2002).
the subtask of retrieval might sometimes be less successfully
executed when learners have to collaborate instead of work-
ing on their own. Benefits and costs of collaboration for inferencing
In a typical collaborative memory experiment, subjects Based on human memory and information processing theo-
encode given learning material and then retrieve the learned ries (e.g., Anderson & Pirolli, 1984; Atkinson & Shiffrin,
information either collaboratively in groups or individually 1968; Kintsch, 1998; Sweller & Sweller, 2006), the benefits
in so-called nominal groups. Comparisons are then con- and costs of collaboration for subjects’ capabilities to draw
ducted between the retrieval performance of collaborative inferences are investigated by cognitively oriented research
groups with that of nominal groups (Marion & Thorley, lines within the fields of group brainstorming and collabora-
2016; Rajaram & Pereira-Pasarin, 2010). Retrieval perform- tive learning. Hence, both research lines provide insights
ance is usually defined as the number of items that are cor- into why the subtask of inferencing may sometimes be less
rectly retrieved by at least one group member in response to successfully executed when learners have to collaborate
an open or closed memory test (i.e., free recall or cued instead of working on their own.
recall/recognition). The retrieved items are summed up to a Group brainstorming research is focused on examining
group retrieval score with items mentioned more than once the impact of collaboration on idea generation. Idea gener-
being counted only once. For the nominal groups, the data ation can be seen as a form of inferencing since subjects
of several individuals are analyzed as if they would have have to combine several pieces of their knowledge in order
worked in a group even though they had worked separately to create an idea fitting with the constraints of a given
(Marion & Thorley, 2016). brainstorming task. Thus, subjects have to go beyond what
Retrieval benefits are claimed to arise from the possibility is already stated in the instructional material (cf. Nijstad &
to profit from each other’s outputs. For example, outputs Stroebe, 2006; Wang & Rose, 2007). Group brainstorming
produced by co-learners during collaborative retrieval may experiments typically compare the idea generation perform-
serve as a reminder for otherwise forgotten items (re-expos- ance of nominal to collaborative groups, similar to the col-
ure; e.g., Rajaram, 2011), as a cue to retrieve additional laborative memory paradigm (Nijstad & Stroebe, 2006;
items otherwise not retrieved (cross-cueing; e.g., Blumen Stroebe et al., 2010). Idea generation performance is usually
et al., 2014), or as feedback to correct retrieval errors (error captured by measuring groups’ ideation productivity, that is,
pruning; e.g., Barber et al., 2012). the sum of non-redundant ideas generated by the group
However, collaborative memory research offers over- members in response to the brainstorming task, such as
whelming evidence for the so-called collaborative inhibition finding as many ways as possible to increase ones’ home-
effect (Weldon & Bellinger, 1997): collaborative groups are town’s attractiveness for tourists (Nijstad & Stroebe, 2006;
reliably revealed to perform worse than nominal groups Stroebe et al., 2010).
(Marion & Thorley, 2016). These findings are explained by Group brainstorming researchers assume that cognitive
retrieval costs arising from being exposed to each other’s benefits of group work arise from stimulation effects, that is,
outputs. The best empirically supported explanation for the the ideas expressed by one group member can cue new ideas
collaborative inhibition effect is the mechanism of retrieval or categories of ideas another group member would not
strategy disruption (Marion & Thorley, 2016; Rajaram & have generated or surveyed on their own (Brown et al.,
Pereira-Pasarin, 2010): subjects develop individual retrieval 1998; Nijstad et al., 2002; Nijstad & Stroebe, 2006; Stroebe
strategies while encoding to-be-learned information. Due to et al., 2010).
these individually different retrieval strategies, the individual However, group brainstorming research offers corrobo-
group members will retrieve the to-be-learned information rating evidence that collaborative groups perform worse
in different orders. When retrieving information together, than nominal groups in terms of idea generation productiv-
they thus disrupt each other’s retrieval strategies (Basden ity (Diehl & Stroebe, 1987; Mullen et al., 1991), an effect
et al., 1997; Marion & Thorley, 2016). This explanation called productivity loss (e.g., Nijstad & Stroebe, 2006). These
receives support, among others, from the finding that the findings are explained by cognitive costs which can arise
type of retrieval test affects the degree to which collaborative from mechanisms such as production blocking (e.g., Nijstad
inhibition appears: In open recall tests the group members & Stroebe, 2006): group members have to take turns during
try to retrieve the previously learned items according to the ideation process and, thus, cannot express their ideas
32 S. MENDE ET AL.

immediately when they arise (Diehl & Stroebe, 1987). another learner to generate an explanation of a to-be-
Consequently, subjects tend to forget their own ideas and/or learned principle. Or a counterargument of a learner may
are deterred from generating new ideas (Nijstad et al., induce another learner to draw new connections between
2003). Further, the dual task of generating ideas and attend- their own and the co-learner’s knowledge to solve the
ing to the ideas of others can overload subjects’ working disagreement.
memory capacities (Dennis & Williams, 2005; Paulus et al., However, research findings reveal considerable variability
2013; Santanen et al., 2004). In addition, although the ideas concerning the superiority or inferiority of collaborative as
available from others can benefit one’s own idea generation, compared to individual learning in view of inferencing indi-
they can also lead to fixation effects: an idea presented by a cators (Andrews & Rapp, 2015; Menekse & Chi, 2019;
group member cues associated parts of another member’s Nokes-Malach et al., 2015). Moreover, in studies which
knowledge, favoring the generation and expression of similar explicitly investigate learners’ collaboration activities, it is
ideas, etc. In this way each single group member’s range of frequently observed that collaborating learners often tend to
explored ideas is restricted (Kohn & Smith, 2011; Nijstad & restate what has already been given instead of drawing infer-
Stroebe, 2006; Wang & Rose, 2007). ences (Chase et al., 2019; Chi & Menekse, 2015; Roscoe &
The focus in collaborative learning research is on the Chi, 2007; Webb, 2013). To explain these findings, cognitive
effects and conditions of collaboration for learning from costs are proposed in terms of collaborative learning putting
educationally relevant materials and tasks (cf. Pociask & higher demands on the learners’ cognitive resources than
Rajaram, 2014). This is examined, among others, according individual learning. Not only information of the instruc-
to (quasi-) experimental designs. Here, several measures tional material, but also information externalized by the co-
reflect learners’ inferencing performance. First, inferencing learners has to be processed, yielding a risk of information
can be indicated by learners’ discourse contributions con- overload and distractions. Hence, learners’ limited working
taining content-related information not already present in memory capacities can be exceeded which may hamper deep
the instructional material (such as knowledge construction learning processes such as inferencing (Dillenbourg &
statements; e.g., King, 1999; or elaboration statements; e.g., Betrancourt, 2006; Kolfschoten et al., 2014).
Roscoe, 2014). Second, it can be reflected in performance
measures on group tasks requiring the learners to generate The role of referencing for the benefits and costs of
and externalize content-relevant outputs that go beyond the collaboration
instructional material, for example, to solve a transfer prob- The role of referencing for the benefits and costs of collabor-
lem (e.g., Huang et al., 2012) or to transform information ation is addressed in collaborative learning research based on
into another representational format (e.g., a text into a con- transactive memory theory (Wegner, 1987), the concept of cog-
cept map or a picture, see for example Engelmann, Tergan nitive group awareness (e.g., Janssen & Bodemer, 2013), as
et al., 2009; Gijlers et al., 2013). Third, achievement in post- well as the collaborative cognitive load theory (Kirschner et al.,
collaborative transfer tests (e.g., Lam & Muldner, 2017) is 2018). Further, facilitators and barriers of referencing are
also considered to indicate learners’ collaborative inferencing investigated, typically in experimental and quasi-experimental
performance since subjects have to draw inferences while research designs (e.g., Engelmann, Tergan et al., 2009; Gijlers
dealing with the to-be-learned information in order to be et al., 2013).
able to apply it to problems or situations not already Referencing is operationalized by measures directly cap-
addressed in the previously accessed instructional material tured from the collaboration process. For example, the num-
(i.e., transfer achievement; Chi et al., 1994; Kintsch, 2004; ber of transactive utterances (Berkowitz et al., 2008; Teasley,
McNamara & Magliano, 2009). 1997) reflects the extent to which learners take up co-learn-
According to the cognitive elaboration perspective, one ers’ externalized content-relevant information. Referencing
benefit of collaborative learning arises from the circumstance can also be assessed by post-collaborative measures of know-
that collaborating learners have to verbalize what has been ledge convergence, that is, the amount of content-relevant
understood from to-be-learned material in order to talk knowledge shared after collaboration (Weinberger et al.,
about it with their co-learners (O’Donnell & O’Kelly, 1994; 2007). Further, it can be assessed through post-collaborative
Webb, 1989). This act of externalization is assumed to knowledge or information awareness tests, capturing a learn-
invoke processes of inferencing, for instance, in the form of er’s degree of knowledge about their co-learners’ knowledge
taking note of and filling in own knowledge gaps, re-organ- (“who knows what;” e.g., Engelmann, Tergan et al., 2009).
izing one’s knowledge, or drawing new connections between Referencing cannot occur in individual learning settings.
learned information (Cohen, 1994; Damnik et al., 2013; van It is considered a core process of successful collaborative
Boxtel et al., 2000; Webb, 2013). Moreover, based on the learning (Chi & Wylie, 2014; Fischer, Kollar, Stegmann, &
work of Piaget (1926) and Vygotsky (1978), some research- Wecker, 2013). On one hand, referencing is the central pre-
ers claim that the contributions of other learners can assist requisite for using co-learners externalizations as learning
the drawing of inferences or can invoke cognitive conflicts resources in addition to the instructional material and, thus,
stimulating inferences the learners would otherwise not have to exploit the potential benefits of collaboration for one’s
generated by themselves (e.g., Cress & Kimmerle, 2008; own retrieval and inferencing (Janssen & Kirschner, 2020;
Jorczak, 2011; King, 1999; Slavin, 2011; Webb, 2013). For Kirschner et al., 2018; Wegner, 1987). On the other hand,
instance, a learner may provide information that enables referencing is associated with transaction costs, that is, the
EDUCATIONAL PSYCHOLOGIST 33

efforts necessary to establish and maintain coordination and Individual preparation is frequently argued as being
communication with other co-learners (Janssen et al., 2010). effective in addressing the difficulties learners are faced with
Thus, the transaction costs which need to be invested in the during collaboration (e.g., Andrews & Rapp, 2015; Tsovaltzi
service of referencing also consume working memory et al., 2015). In the following sections, we revisit the collab-
resources, leaving less capacity for processing the instruc- orative memory, group brainstorming, and collaborative
tional material (Janssen & Kirschner, 2020; Kirschner learning literature in order to develop hypotheses regarding
et al., 2018). the advantages and disadvantages of having versus not hav-
Research shows that learners often fail to take up each ing learners prepare individually prior to collaboration.
other’s contributions or tend to build quick consensus Afterward, we address the issue of how design features of
(Barron, 2003; Jeong & Hmelo-Silver, 2016; Kirschner & the individual preparation phase, namely the preparation
Erkens, 2013; Weinberger, 2011). Hence, learners often task type and the provision of awareness induction support,
struggle with the interactive activity subtask of referencing. relate to these advantages and disadvantages.
To account for these findings, facilitators and barriers of
referencing are claimed and investigated (e.g., Janssen &
Individual preparation for collaboration compared to
Kirschner, 2020).
immediate collaboration
Referencing is stimulated by positive interdependence
Based on the research revisited, we hypothesize individual
(e.g., Johnson et al., 1989), that is, the experience that the
preparation for collaboration—as compared to immediate
learning task can be better solved by working collaboratively
collaboration—to yield an information processing advantage
rather than individually. Positive interdependence may
and at the same time a coordination disadvantage. The infor-
increase with increasing task complexity (which depends on
mation-processing advantage is that learners can initially
the task demands themselves and the learners’ available
process the instructional material with full cognitive capacity
prior domain knowledge) and with increasing divergence of
and without the costs resulting from being exposed to co-
task-relevant domain knowledge among the group members
learners’ externalizations. Moreover, they do not have to
(Janssen & Kirschner, 2020; Kirschner et al., 2018).
cope with transaction costs (cf. Andrews & Rapp, 2015;
Referencing is negatively affected by the degree of the
Blumen & Rajaram, 2008; Kohn et al., 2011; Korde &
associated transaction costs. The transaction costs are likely
Paulus, 2017; Tsovaltzi et al., 2015). Consequently, learners
to increase with a higher divergence of knowledge among
might be better able to activate prior knowledge in order to
the group members as well as with group size elaborate it more effectively in the subsequent collaboration
(Hollingshead, 2001; Kirschner et al., 2018). By contrast, (Lam & Kapur, 2017; Lam & Muldner, 2017; cf. Schwartz
transaction costs decrease with the degree to which group et al., 2007). Hence, having learners prepare individually
members have developed a shared knowledge about who rather than collaborate immediately could be assumed to
knows what, that is, a transactive memory system (Wegner, foster their retrieval and inferencing performance. Retrieval
1987) or cognitive group awareness (Janssen & Bodemer, may be fostered, since having subjects encode to-be-learned
2013). Accordingly, referencing is facilitated to the extent to information or practice retrieval individually can strengthen
which the collaborating learners possess knowledge regard- their personal memory organization. This in turn could
ing other group members’ knowledge, ideas, and perspec- benefit subsequent group retrieval performance (cf. Barber
tives as well as how far there is a shared understanding et al., 2010; Blumen et al., 2014; Blumen & Rajaram, 2008).
among the group members (Gijlers & de Jong, 2009; Janssen Inferencing may be fostered since learners can build up a
et al., 2007; Janssen & Bodemer, 2013; Janssen & richer initial knowledge and develop a wider range of ideas
Kirschner, 2020). and perspectives around the learning content. This in turn
could foster the capabilities to draw inferences during subse-
Individual preparation as a potential aid to improve quent collaboration (cf. Best et al., 2005; Kintsch, 1998; Lam
subsequent collaboration & Kapur, 2017; Schwartz et al., 2007).
Concerning learners’ referencing performance, the poten-
As outlined so far, interactive activities involve simultaneous tial effect of having versus not having learners individually
retrieval, inferencing, and referencing. When subjects are prepare for collaboration seems to be less clear. Learners
asked to collaborate, they often struggle with the demands may benefit from the information processing advantage
of these cognitive subtasks due to the costs arising from the since they can process the instructional material before their
collaboration: individual information processing can be cognitive resources are burdened with the transaction costs.
impeded by co-learners’ externalizations, for example by dis- Thus, more cognitive capacities might be available to deal
ruption (e.g., retrieval strategy disruption, production block- with these costs during collaboration (cf. Dillenbourg &
ing), interference (e.g., retrieval inhibition, fixation), and/or Betrancourt, 2006; Janssen & Kirschner, 2020).
information overload. In addition, learners’ limited working Due to the coordination disadvantage, however, referenc-
memory capacities are burdened by the need to invest trans- ing may also be hampered: spending a part of one’s learning
action costs. Group learning situations are therefore often time individually before collaborating results in a lack of
more complex and resource-demanding than situations of coordination opportunities and, thus, hinders the develop-
individual learning. Hence, the potential benefits of collab- ment of cognitive group awareness or a transactive memory
orative learning often do not materialize. system, respectively. Consequently, in subsequent
34 S. MENDE ET AL.

collaboration, the transaction costs to be invested in the ser- A generative preparation task can also have beneficial
vice of referencing might be increased (cf. Janssen & effects on learners’ referencing. It can lead the different
Bodemer, 2013; Janssen & Kirschner, 2020; Wegner, 1987). group members to build more divergent ideas, perspectives,
Hence, individual preparation for collaboration may not and knowledge before the start of the collaboration (cf.
necessarily foster learners’ referencing performance as com- Janssen & Kirschner, 2020; Jorczak, 2011). This may increase
pared to immediate collaboration arrangements. As outlined the perceived interdependence which could positively affect
previously, effective mutual referencing (i.e., attending to referencing performance during subsequent collaboration.
and taking up co-learners’ externalizations) is a prerequisite However, a higher knowledge divergence is also associated
to benefit from other co-learners as additional learning with higher transaction costs which could hamper learners’
resources in terms of retrieval and inferencing (cf. Janssen & referencing (cf. Janssen & Kirschner, 2020).
Kirschner, 2020; Wegner, 1987). Although the information An individual preparation phase could not only be used
processing advantage may contribute to attenuate the costs to have the learners individually process the instructional
of collaboration for retrieval and inferencing, the coordin- material but also to have them inspect each other’s individ-
ation disadvantage may thus restrict the extent to which the ual preparation products before the collaboration. This then
potential benefits of collaboration for these subtasks come represents a specific form of awareness induction
into effect. (Weinberger, 2011) or group awareness support (Janssen &
Bodemer, 2013). Awareness induction support is understood
as informing group members about each other’s knowledge,
The role of the individual preparation design: Task type ideas, and perspectives (e.g., Engelmann, Dehler et al., 2009;
and awareness induction support Janssen & Bodemer, 2013). In this way, awareness induction
The arguments outlined above raise the question of which support can reduce transaction costs (Janssen & Bodemer,
factors in the design of individual preparation may influence 2013; Jeong & Hmelo-Silver, 2016; Weinberger, 2011).
the manifestation of its potentials and risks. These design Hence, implementing such support might be particularly
factors should have an impact on the extent to which the suited to compensate for the coordination disadvantage of
information processing advantage and the coordination dis- individual preparation so that referencing could be fostered
advantage come into effect. Therefore, we consider the fol- (cf. Janssen & Bodemer, 2013). This in turn may enhance
lowing two design factors: first, the type of the individual the extent to which learners can profit from co-learners’
preparation task which may affect how the information con- externalizations with respect to retrieval and inferencing: a
tained in the instructional material is processed during prep- well-developed transactive memory system or group aware-
aration. Second, the provision of group awareness support ness can foster cross-cueing and facilitate the alignment of
which may affect coordination through stimulating the retrieval strategies in order to overcome retrieval strategy
learner to process the individual preparation products of the disruption (cf. Blumen & Rajaram, 2008; Marion & Thorley,
future co-learners before the collaboration (e.g., Engelmann, 2016; Moreland & Myaskovsky, 2000; Wegner, 1987).
Tergan et al., 2009; Janssen & Bodemer, 2013). Further, awareness induction support may increase the
Existing research particularly stresses the role of genera- extent to which learners can stimulate, provoke, and assist
tive tasks for preparing students for subsequent (collabora- each other’s inferencing during subsequent collaboration (cf.
tive) learning (e.g., Kapur, 2015; Lam & Kapur, 2017; Dugosh et al., 2000; Erkens et al., 2016; Noroozi
Schwartz et al., 2011). A generative task requires the learner et al., 2013).
to infer new knowledge by connecting different information
from the instructional material and/or by linking this infor-
mation with preexisting knowledge (e.g., transforming an Summary and research questions
expository text into a concept map or generating examples Learners often struggle with the cognitive subtasks of inter-
for a concept; Fiorella & Mayer, 2016; Grabowski, 2004; active activities in terms of retrieval, inferencing, and refer-
Wittrock, 1989). Research on individual learning shows that encing during collaboration. Preceding the collaboration
generative tasks foster a) retrieval performance due to with individual preparation is considered to aid learners in
deeper encoding (Craik & Tulving, 1975; Fiorella & Mayer, overcoming these difficulties. However, the research outlined
2016; Grabowski, 2004; Kintsch, 2004) and b) inferencing above suggests that individual preparation for collaboration
performance due to the acquisition of better-connected may at the same time yield an information processing
knowledge that can be used more flexibly (Alfieri et al., advantage and a coordination disadvantage. This attracts
2013; Chi, 2000; Kintsch, 2004; McNamara, 2004; Schwartz attention to the individual preparation design, that is,
et al., 2011). Thus, requiring learners to individually prepare whether (a) a generative, as opposed to a non-generative
with a generative task may be suited to enforce the informa- preparation task, may be particularly suited to enforce the
tion processing advantage since learners can use their undiv- information processing advantage and (b) providing learners
ided cognitive capacities in the service of deep information with their future co-learners’ individual preparation prod-
processing. Consequently, a generative task may enhance the ucts (i.e., awareness induction support) can compensate for
benefits of individual preparation for subjects’ retrieval and the coordination disadvantage.
inferencing performance in later collaboration (cf. Lam & To the best of our knowledge, none of these questions
Kapur, 2017). have been subject to a systematic review so far. Accordingly,
EDUCATIONAL PSYCHOLOGIST 35

the present work aims at systematically reviewing the (3.1) individual preparation designs involving generative versus
research bodies of collaborative learning, collaborative mem- non-generative tasks?
ory, and group brainstorming with regard to the following
(3.2) individual preparation designs providing versus not
research questions (RQ):
providing awareness induction support?
RQ 1: What effects on (a) retrieval, (b) inferencing, and (c)
referencing are revealed by studies comparing individual We expect individual preparation for collaboration employ-
preparation for collaboration to immediate collaboration? ing generative preparation tasks to be superior to individual
preparation for collaboration employing non-generative prepar-
According to the hypothesized information processing
ation tasks in terms of retrieval and inferencing. We expect indi-
advantage, we expect individual preparation for collabor-
vidual preparation for collaboration involving awareness
ation to foster learners’ retrieval and inferencing perform-
induction support to be superior to individual preparation with-
ance as compared to immediate collaboration. Since
out awareness induction support in terms of retrieval, inferenc-
inferencing is a more complex, resource-demanding cogni-
ing, and referencing. Further, for studies varying task type and
tive task than the mere retrieval of stored knowledge (e.g.,
awareness induction support, we expect individual preparation
Kintsch, 2004), we expect the benefits of individual prepar-
for collaboration involving generative preparation tasks and
ation to be more pronounced for inferencing than for
awareness induction support to be superior to conditions where
retrieval (cf. Janssen & Kirschner, 2020). Further, learners’
only one or none of these design features are applied.
referencing performance may also benefit from the informa-
In addition, we will consider participants’ age level and the
tion processing advantage but at the same time suffer from
group size as control variables. Group size is likely to increase
the coordination disadvantage. Consequently, we have no
the transaction costs associated with referencing (Janssen &
clear expectations concerning the effect of individual prepar-
Kirschner, 2020). Age level can be seen as a raw indicator for
ation for collaboration on learners’ referencing performance.
learners’ prior knowledge which may affect the degree to which
RQ 2: Are the effects revealed by studies comparing individual learners profit from support strategies such as individual prep-
preparation for collaboration to immediate collaboration
aration (cf. Lam, 2019). For studies capturing measures of
dependent on the design of the individual preparation
phase, namely retrieval performance, we will additionally consider the collab-
orative retrieval test type as a control variable since it has been
(2.1) the individual preparation task type (i.e., generative versus evidenced to affect the degree to which retrieval strategy disrup-
non-generative)? tion appears, in turn impacting collaborative retrieval perform-
ance (Marion & Thorley, 2016).
(2.2) the provision of awareness induction support (i.e.,
provided versus not provided)?

Generative tasks may be suited to enhance the informa- Method


tion processing advantage of individual preparation. The literature research was conducted in four steps: keyword
Concerning retrieval and inferencing, we therefore expect extraction, database search, abstract screening, and full-text
the positive effects of individual preparation for collabor- assessment. Figure 1 summarizes the results of the search
ation as compared to immediate collaboration to be more and selection process.
pronounced in studies using generative tasks than in studies
using non-generative tasks. Yet, regarding the effects on
referencing we consider it premature to formulate concrete Search term extraction
expectations. Awareness induction support may be suited to
Relevant search terms were extracted from reviews, meta-
compensate for the coordination disadvantage of individual
analyses, overview articles, and representative studies of the
preparation. Regarding referencing we, therefore, expect that
research fields of collaborative memory, group brainstorm-
studies providing awareness support reveal positive effects of
ing, and collaborative learning. These search terms were
individual preparation for collaboration as compared to
complemented by database thesaurus consultation.
immediate collaboration. Since effective mutual referencing
In order to reduce the risk of overlooking relevant studies,
should also facilitate retrieval and inferencing, we expect the
we defined a separate set of search terms for each research
positive effects of individual preparation for collaboration as
field. In addition, we defined a separate search term set for the
compared to immediate collaboration on these subtasks to
group awareness support literature although we considered it
be more pronounced in studies providing than in studies
as part of the collaborative learning research field. This was
not providing awareness induction support. Moreover, we
necessary due to some specifics in the relevant keywords.
expect the most pronounced positive effects of individual
For the field of collaborative memory, we conjunctively
preparation for collaboration as compared to immediate col-
combined search terms referring to the social mode of inter-
laboration regarding all three subtasks of interactive activ-
est, such as “collaborati,” “cooperati,” “peer,” “dyad,” or
ities when generative preparation tasks and awareness
induction support are applied. “group,” with terms referring to the relevant dependent vari-
able, such as “encod,” “inhibition,” “memory,” “recall,” or
RQ 3: What effects on (a) retrieval, (b) inferencing, and (c) “retriev.” For the field of brainstorming, we defined single
referencing are revealed by studies comparing different designs
search terms such as “brainstorm” or search term combina-
of individual preparation for collaboration, namely
tions such as “idea” and “generat.” For the field of group
36 S. MENDE ET AL.

Figure 1. Summary of the results of the literature research and selection process.

awareness, we combined search terms referring to the rele- basic criteria: (a) original empirical research reported in (b)
vant social mode with the term “aware.” a peer-reviewed research article in (c) English or German
For these three research fields, no specific restrictions language which was (d) published since 1987, and (e) cate-
were made. For the collaborative learning search term set gorized in a discipline or subject concerned with cognitive
we limited the findings to studies with terms like and/or group performance by the database providers (e.g.,
“preparation” in the title or keywords because, otherwise, “group performance,” “Educational Psychology,” “cognitive
the number of hits would have exceeded our analytical processes,” or “learning and memory”). These searches
capacities. Consequently, this search term set consisted not resulted in an initial number of 2,947 records (see Figure 1).
only of a conjunctive combination of terms referring to the
relevant social mode and terms referring to the dependent
variable of interest (e.g., “learn,” “comprehen,” or Abstract screening
“discourse”). To limit the findings, we complemented the set Inclusion criteria were developed in line with the PICOS
with terms referring to the relevant intervention (e.g., approach (Liberati et al., 2009; see Table 1). We screened
“prepar,” “sequenc,” or “structur”) and terms referring to the titles and abstracts of the studies according to these
the relevant social mode of the intervention (e.g., “individ,” inclusion criteria. Due to the limited information provided
“single,” or “alone”). in abstracts and titles, studies were excluded at this stage
only if the given information was sufficient to clearly con-
clude that a criterion was not fulfilled. This process nar-
Database search
rowed the list to 487 records. One study (Lam & Muldner,
The literature search was executed in the databases 2017), a priori known to be potentially relevant by the
PsycINFO, PsycARTICLES, PSYNDEX, and Web of Science authors, was initially not part of the search results due to a
between May and August 2017 according to the following missing entry in the database PSYCINFO. However, the
EDUCATIONAL PSYCHOLOGIST 37

Table 1. Inclusion criteria according to the PICOS approach (Liberati et al., 2009).
Review questions 1. What effects on (a) retrieval, (b) inferencing, and (c) referencing are revealed by studies comparing individual
preparation for collaboration to immediate collaboration?
2. Are the effects revealed by studies comparing individual preparation for collaboration to immediate collaboration
dependent on the design of the individual preparation phase (i.e., task type and awareness induction support)?
3. What effects on (a) retrieval, (b) inferencing, and (c) referencing are revealed by studies comparing different designs
of individual preparation for collaboration (i.e., individual preparation designs differing in task type or the provision
of awareness induction support)?

Population  Adults and school children.


 Non-clinical populations.

Intervention Collaboration with a preceding individual preparation.

Comparator Collaboration without a preceding individual preparation (i.e., immediate collaboration) OR


Collaboration with a preceding individual preparation with the latter differing from the intervention either with
respect to
 the preparation task type (generative versus non-generative) or
 the provision of awareness induction support (present versus absent).

Further specifications The individual preparation phase …


 … has to be content-specific, that is, subjects are required to deal with the same subject matter or with the same
(superordinate) task as in the subsequent collaboration phase.
 … has to require subjects to execute observable actions focused on dealing with the subject matter or task (i.e., not
only passive reception).

Task type and the presence of awareness induction support are determined according to the following definitions:
 A generative task requires the subject to create meaningful connections not already stated in the available
instructional material. Such connections could be drawn between different pieces of information from the
instructional material and/or between the information from the instructional material and preexisting knowledge (Chi
& Wylie, 2014; Fiorella & Mayer, 2016; Grabowski, 2004; Wittrock, 1989).
 An individual preparation is considered to be complemented by awareness induction support if learners are explicitly
provided with and instructed to inspect the preparation products of their (future) group members before
collaboration (cf. Engelmann, Dehler et al., 2009; Janssen & Bodemer, 2013; Weinberger, 2011).

Outcomes At least one quantitative, non-subjective measure indicative of subjects’ retrieval, inferencing, or referencing performance
during collaboration
 Retrieval: ability to successfully retrieve previously studied content-relevant information during collaboration (e.g.,
Rajaram & Pereira-Pasarin, 2010).
 Inferencing: ability to generate content-relevant outputs beyond the given instructional material during collaboration
(e.g., Roscoe, 2014).
 Referencing: ability to attend to and to take up content-relevant information externalized by co-learners (e.g.,
Teasley, 1997).

For at least one comparison relevant to our review questions: report of summary data necessary to compute the
respective effect size and confidence interval.

Setting  Collaborative tasks requiring primarily cognitive performance.


 Small groups.

Study design  Empirical studies with experimental or quasi-experimental design.


 Within-designs only if counterbalanced.
 No confounding of individual preparation effects.

technical support of the database has assured the authors preparation was implemented, the control condition
that this was an unsystematic error. This study was added to involved an individual phase between several collaborative
the pool of records manually. phases or individual preparation for collaboration was only
the general context rather than the subject of examination.
A further 29 studies were excluded because they did not cor-
Full-text assessment
respond with the criteria we derived for the present work from
After eliminating duplicates, a set of 288 records were sub- the definition of individual preparation for collaboration. First,
jected to a full-text assessment. With this last step, the final we excluded studies that involved individually preparing sub-
inclusion decisions were made for the remaining studies. jects in a non-content specific manner for the subsequent col-
Some studies (n ¼ 5) were excluded as a result of unclear laboration task (e.g., in the form of a general collaboration or
reporting. We also excluded studies revealed to be non-ori- problem-solving skills training). In addition, studies were dese-
ginal (n ¼ 10), non-empirical (n ¼ 2), or non-experimen- lected which implemented individual preparations only in the
tal (n ¼ 18). form of requiring subjects to passively receive information
The largest group of excluded papers was removed rather than to deal with them in an observable manner.
because their research designs did not match our research Moreover, we excluded studies comparing different individual
questions (n ¼ 186), for example, because no individual preparation for collaboration arrangements that did not focus
38 S. MENDE ET AL.

on non-generative versus generative preparation tasks or the sample sizes as reported in the articles according to the formu-
presence versus absence of awareness induction support (see lae given in Borenstein et al. (2009, pp. 26–27). Please note that
the respective definitions in Table 1). For example, studies were these summary data could either refer to the individual or to
excluded in which all conditions involved individual prepar- the group level dependent on the unit of analysis chosen by the
ation for collaboration, but the manipulations consisted of hav- authors of the respective study (see Appendix A, column 5).
ing the group members process the same versus different Based on the standard errors, we further computed 95% confi-
contents or tasks during preparation, or prescribing different dence intervals for each effect size according to the procedure
sequences or numbers of individual generative or non-genera- provided in Nakagawa and Cuthill (2007, p. 599). One study
tive task processing phases. reported findings based on a pretest-posttest-control design
We also excluded studies that did not fulfill the methodo- (Erkens et al., 2016). Following the recommendations of the
logical criterion of internal validity regarding our research ques- What Works Clearinghouse Procedures Handbook (Institute of
tions (n ¼ 12). For instance, we excluded studies applying a Education Sciences, 2020, p. E-5) we inserted the between-con-
within-design without counter-balancing. In addition, studies dition difference of the pretest-posttest-differences in the
were excluded if the manipulated conditions not only differed numerator and the pooled posttest standard deviation in the
concerning the presence or the kind of individual preparations denominator of the Hedge’s g formula to compute the effect
but also concerning the subsequent collaboration phases. For size here. Since no pretest-posttest-correlation was reported, we
example, some studies compared a collaboration only control assumed a pre-post-intercorrelation of 0.50 for the standard
condition to a collaboration script condition, with the latter error calculation (Institute of Education Sciences, 2020, p. E-6).
involving not only an individual preparation phase but also Two further studies (Rajaram & Pereira-Pasarin, 2007, experi-
additional instructions and scaffolds in the collaboration phase. ment 1; Weldon & Bellinger, 1997, experiment 1) reported find-
In contrast, we included studies comparing conditions where ings based on mixed ANOVAS where the within subject factor
the products of individual preparations (e.g., problem solutions, effects were of interest. Here we applied the formula provided
generated explanations, or concept maps) were available to the in Borenstein et al. (2009, p. 29). For standard error calculation,
subjects during subsequent collaboration with conditions where a within-subjects’ correlation of 0.50 was assumed (cf. Institute
the products were not available during collaboration. Giving of Education Sciences, 2020). Details on how we dealt with
learners the opportunity to further use, discuss, and develop problems and exceptions regarding the calculation of effect sizes
each other’s preparation products is an inherent affordance of can be found in Appendix B.
individual preparation for collaboration.
A further set of inclusion criteria addressed the measures
applied in the studies. Generally, we accepted only quantified Data analysis
measures which were non-subjective in nature, that is, meas- Because of the relatively small amount of quite heteroge-
ures not relying on the personal opinion or evaluation of the neous studies available, conducting a meta-analysis by com-
subjects. Further, in order to be included a study had to cap- puting pooled summary effect estimates in response to our
ture at least one measure indicative of the subjects’ retrieval, research questions did not prove to be an adequate method
inferencing, or referencing performance during collaboration (Boland et al., 2017; McKenzie & Brennan, 2019; Verbeek
(see Table 1 for the definitions). In this regard, the measures et al., 2012). Consequently, we conducted a narrative synthe-
were assessed according to the principle of caution: if we sis based on a structured overview of the direction, size, and
could not clearly establish whether a measure reflects, for precision (i.e., confidence intervals) of effects obtained from
instance, the mere repetition of previously studied material in the single studies separately for each research question and
the sense of retrieval, the drawing of inferences going beyond dependent variable (cf. McKenzie & Brennan, 2019; Popay
this information, or a mixture of both, it was not further con- et al., 2006; Verbeek et al., 2012).
sidered for inclusion. A total of n ¼ 4 studies were excluded In line with the guidelines provided by Cohen (1988), we
because they did not provide any measure fulfilling our crite- considered effect sizes (Hedge’s g) of 0.20, 0.50, and 0.80 to
ria. One further study was excluded because it reported nei- be small, medium, and large, respectively. An effect was
ther a statistical analysis nor summary data relevant to our considered to be significant at a 5% significance level if the
research questions. The full-text assessment and our subse-
95% confidence interval did not include zero (Sch€ unemann
quent selection process resulted in a final set of 20 records,
et al., 2019). This has been indicated by an asterisk in
including 22 criteria-compliant experiments. An overview of
the text. For example, an effect size statement of Hedge’s
g ¼ 0.85 indicates the effect to be large and statistically sig-
the selected studies and their characteristics is given in
Appendix A. Please note that only conditions that allow for at
nificant at the 5% level.
least one comparison relevant to our research questions and
correspond to our inclusion criteria are presented.
Results
Effect size calculation RQ 1: Effects of individual preparation for collaboration
versus immediate collaboration
In general, for each dependent variable in each study the effect
size index Hedge’s g and the related standard error was calcu- Table 2 presents an overview of the effects of the studies
lated based on the means, standard deviations, and condition comparing conditions of individual preparation for
EDUCATIONAL PSYCHOLOGIST 39

collaboration to conditions of immediate collaboration awareness induction support (see columns 2 and 3).
(RQ1) separately for each dependent variable (i.e., retrieval, Moreover, we consider whether the variability of the effects
inferencing, and referencing). can be further explained by variations in additional study
features, especially the size of the learning groups and par-
ticipants’ age level (see columns 4 and 5 in Table 2). For
Retrieval studies capturing measures of retrieval performance we add-
Eight studies capture learners’ retrieval performance with itionally consider the type of the collaborative retrieval test
effect sizes ranging from large negative (g ¼ 1.13) to (indicated by superscript a or b in Table 2).
large positive (g ¼ 0.87). Three of these effect sizes are
statistically significant. One study shows a large, statistically
significant positive effect. Seven studies reveal negative RQ 2.1: The role of the preparation task type for the
effects; two with large significant effect sizes and five with effects of individual preparation for collaboration versus
small non-significant effect sizes. immediate collaboration
Retrieval. Out of the eight studies capturing measures of
retrieval, four apply a generative and four a non-generative
Inferencing preparation task (see Table 2, column 2). Among the studies
The effect sizes of the seven studies with measures on sub- employing generative preparation tasks, one reveals a large
jects’ inferencing performance range from small negative and significant positive effect. Three studies show a negative
(g ¼ 0.28) to large positive (g ¼ 1.15). Two of these effect (one significant large and two non-significant small
effect sizes are statistically significant. Six studies reveal posi- effects). All studies applying non-generative preparation
tive effects; one of large, one of moderate, and four of small tasks reveal negative effects (one significant large, one sig-
size. Two of these positive effects are statistically significant. nificant moderate, and two non-significant small effects).
One study shows a small non-significant negative effect. Hence, the comparisons reveal no discernable trend con-
cerning the role of the preparation task type (generative vs.
non-generative) for the effects of individual preparation for
Referencing
collaboration as compared to immediate collaboration on
Among the five studies measuring subjects’ referencing per-
retrieval. However, considering additional study features
formance, effect sizes range from moderate negative (g ¼
sheds more light on this picture:
0.53) to large positive (g ¼ 1.20). Two of these effect
When comparing those studies that differ in preparation
sizes are statistically significant. Three studies reveal positive
task type but are similar in terms of yielding a relatively low risk
effects; one with a large, statistically significant effect size
of retrieval strategy disruption (closed collaborative retrieval
and two with small non-significant effect sizes. Two studies
tests and group sizes of 2), studies with non-generative prepar-
show negative effects (one significant moderate and one
ation tasks show two small, not significantly negative effects
non-significant small effect size).
(Allwood et al., 2003, experiment 1; Stephenson & Wagner,
1989). Studies employing generative preparation tasks show
To sum up, the vast majority of studies on retrieval per-
one large significant positive effect (Barber et al., 2010, experi-
formance reveal negative effects, indicating individual prep-
ment 1) and one small, not significantly negative effect
aration for collaboration is inferior to immediate
(Andersson & R€ onnberg, 1997, experiment 1). A notable differ-
collaboration. With respect to subjects’ inferencing perform-
ence between the Barber et al. (2010) and the Andersson and
ance, the opposite is true: here, most of the studies show R€onnberg (1997) studies consists in the complexity of the
positive effects, favoring individual preparation for collabor- applied generative preparation task: while the subjects of the
ation over immediate collaboration. For referencing per- former study had to create sentences from given word pairs,
formance no clear trend is discernable. the participants of the latter study had to generate an associated
word for each given target word. Hence, the task applied by
RQ 2: Effects of individual preparation for collaboration Barber et al. involves a higher amount of information to be
versus immediate collaboration—role of task type and meaningfully connected, probably evoking richer and more ela-
awareness induction support borated encoding (cf. Craik & Tulving, 1975).
When comparing the studies that differ in preparation task
To address RQ 2 we compare the effects between different type but are similar in terms of yielding a higher risk of
subgroups of studies that compare conditions of individual retrieval strategy disruption (open collaborative retrieval tests
preparation for collaboration to conditions of immediate and group sizes of 3), studies with non-generative preparation
collaboration. For RQ 2.1 the effects of studies employing a tasks show moderate to large significant negative effect sizes
generative preparation task are compared to the effects of (Moreland & Myaskovsky, 2000; Weldon & Bellinger, 1997,
studies employing a non-generative preparation task, while experiment 2). The study employing a generative preparation
for RQ 2.2 the effects of studies where awareness induction task reveals a small and non-significant negative effect size
support is provided are compared to the effects of studies (Barber et al., 2012). One further study also involving a gen-
where such support is not provided. To this end, the studies erative preparation task and an open collaborative retrieval
in each dependent variable section are arranged in Table 2 test but examining dyads shows a large significant negative
according to the modifiers preparation task type and effect (Lambiotte et al., 1987). However, this effect is likely to
40 S. MENDE ET AL.

Table 2. Effects of individual preparation for collaboration (IC) versus immediate collaboration (C) on subjects’ retrieval, inferencing, and referencing
performance.
Awareness
Preparation induction
Study task type support Group size Age level Hedge’s g [95% CI]
Retrieval
Barber et al. (2010), experiment 1b Generative No 2 University students 0.87 [0.37, 1.38]
Andersson and R€onnberg (1997), Generative No 2 University students 0.17 [0.79, 0.45]
experiment 1b
Lambiotte et al. (1987)a Generative No 2 University students 1.13 [1.79, 0.46]
Barber et al. (2012)a Generative No 3 University students 0.42 [1.12, 0.29]
Allwood et al. (2003), Non-generative No 2 University students 0.14 [0.75, 0.47]
experiment 1b
Stephenson and Wagner (1989)b Non-generative No 2 University students 0.24 [1.12, 0.64]
Weldon and Bellinger (1997), Non-generative No 3 University students 1.04 [2.03, 0.05]
experiment 2a
Moreland and Myaskovsky (2000)a Non-generative No/yes 3 University students 0.31 [0.82, 0.20]c
0.61 [1.21, 0.01]d,e
for IC without awareness
induction support
0.02 [0.61, 0.56]e
for IC with awareness
induction support
Inferencing
Lam and Muldner (2017) Generative Yes 2 University students 0.49 [0.08, 0.91]
Gijlers et al. (2013) Generative Yes 2 School (primary) 0.23 [0.40, 0.86]
Huang et al. (2012) Generative Yes 3 University students 1.15 [0.24, 2.06]
Kohn et al. (2011), Generative Yes 3 University students 0.69 [0.11, 1.50]
experiment 1
Van Boxtel et al. (2000) Generative No 2 School (secondary) 0.28 [1.12, 0.56]
Korde and Paulus (2017), Generative No 3 University students 0.40 [0.31, 1.10]
experiment 1
Korde and Paulus (2017), Generative No 3 University students 0.20 [0.31, 0.70]
experiment 2
Referencing
Gijlers et al. (2013) Generative Yes 2 School (primary) 1.20 [0.50, 1.89]
Kohn et al. (2011), Generative Yes 3 University students 0.32 [0.57, 1.20]
experiment 1
Tsovaltzi et al. (2015), Generative No 2 University students 0.53 [1.03, 0.04]
experiment 3
Van Boxtel et al. (2000) Generative No 2 School (secondary) 0.34 [0.51, 1.18]
Barber et al. (2012) Generative No 3 University students 0.24 [0.94, 0.46]
Note. Positive effect sizes indicate individual preparation for collaboration (IC) is superior to immediate collaboration (C).
a
Open collaborative retrieval test (open recall). bNon-open collaborative retrieval test (cued recall or recognition). cThis effect size is considered when addressing
RQ 1. dThis effect size is considered when addressing RQ 2.1. eThese effect sizes are considered when addressing RQ 2.2.
p < 0.05.

have occurred due to issues of over-scripting the individually RQ 2.2: The role of awareness induction support for the
preparing learners, as the authors acknowledge themselves effects of individual preparation for collaboration versus
(Lambiotte et al., 1987). immediate collaboration
To summarize, in collaborative retrieval situations where Retrieval. Out of the eight studies capturing retrieval measures,
the risk of retrieval strategy disruption is low, generative only one study employs an individual preparation for collabor-
preparation tasks may render individual preparation for col- ation condition with awareness induction support (Moreland &
laboration to be superior to immediate collaboration. In sit- Myaskovsky, 2000; see Table 2, column 3). The comparison
uations where the risk of retrieval strategy disruption is between this condition and the immediate collaboration condi-
high, generative preparation tasks may diminish the negative tion results in a small negative non-significant effect size. The
effects of individual preparation for collaboration as com- same study also includes an individual preparation for collabor-
pared to immediate collaboration. The results further indi- ation condition without awareness induction support. When
cate that for this to occur also the complexity of the this condition is compared to the immediate collaboration con-
generative preparation task and the risk of over-scripting dition, a significantly negative effect of moderate size occurs.
may have to be considered. Finally, from the seven studies not involving awareness induc-
tion support, one is comparable to the Moreland and
Inferencing. We identified no studies employing non-genera- Myaskovsky study in view of preparation task type, retrieval
tive preparation tasks here. Hence, RQ 2.1 cannot be test type, and group size. This study shows a large, significantly
addressed in view of subjects’ inferencing performance. negative effect (Weldon & Bellinger, 1997, experiment 2).

Referencing. The same is true concerning subjects’ referenc- Inferencing. Out of the seven studies capturing subjects’
ing performance. inferencing performance, four apply and three do not apply
EDUCATIONAL PSYCHOLOGIST 41

Table 3. Effects resulting from comparisons between individual preparation for collaboration designs differing in the preparation task type or the provision of
awareness induction support on subjects’ retrieval, inferencing, and referencing performance.
Study Group size Age level Hedge’s g [95% CI]
Retrieval: Individual preparation with generative vs. non-generative task
Rajaram and Pereira-Pasarin (2007), experiment 1b 3 University 1.24 [0.77, 1.71]
Weldon and Bellinger (1997), experiment 1a 3 University 0.70 [0.18, 1.22]
Inferencing: Individual preparation with generative vs. non-generative task
No studies obtained
Referencing: Individual preparation with generative vs. non-generative task
No studies obtained
Retrieval: Individual preparation with vs. without awareness induction support
Moreland and Myaskovsky (2000)a 3 University 0.63 [0.01, 1.25]
Inferencing: Individual preparation with vs. without awareness induction support
Erkens et al. (2016) 2 School (high) 0.40 [0.13, 0.94]
Engelmann, Tergan et al. (2009) 3 University students 0.76 [1.39, 0.14]
Kablan (2014) 4 School (primary) 0.20 [0.26, 0.66]
Referencing: Individual preparation with vs. without awareness induction support
Erkens et al. (2016) 2 School (high) 0.66 [0.12, 1.20]
Engelmann, Tergan et al. (2009) 3 University students 1.51 [0.72, 2.31]
Engelmann et al. (2014) 3 University students 0.61 [0.01, 1.23]
Note. A positive effect size indicates an individual preparation for collaboration condition involving a generative preparation task or awareness induction support
to outperform an individual preparation for collaboration condition with a non-generative task or without awareness support, respectively.
a
Open collaborative retrieval test (open recall). bNon-open collaborative retrieval test (cued recall or recognition).
 p < 0.05.

awareness induction support. All of the studies with aware- show negative effects (one medium significant effect; one
ness induction support reveal positive effects (one significant small non-significant effect).
large, one not significant moderate, one significant small, Hence, this result pattern suggests that awareness induc-
and one non-significant small effect size). The studies with- tion support can improve the effects of individual prepar-
out awareness induction support reveal two small non-sig- ation for collaboration as compared to immediate
nificant positive effects and one small non-significant collaboration concerning subjects referencing performance.
negative effect. Considering additional study features further supports this
Hence, the results indicate that awareness induction sup- view: when comparing the studies that either provide aware-
port improves the effects of individual preparation for col- ness induction support or not, but are similar concerning
laboration as compared to immediate collaboration. group size (2 or 3) and age level (school or university stu-
Considering additional study features further supports this dents), studies involving awareness induction support sys-
view: when comparing the studies that either provide aware- tematically reveal higher effect sizes than studies without
ness induction support or not, but are similar concerning awareness induction support.
group size (2 or 3) and age level (school or university stu-
dents), studies involving awareness induction support sys-
tematically reveal higher effect sizes than studies without RQ 3: Comparing different designs of individual
awareness induction support. Further, regarding the Gijlers preparation for collaboration—effects of preparation
et al. (2013) study that reveals only a small positive effect, it task type and awareness induction support
has to be considered that the high complexity of the pre-
sented awareness information and procedure is likely to While RQ 2 examines whether generative preparation tasks
have exceeded the learners’ cognitive capacities, as suggested or awareness induction support could improve the effects of
by the authors themselves. This may partly account for the individual preparation for collaboration as compared to
small effect size. immediate collaboration, RQ 3 examines whether an indi-
To sum up, the results suggest awareness induction sup- vidual preparation for collaboration condition involving a
port further enhances the positive effects of individual prep- generative preparation task or awareness induction support
aration for collaboration as compared to immediate might outperform an individual preparation for collabor-
collaboration in view of subjects’ inferencing performance. ation condition with a non-generative task or without
For this to occur, the complexity of the presented awareness awareness support, respectively.
information may be a further factor that needs to Table 3 presents an overview of the studies which allow
be considered. for such comparisons. These studies are organized according
to the dependent variable (retrieval, inferencing, or referenc-
Referencing. Of the five studies capturing subjects’ referenc- ing) and the type of comparisons conducted: individual
ing performance, two apply and three do not apply aware- preparation with a generative versus non-generative prepar-
ness induction support. Both of the studies applying ation task (RQ 3.1) and individual preparation with versus
awareness induction support reveal positive effects (one without awareness induction support (RQ 3.2). Again, we
large significant effect; one small non-significant effect). consider whether additional study features (i.e., group size,
Among the studies without awareness induction support, participants’ age level, and retrieval test type) can further
one shows a small non-significant positive effect while two explain the effects (see Table 3).
42 S. MENDE ET AL.

RQ 3.1: The effects of generative preparation tasks summary of the partner’s individual preparation product for
Retrieval. We identified two studies measuring subjects’ the subsequent discussion. Further, it is worth considering
retrieval performance and involving individual preparation that in the study of Kablan (2014) the participants assigned
for collaboration conditions which vary concerning the to the condition without awareness induction also received
preparation task type (Rajaram & Pereira-Pasarin, 2007, the instruction to share and discuss each other’s preparation
experiment 1; Weldon & Bellinger, 1997, experiment 1). products, albeit only during the collaboration phase. Taken
Both studies reveal significant positive effects of generative together, we find mixed results on the effects of individual
as compared to non-generative individual preparation tasks, preparation for collaboration with versus without awareness
one of large and one of medium size. Both studies are simi- induction support on inferencing.
lar concerning group size and age level (groups of three uni-
versity students), but differ with respect to the collaborative Referencing. We identified three studies in which subjects’
retrieval test type: the former study applies a closed and the referencing performance is assessed and the provision of
latter study an open retrieval test. This indicates individual awareness induction support is manipulated among individual
preparation prompted by a generative task is superior to preparation for collaboration conditions. All of these studies
individual preparation prompted by a non-generative task in reveal positive effects: one significant large (Engelmann,
terms of subjects’ retrieval performance, particularly when a Tergan et al., 2009), one significant medium (Erkens et al.,
closed rather than an open collaborative retrieval test is 2016), and one non-significant medium effect (Engelmann
administered. et al., 2014). The study of Erkens et al. differs from the studies
of Engelmann et al. regarding group size (2 or 3) and age level
Inferencing. We identified no studies involving individual (high schoolers or university students). A notable difference
preparation for collaboration conditions which vary con- between the two Engelmann et al. studies consists in the
cerning the preparation task type here. implementation of the awareness induction support: while the
participants of the condition with awareness induction sup-
Referencing. The same is true concerning subjects referenc- port had access to the evolving individual preparation prod-
ing performance. ucts of their future co-learners right from the start of the
individual preparation in the study of Engelmann et al.
(2014), the participants in the study of Engelmann, Tergan
RQ 3.2: The effects of awareness induction support et al. (2009) received the products of their co-learners only
Retrieval. We identified one experimental comparison in after they had completed their own learning product.
which collaborative retrieval performance is captured and To sum up, the results indicate that individual prepar-
the provision of awareness induction support is manipulated ation for collaboration with awareness induction support is
among two individual preparation for collaboration condi- superior to individual preparation for collaboration without
tions. This comparison stems from the Moreland and awareness induction support in terms of subjects’ referenc-
Myaskovsky (2000) study and reveals a medium-sized, statis- ing performance. This is found for high schoolers in groups
tically significant positive effect, indicating individual prep- of two and university students in groups of three. A further
aration for collaboration with awareness induction support factor that may be worth considering relates to the time
is superior to individual preparation for collaboration with- point of providing learners with information about other
out awareness induction support. It is worth considering learners in terms of their (evolving) individual preparation
that this study yields a relatively high risk of retrieval strat- products: before or only after one’s own individual prepar-
egy disruption (open collaborative retrieval test and group ation product is completed.
sizes of three).

Inferencing. We obtained three studies comparing individual Discussion


preparation for collaboration with versus without awareness Individual preparation has often been suggested to benefit
induction support in which learners’ inferencing perform- collaborative learning. Yet, so far, the theoretical bases and
ance is captured. Two reveal small non-significant positive the empirical evidence have not been systematically
effects (Erkens et al., 2016; Kablan, 2014) and one reveals a addressed in previous research. This review addressed these
medium-sized significant negative effect (Engelmann, gaps by developing and testing hypotheses regarding the
Tergan et al., 2009). The three studies vary considerably claim that individual preparation is associated with an infor-
regarding the group sizes and the participants’ age levels mation processing advantage and at the same time with a
(see Table 3). Furthermore, a notable difference between the coordination disadvantage. We investigated whether individ-
Engelmann et al. and the Erkens et al. studies consists in the ual preparation for collaboration has different consequences
complexity of the awareness information and procedure: for learners’ execution of the interactive activity subtasks
While the participants in the Engelmann, Tergan et al. retrieval, inferencing, and referencing during subsequent col-
(2009) study had to deal with several complex information laboration. We further examined whether a generative indi-
sources (own concept map, co-learners maps, developing vidual preparation task and the provision of awareness
joint map) before and during collaboration, the subjects of induction support can enforce the advantage or mitigate the
Erkens et al. (2016) received a relatively concise visual disadvantage.
EDUCATIONAL PSYCHOLOGIST 43

Figure 2. Heuristic model of the conditions and effects of individual preparation on the cognitive subtasks of interactive activity in subsequent collaboration.

Our review reveals that the research on individual prep- First, if the cognitive demands of the collaboration phase
aration for collaboration is still in its infancy. Accordingly, do not at least involve inferencing but only mere retrieval,
drawing definite conclusions would be premature: the information processing advantage may be smaller than
Regarding some of our hypotheses, no final decision can the coordination disadvantage (cf. Janssen & Kirschner,
be made due to a very low number of relevant studies. In 2020; Kintsch, 2004). If only retrieval is required, it might
addition, many studies employ rather small sample sizes be more useful for the learners to collaborate from the out-
which may or may not be a reason for non-significant set in order to share retrieval strategies, which can enhance
effects. Other hypotheses cannot be examined at all due to cross-cuing and attenuate retrieval strategy disruption dur-
a lack of studies. Nevertheless, the findings of the existent ing subsequent collaboration (cf. Barber et al., 2012; Blumen
studies inspire us to reconsider our initial hypotheses and et al., 2014; Blumen & Rajaram, 2008). This explanation
to draft a preliminary model of how individual preparation receives some support from the finding that the negative
may affect interactive activities in subsequent collaboration effects of individual preparation for collaboration on
(see Figure 2). retrieval are more pronounced in studies where the risk of
retrieval strategy disruption is higher due to higher group
sizes and/or open collaborative retrieval tests (cf. Marion &
Summary of theoretical considerations and empirical Thorley, 2016). Conversely, the positive effects of individual
findings: A preliminary model preparation for collaboration in terms of inferencing tend to
be more pronounced for larger groups where more co-learn-
We claim within our preliminary model (Figure 2) that indi- ers could harm initial information processing through dis-
vidual preparation per se does not automatically promote ruption and overload.
interactive activities since not all cognitive subtasks of inter- Second, the information processing advantage of individ-
active activities are fostered. We further claim that the net ual preparation may only exceed its coordination disadvan-
effect of individual preparation for a collaboration depends tage if the preparation task is sufficiently complex and
on whether its information processing advantage or its cognitively resource demanding. This interpretation is sup-
coordination disadvantage prevails. ported by two findings of the review. First, the results in
The results of this review suggest that, when compared to relation to RQ 2.1 and 3.1 indicate that generative as com-
immediate collaboration, individual preparation for collabor- pared to non-generative preparation tasks improve the
ation reduces learners’ retrieval performance, fosters their effects of individual preparation for collaboration on learn-
inferencing performance, and has no clear impact on refer- ers’ collaborative retrieval performance. Second, in the
encing (RQ 1). Since inferencing requires successful course of investigating RQ 2.1 we found some hints suggest-
retrieval, this may seem somewhat contradictory. At least ing that not only the type of the preparation task (generative
two possibly complementary mechanisms could explain this vs. non-generative) but also the generative task complexity—
pattern of results. in terms of the amount of to-be-memorized information to
44 S. MENDE ET AL.

be meaningfully connected—may play a role in this regard hinder the drawing of inferences during collaboration
(see Andersson & R€ onnberg, 1997; Barber et al., 2010). (cf. Engelmann, Tergan et al., 2009; Gijlers et al., 2013).
From prior research, we know that a higher amount of To conclude, based on the review we include the follow-
information to be linked improves memory, but also ing claims into our preliminary model of the effects of indi-
increases the burdens on learners’ cognitive capacities (Craik vidual preparation for subsequent collaboration: the net
& Tulving, 1975; Kalyuga, 2007; Kintsch, 2004; van effect of individual preparation for collaboration increases
Merri€enboer & Sweller, 2005). Hence, it may be that indi-
vidual preparation for collaboration is only superior to  with the degree to which the information processing
immediate collaboration when a task requires the generation advantage is actually used during preparation and actually
of so many meaningful connections that only individually useful for the subsequent collaboration and
preparing learners are able to invest the necessary cognitive  with the degree to which the coordination disadvantage
resources, while immediately collaborating learners are not. is mitigated.
Among the studies comparing individual preparation for
collaboration to immediate collaboration, only the studies The use of the information processing advantage during
capturing retrieval indicators vary concerning the prepar- the preparation seems to depend on the task type and pos-
ation task type while all studies capturing inferencing indica- sibly further task characteristics such as the task complexity,
tors employ generative preparation tasks (see Figure 2). whereas the cognitive requirements of the collaboration task
Hence, we cannot further disentangle whether the mainly seem to influence the usefulness of the information process-
negative retrieval and the mainly positive inferencing effects ing advantage for the subsequent collaboration. To mitigate
of individual preparation are due to the preparation task the coordination disadvantage, the provision of awareness
characteristics, due to the complexity of the collaborative induction support per se, as well as further parameters such
learning requirements, or a mixture of both. as the timing and the complexity of the awareness informa-
In general, the review results are also in line with the tion presentation may play a role. In addition, the net effect
hypothesis that reducing the coordination disadvantage of of individual preparation for collaboration may be influ-
individual preparation via the provision of awareness induc- enced by further variables including learner (e.g., age level,
tion support may foster referencing, in turn raising the ben- prior knowledge, collaboration skills) and group characteris-
efits of collaboration for retrieval and inferencing (RQ 2.2 tics (e.g., group size, knowledge divergence).
and 3.2). However, some restrictions and specifications have
to be considered here.
Limitations
First, we found some hints that the timing of awareness
induction support may have an impact on its effectivity This work is subject to several limitations that stem from
regarding learners referencing performance (see Engelmann, methodological issues concerning a) the single studies
Tergan et al., 2009; Engelmann et al., 2014): when provided reviewed, b) the review itself as well as c) the generalizability
after learners have completed their own individual prepar- of the reviewed results.
ation the group may possess a higher divergence of content-
related perspectives and ideas than when learners are
granted with access to others’ evolving products or solutions Methodological issues at the level of the single studies
First, the sample sizes of the reviewed studies are often rela-
right from the start of the preparation. Hence, in the former
tively small, making the effect size confidence intervals quite
case learners may perceive a higher interdependence so that
wide. This is likely to increase the risk of false positives and
mutual referencing is encouraged to a higher degree (cf.
false negatives (e.g., Button et al., 2013).
Janssen & Kirschner, 2020; Jorczak, 2011).
Second, some of the comparisons between conditions
Second, we obtained only one study providing evidence
involving collaboration with versus without individual prepar-
regarding the hypothesis that awareness induction support
ation are potentially confounded. That is, the condition with-
might improve the individual preparation effects in view of
out individual preparation sometimes does not include the
learners’ collaborative retrieval performance (Moreland &
task required in the individual preparation. For example, Van
Myaskovsky, 2000). Although this study supports our
Boxtel et al. (2000) had their participants construct a learning
hypothesis, more evidence is needed for clear confirmation.
product in dyads either immediately (immediate collaboration
Third, the results suggest awareness induction improves the
condition) or after individually making a design for the learn-
positive effects of individual preparation for collaboration as
ing product (individual preparation for collaboration condi-
compared to immediate collaboration in view of learners’
tion). Thus, it cannot clearly be concluded whether the effects
inferencing performance (RQ 2.2). However, with regard to
are due to the individual preparation itself, the task to make
the studies comparing individual preparation arrangements
a design that the immediate collaboration participants never
with and without awareness induction support (RQ 3.2), the
received, or a mixture of both.
results reveal no clear picture. Here, the findings indicate
that a further factor of the individual preparation design
may play a role: the complexity of the presented awareness Methodological issues at the level of the review
information and instructions. When too complex, the sub- At least two limitations have to be mentioned here. First, we
jects’ cognitive capacities may be overstrained and, thus, considered not only direct process measures but also
EDUCATIONAL PSYCHOLOGIST 45

post-collaborative measures to reflect subjects’ retrieval, Directions for future research


inferencing, or referencing performance during collabor-
The present review helped us to formulate a preliminary model
ation. Post-collaborative measures can only serve as indirect
of whether and when individual preparation may foster the cog-
indicators for the degree to which related collaborative proc-
nitive subtasks of interactive activity during subsequent collab-
esses have occurred. For example, better achievement in a
oration. More research is needed to examine its hypotheses and
transfer posttest may allow for the conclusion that more
to investigate the role of additional factors that are indicated to
inferences were drawn during the previous learning process
be also potentially relevant by the present results.
(e.g., Chi et al., 1994; Kintsch, 2004). Yet, without measures
First, in some cases none, and in many cases, only a few
directly captured from the collaboration process, it cannot
studies were obtained that dealt with the independent and
be fully determined whether such inferences were actually
dependent variables relevant to our research questions. Thus,
carried out and when they were drawn (i.e., during collabor-
the current research state does not allow disentangling the
ation or another time such as during individual prepar-
effects of task type and awareness induction support and does
ation). Unfortunately, our literature search revealed only a
not provide enough insights into the effects of a combination
few studies capturing process measures.
of these design features. To address these gaps on a secondary
Second, in some cases, the reported effect sizes and confi-
research level, further studies are needed that compare indi-
dence intervals could not directly be computed due to missing
vidual preparation for collaboration to immediate collabor-
summary data in the primary studies. We followed established
ation and (a) capture retrieval and apply awareness induction
recommendations to make assumptions about unknown sta-
support (see Table 2, third column) or (b) capture inferencing
tistics and applied the best possible methods to estimate the
and referencing and apply non-generative preparation tasks
indices in case of missing values as exact as possible based on
(see Table 2, second column). Furthermore, studies are
the given information (see Appendix B). However, it still has
to be acknowledged that some results are based on approxi- needed that compare individual preparations according to
mations of the true values, potentially limiting their validity. generative tasks with individual preparations according to
non-generative tasks in view of learners’ inferencing and
referencing performance (see Table 3). In addition, studies
Generalizability of the review results employing a 2  2  2 design, varying (a) individual prepar-
At least two issues may limit the generalizability of our results. ation for collaboration versus immediate collaboration, (b)
First, half of the studies included in this review employ authen- generative versus non-generative preparation tasks, and (c)
tic educational materials and relatively complex preparation the provision of awareness induction support would allow for
and collaboration tasks (e.g., concept mapping, application a systematic examination of the role of each independent vari-
questions, essay writing). The other half of the studies—stem- able, as well as their interactions at the primary research level.
ming primarily from group brainstorming and collaborative Second, the reviewed studies measure subjects’ performance
memory research—involve more artificial tasks (e.g., memoriza- in terms of the interactive activity subtasks, but not as an inte-
tion and retrieval of word lists). Such tasks are not necessarily grated behavior. Hence, so far it is not possible to draw con-
group-worthy (e.g., Cohen, 1994; Janssen & Kirschner, 2020) clusions concerning the question of whether and under what
and the respective results are not directly generalizable to the conditions an individual preparation might foster the execution
effects of individual preparation for collaborative learning at of interactive activities as such during subsequent collabor-
school, university, or the workplace. However, these studies ation. Consequently, studies are needed which capture behav-
apply highly controlled experiments suited to carve out the ioral data from the collaboration process according to coding
basic factors and cognitive mechanisms underlying the per- schemes reflecting the exact definition of interactive activities
formance concerning important cognitive subtasks of meaning- (Deiglmayr et al., 2015; Mende et al., 2017).
ful collaborative learning processes. For instance, although Third, future (meta-)studies could look at the effects of
group retrieval per se may be an artificial, not group worthy individual preparation for collaboration not only through a
task, insights into the basic factors facilitating or hindering cognitive but also a motivational or metacognitive lens by
group retrieval can contribute to clarifying when and why indi- considering further dependent variables. For example, an
vidual preparation may (not) foster interactive activities during interesting question would be how individual preparation
collaboration in authentic learning settings. affects the learners’ capabilities to regulate their own and
Second, only four of the reviewed studies investigate sam- each other’s learning on a metacognitive level and how this
ples of K-12 students while the remaining studies examine relates to mutual cognitive knowledge construction in terms
university students. Hence, based on the given evidence we of interactive activities (e.g., Miller & Hadwin, 2015).
cannot draw reliable conclusions about whether the effects Fourth, our results indicate that further factors may play
of individual preparation for collaboration are comparable a role regarding the effects of individual preparation for col-
for K-12 and university students or may vary as a function laboration. In future studies employing the above mentioned
of, for instance, developmental factors in terms of intellec- 2  2  2 design, the collaboration task demands could be
tual ability or social competence (cf. Wentzel & Watkins, varied to examine whether individual preparation and its
2002). A similar issue meets the size of the collaborating design features may be more or less effective in dependence
groups: except one, all studies employ either dyads or triads. of what the learners have to subsequently achieve together.
Hence, the present results allow no conclusions regarding The same applies to potentially relevant parameters of the
individual preparation effects for larger groups. individual preparation design: the amount of information to
46 S. MENDE ET AL.

be linked in order to process a generative preparation task, for instance, in terms of generating new conclusions, connec-
the complexity of the presented awareness information, and tions, or ideas which could then be discussed and further
the timing of providing awareness information. Hence, developed during later collaboration (i.e., generative prepar-
besides the question of whether a generative preparation task ation task), and (c) have a phase implemented in which the
or awareness induction support is provided, it may also single members of the future group inspect each other’s indi-
matter how these support strategies are themselves designed. vidual preparation products before starting collaboration (i.e.,
Fifth, future research should more explicitly relate individual awareness induction support). Teachers can use such strategic
preparation to other support strategies and tools. For example, suggestions in two respects: First, they can apply them to
collaboration scripts provide learners with external rules to orchestrate effective collaboration in the classroom in terms
guide them to conducive collaborative practices (e.g., Fischer, of fostering interactive activities to promote deep knowledge
Kollar, Stegmann, & Wecker, 2013). As such, scripts can pre- acquisition. Second, they can train their students to apply the
scribe individual preparation for collaboration but can (also) suggestions in terms of effective collaboration strategies and,
provide more fine-grained scaffolds to directly support the col- by doing so, empower them for the future as successful team
laboration phase. Similarly, external representation tools can be players in modern working environments.
applied during collaboration to help learners off-loading the
cognitive demands (e.g., Dillenbourg & Betrancourt, 2006; Van
Bruggen et al., 2002) but could also be provided already during Acknowledgments
an individual preparation. Hence, future research could exam- We would like to express our sincere gratitude to the editor Professor
ine whether and under what conditions (a) individual prepar- Kathryn Wentzel as well as three anonymous reviewers for their critical
ation or direct collaboration support might be superior and/or reading of former versions of the manuscript and for challenging us to
(b) a combination might be best. deepen and strengthen our arguments with their very valuable critiques
and suggestions.
Finally, it cannot be taken for granted that the effects of indi-
vidual preparation for collaboration and the mentioned prepar-
ation design parameters are the same regardless of individual References
learner characteristics (e.g., Mende et al., 2017). In the field of
Alfieri, L., Nokes-Malach, T. J., & Schunn, C. D. (2013). Learning
collaboration scripting research, it is frequently argued that a through case comparisons: A meta-analytic review. Educational
given degree of guidance may be adequate to enable some learn- Psychologist, 48(2), 87–113. https://doi.org/10.1080/00461520.2013.775712
ers to execute the desired learning or collaboration activity, while Allwood, C. M., Granhag, P. A., & Johansson, M. (2003). Increased
it may be insufficient or redundant for other learners dependent, realism in eyewitness confidence judgements: The effect of dyadic
for instance, on the prior knowledge or the collaboration skills collaboration. Applied Cognitive Psychology, 17(5), 545–561. https://
they already possess (e.g., Kollar et al., 2006; Vogel et al., 2017). doi.org/10.1002/acp.888
Anderson, J. R., & Pirolli, P. L. (1984). Spread of activation. Journal of
However, especially the role of learners’ prior knowledge is Experimental Psychology: Learning, Memory, and Cognition, 10(4),
under-researched in this concern and thus deserves further 791–798. https://doi.org/10.1037/0278-7393.10.4.791
research (cf. Vogel et al., 2017). Moreover, whether individual Andersson, J., & R€ onnberg, J. (1997). Cued memory collaboration:
preparation with a certain degree of support would be adequate Effects of friendship and type of retrieval cue. European Journal of
for a learner with a given prior knowledge level should also Cognitive Psychology, 9(3), 273–287. https://doi.org/10.1080/713752558
Andrews, J. J., & Rapp, D. N. (2015). Benefits, costs, and challenges of
depend on the complexity of the subsequent collaborative task
collaboration for learning and memory. Translational Issues in
(Janssen & Kirschner, 2020; Kirschner et al., 2018). Psychological Science, 1(2), 182–191. https://doi.org/10.1037/tps0000025
Arvaja, M., H€akkinen, P., & Kankaanranta, M. (2008). Collaborative
learning and computer-supported collaborative learning environ-
Conclusions and significance ments. In J. Voogt & G. Knezek (Eds.), International handbook of
The findings and conclusions of this work are relevant to information technology in primary and secondary education (pp.
267–279). Springer.
both researchers and teachers. Adopting a researcher’s per- Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed
spective, this review adds a suggestion for a theoretical system and its control processes. In K. W. Spence & J. T. Spence
model elaborating on the cognitive mechanisms which may (Eds.), The psychology of learning and motivation - Advances in
explain when and how individual preparation for collabor- research and theory (Vol. 2, pp. 89–195). Academic Press.
ation affects the execution of interactive activities during Barber, S. J., Rajaram, S., & Aron, A. (2010). When two is too many:
subsequent collaboration. It further highlights research strat- Collaborative encoding impairs memory. Memory & Cognition,
38(3), 255–264. https://doi.org/10.3758/MC.38.3.255
egies for investigations necessary to confirm and to extend Barber, S. J., Rajaram, S., & Fox, E. B. (2012). Learning and remember-
our preliminary conclusions. ing with others: The key role of retrieval in shaping group recall
For teachers, it is of interest whether and under what con- and collective memory. Social Cognition, 30(1), 121–132. https://doi.
ditions individual preparation should be included in their org/10.1521/soco.2012.30.1.121
instructional toolbox. Some tentative suggestions can be for- Barron, B. (2003). When smart groups fail. Journal of the Learning
mulated: (a) apply individual preparation only if the subse- Sciences, 12(3), 307–359. https://doi.org/10.1207/S15327809JLS1203_1
Basden, B., Basden, D., Bryner, S., & Thomas, R. (1997). A comparison
quent collaboration requirements involve deep processing of group and individual remembering: Does collaboration disrupt
(i.e., inferencing) of complex information not already well retrieval strategies? Journal of Experimental Psychology. Learning,
known or understood, (b) use the individual preparation time Memory, and Cognition, 23(5), 1176–1189. https://doi.org/10.1037/
to ask students to go beyond the given instructional material, 0278-7393.23.5.1176
EDUCATIONAL PSYCHOLOGIST 47

Basden, B., Reysen, M. B., & Basden, D. (2002). Transmitting false Behavior, 11(6), 671–684. https://doi.org/10.1016/S0022-
memories in social groups. The American Journal of Psychology, 5371(72)80001-X
115(2), 211–231. https://doi.org/10.2307/1423436 Craik, F. I. M., & Tulving, E. (1975). Depth of processing and the
Berkowitz, M. W., Althof, W., Turner, V. D., & Bloch, D. (2008). retention of words in episodic memory. Journal of Experimental
Discourse, development, and education. In F. K. Oser & W. Psychology: General, 104(3), 268–294. https://doi.org/10.1037/0096-
Veugelers (Eds.), Getting involved: Global citizenship development 3445.104.3.268
and sources of moral values (pp. 189–201). Sense Publishers. Cress, U., & Kimmerle, J. (2008). A systemic and cognitive view on
Best, R. M., Rowe, M., Ozuru, Y., & McNamara, D. S. (2005). Deep- collaborative knowledge building with wikis. International Journal of
level comprehension of science texts: The role of the reader and the Computer-Supported Collaborative Learning, 3(2), 105–122. https://
text. Topics in Language Disorders, 25(1), 65–83. https://doi.org/10. doi.org/10.1007/s11412-007-9035-z
1097/00011363-200501000-00007 Cuc, A., Koppel, J., & Hirst, W. (2007). Silence is not golden: A case
Blumen, H. M., & Rajaram, S. (2008). Influence of re-exposure and for socially shared retrieval-induced forgetting. Psychological Science,
retrieval disruption during group collaboration on later individual recall. 18(8), 727–733. https://doi.org/10.1111/j.1467-9280.2007.01967.x
Memory, 16(3), 231–244. https://doi.org/10.1080/09658210701804495 Damnik, G., Proske, A., Narciss, S., & K€ orndle, H. (2013). Informal
Blumen, H. M., Young, K. E., & Rajaram, S. (2014). Optimizing group learning with technology: The effects of self-constructing externaliza-
collaboration to improve later retention. Journal of Applied Research tions. The Journal of Educational Research, 106(6), 431–440. https://
in Memory and Cognition, 3(4), 244–251. https://doi.org/10.1016/j. doi.org/10.1080/00220671.2013.832978
jarmac.2014.05.002 Deiglmayr, A., & Spada, H. (2010a). Training for fostering knowledge
Boland, A., Cherry, G., & Dickson, R. (2017). Doing a systematic co-construction from collaborative inference-drawing. Learning and
review. Sage. Instruction, 21(3), 441–451. https://doi.org/10.1016/j.learninstruc.
Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2009). 2010.06.004
Introduction to meta-analysis. John Wiley & Sons. Deiglmayr, A., & Spada, H. (2010b). Collaborative problem-solving
Bromme, R., Hesse, F. W., & Spada, H. (2005). Barriers and biases in with distributed information: The role of inferences from inter-
computer-mediated knowledge communication. And how they may be dependent information. Group Processes & Intergroup Relations,
overcome. Springer. 13(3), 361–378. https://doi.org/10.1177/1368430209342259
Brown, V., Tumeo, M., Larey, T. S., & Paulus, P. B. (1998). Modeling Deiglmayr, A., Rummel, N., & Loibl, K. (2015, June 7–11). The media-
cognitive interactions during group brainstorming. Small Group ting role of interactive learning activities in CSCL: An INPUT-
Research, 29(4), 495–526. https://doi.org/10.1177/1046496498294005 PROCESS-OUTCOME Model [Paper presentation]. The 11th
Button, K. S., Ioannidis, J. P. A., Mokrysz, C., Nosek, B. A., Flint, J., Conference on Computer Supported Collaborative Learning (CSCL),
Robinson, E. S. J., & Munaf o, M. R. (2013). Power failure: Why Gothenburg, Sweden.
Dennis, A. R., & Williams, M. L. (2005). A meta-analysis of group size
small sample size undermines the reliability of neuroscience. Nature
effects in electronic brainstorming: More heads are better than one.
Reviews. Neuroscience, 14(5), 365–376. https://doi.org/10.1038/nrn3475
International Journal of e-Collaboration, 1(1), 24–42. https://doi.org/
Chase, C. C., Marks, J., Malkiewich, L. J., & Connolly, H. (2019). How
10.4018/9781599043937.ch013 https://doi.org/10.4018/jec.2005010102
teacher talk guidance during Invention activities shapes students’
Diehl, M., & Stroebe, W. (1987). Productivity loss in brainstorming
cognitive engagement and transfer. International Journal of STEM
groups: Toward the solution of a riddle. Journal of Personality and
Education, 6(14), 1–22.
Social Psychology, 53(3), 497–509. https://doi.org/10.1037/0022-3514.
Chi, M. T. H. (2000). Self-explaining: The dual processes of generating
53.3.497
inference and repairing mental models. In R. Glaser (Ed.), Advances in
Dillenbourg, P. (1999). What do you mean by collaborative learning?
instructional psychology (pp. 161–238). Lawrence Erlbaum Associates
In P. Dillenbourg (Ed.), Collaborative-learning: Cognitive and com-
Publishers. putational approaches (pp. 1–19). Elsevier.
Chi, M. T. H. (2009). Active-constructive-interactive: A conceptual Dillenbourg, P., & Betrancourt, M. (2006). Collaboration load. In J.
framework for differentiating learning activities. Topics in Elen & R. E. Clark (Eds.), Handling complexity in learning environ-
Cognitive Science, 1(1), 73–105. https://doi.org/10.1111/j.1756-8765. ments: Research and theory (pp. 142–163). Elsevier.
2008.01005.x Dillenbourg, P., J€arvel€a, S., & Fischer, F. (2009). The evolution of
Chi, M. T. H., & Menekse, M. (2015). Dialogue patterns in peer collab- research on computer-supported collaborative learning: From design
oration that promote learning. In L. B. Resnick, C. Asterhan, & S. to orchestration. In N. Balacheff, S. Ludvigsen, T. de Jong, A.
Clarke (Eds.), Socializing intelligence through academic talk and dia- Lazonder, & S. Barnes (Eds.), Technology-enhanced learning:
logue (pp. 263–274). American Educational Research Association. Principles and products (pp. 3–19). Springer.
Chi, M. T. H., & Wylie, R. (2014). The ICAP Framework: Linking cogni- Dugosh, K. L., Paulus, P. B., Roland, E. J., & Yang, H. C. (2000).
tive engagement to active learning outcomes. Educational Psychologist, Cognitive stimulation in brainstorming. Journal of Personality and
49(4), 219–243. https://doi.org/10.1080/00461520.2014.965823 Social Psychology, 79(5), 722–735. https://doi.org/10.1037/0022-3514.
Chi, M. T. H., Adams, J., Bogusch, E. B., Bruchok, C., Kang, S., 79.5.722
Lancaster, M., Levy, R., Li, N., McEldoon, K. L., Stump, G. S., Engelmann, T., Dehler, J., Bodemer, D., & Buder, J. (2009). Knowledge
Wylie, R., Xu, D., & Yaghmourian, D. L. (2018). Translating the awareness in CSCL: A psychological perspective. Computers in
ICAP theory of cognitive engagement into practice. Cognitive Human Behavior, 25(4), 949–960. https://doi.org/10.1016/j.chb.2009.
Science, 42(6), 1777–1832. https://doi.org/10.1111/cogs.12626 04.004
Chi, M. T. H., De Leeuw, N., Chiu, M.-H., & Lavancher, C. (1994). Engelmann, T., Kozlov, M. D., Kolodziej, R., & Clariana, R. B. (2014).
Eliciting self-explanations improves understanding. Cognitive Fostering group norm development and orientation while creating
Science, 18(3), 439–477. https://doi.org/10.1207/s15516709cog1803_3 awareness contents for improving net-based collaborative problem
Cohen, E. (1994). Restructuring the classroom: Conditions for product- solving. Computers in Human Behavior, 37, 298–306. https://doi.org/
ive small groups. Review of Educational Research, 64(1), 1–35. 10.1016/j.chb.2014.04.052
https://doi.org/10.3102/00346543064001001 Engelmann, T., Tergan, S. O., & Hesse, F. (2009). Evoking knowledge
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. and information awareness for enhancing computer-supported col-
Erlbaum. laborative problem solving. The Journal of Experimental Education,
Cowan, N. (2014). Working memory underpins cognitive development, 78(2), 268–290. https://doi.org/10.1080/00220970903292850
learning, and education. Educational Psychology Review, 26(2), Erkens, M., Bodemer, D., & Hoppe, H. U. (2016). Improving collabora-
197–223. https://doi.org/10.1007/s10648-013-9246-y tive learning in the classroom: Text mining based grouping and rep-
Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A frame- resenting. International Journal of Computer-Supported Collaborative
work for memory research. Journal of Verbal Learning and Verbal Learning, 11(4), 387–415. https://doi.org/10.1007/s11412-016-9243-5
48 S. MENDE ET AL.

Fiorella, L., & Mayer, R. E. (2016). Eight ways to promote generative Designing for CSCL regulation tools. Educational Technology
learning. Educational Psychology Review, 28(4), 717–741. https://doi. Research and Development, 63(1), 125–142. https://doi.org/10.1007/
org/10.1007/s10648-015-9348-9 s11423-014-9358-1
Fischer, F., Kollar, I., Stegmann, K., & Wecker, C. (2013). Toward a Jeong, H., & Hmelo-Silver, C. E. (2016). Seven affordances of com-
script theory of guidance in computer-supported collaborative learn- puter-supported collaborative learning: How to support collaborative
ing. Educational Psychologist, 48(1), 56–66. https://doi.org/10.1080/ learning? How can technologies help? Educational Psychologist,
00461520.2012.748005 51(2), 247–265. https://doi.org/10.1080/00461520.2016.1158654
Fischer, F., Kollar, I., Stegmann, K., Wecker, C., Zottmann, J., & Johnson, D. W., Johnson, R. T., & Stanne, M. B. (1989). Impact of goal
Weinberger, A. (2013). Collaboration scripts in computer-supported and resource interdependence on problem-solving success. The
collaborative learning. In C. E. Hmelo-Silver, C. A. Chinn, C. K. K. Journal of Social Psychology, 129(5), 621–629. https://doi.org/10.
Chan, & A. O’Donnell (Eds.), The international handbook of collab- 1080/00224545.1989.9713780
orative learning (pp. 403–419). Routledge. https://doi.org/10.4324/ Jorczak, R. L. (2011). An information processing perspective on diver-
9780203837290.ch23 gence and convergence in collaborative learning. International
Fritz, C. O., Morris, P. E., & Richler, J. J. (2012). Effect size estimates: Journal of Computer-Supported Collaborative Learning, 6(2),
Current use, calculations, and interpretation. Journal of Experimental 207–221. https://doi.org/10.1007/s11412-010-9104-6
Psychology. General, 141(1), 2–18. https://doi.org/10.1037/a0024338 Kablan, Z. (2014). Comparison of individual answer and group answer
Gijlers, H., & de Jong, T. (2009). Sharing and confronting propositions with and without structured peer assessment. Research in Science &
in collaborative inquiry learning. Cognition and Instruction, 27(3), Technological Education, 32(3), 251–262. https://doi.org/10.1080/
239–268. https://doi.org/10.1080/07370000903014352 02635143.2014.931840
Gijlers, H., Weinberger, A., van Dijk, A. M., Bollen, L., & van Kalyuga, S. (2007). Enhancing instructional efficiency of interactive E-
Joolingen, W. (2013). Collaborative drawing on a shared digital canvas learning environments: A cognitive load perspective. Educational
in elementary science education: The effects of script and task awareness Psychology Review, 19(3), 387–399. https://doi.org/10.1007/s10648-007-
support. International Journal of Computer-Supported Collaborative 9051-6
Learning, 8(4), 427–453. https://doi.org/10.1007/s11412-013-9180-5 Kapur, M. (2015). Learning from productive failure. Learning: Research
Grabowski, B. L. (2004). Generative learning contributions to the and Practice, 1(1), 51–65. https://doi.org/10.1080/23735082.2015.1002195
design of instruction and learning. In D. J. Jonassen (Ed.), King, A. (1999). Discourse patterns for mediating peer learning. In
Handbook of research on educational communications and technology A. M. O’Donnell & A. King (Eds.), Cognitive perspectives on peer
(Issue 2, pp. 719–743). Lawrence Erlbaum Associates, Publishers. learning (pp. 87–115). Lawrence Erlbaum Associates Publishers.
Graesser, A. C., Fiore, S. M., Greiff, S., Andrews-Todd, J., Foltz, P. W., Kintsch, W. (1998). Comprehension: A paradigm for cognition.
& Hesse, F. W. (2018). Advancing the science of collaborative prob-
Cambridge University Press.
lem solving. Psychological Science in the Public Interest, 19(2), 59–92. Kintsch, W. (2004). The Construction – Integration Model of text
https://doi.org/10.1177/1529100618808244
comprehension and its implications for instruction. In R. Unrau &
Higgins, J., Li, T., & Deeks, J. (2019). Choosing effect measures and
N. Rudell (Eds.), Theoretical models and processes of reading (5th
computing estimates of effect. In J. Higgins, J. Thomas, J. Chandler,
ed., pp. 1270–1329). International Reading Association.
M. Cumpston, T. Li, M. Page, & V. Welch (Eds.), Cochrane hand-
Kirschner, F., Paas, F., & Kirschner, P. (2009). A cognitive load
book for systematic reviews of interventions version 6.0 (updated July
approach to collaborative learning: United brains for complex tasks.
2019). Chochrane.
Educational Psychology Review, 21(1), 31–42. https://doi.org/10.1007/
Hogan, K., Nastasi, B. K., & Pressley, M. (1999). Discourse patterns
s10648-008-9095-2
and collaborative scientific reasoning in peer and teacher-guided dis-
Kirschner, P., & Erkens, G. (2013). Toward a framework for CSCL
cussions. Cognition and Instruction, 17(4), 379–432. https://doi.org/
research. Educational Psychologist, 48(1), 1–8. https://doi.org/10.
10.2307/3233840 https://doi.org/10.1207/S1532690XCI1704_2
Hollingshead, A. B. (2001). Cognitive interdependence and convergent 1080/00461520.2012.750227
expectations in transactive memory. Journal of Personality and Kirschner, P., Sweller, J., Kirschner, F., & Zambrano, J. R. (2018).
Social Psychology, 81(6), 1080–1089. https://doi.org/10.1037/0022- From cognitive load theory to collaborative cognitive load theory.
3514.81.6.1080 International Journal of Computer-Supported Collaborative Learning,
Huang, H. W., Wu, C. W., & Chen, N. S. (2012). The effectiveness of 13(2), 213–233. https://doi.org/10.1007/s11412-018-9277-y
using procedural scaffoldings in a paper-plus-smartphone collabora- Kohn, N. W., & Smith, S. M. (2011). Collaborative fixation: Effects of
tive learning context. Computers & Education, 59(2), 250–259. others’ ideas on brainstorming. Applied Cognitive Psychology, 25(3),
https://doi.org/10.1016/j.compedu.2012.01.015 359–371. https://doi.org/10.1002/acp.1699
Institute of Education Sciences. (2020). What Works Clearinghouse: Kohn, N. W., Paulus, P. B., & Choi, Y. (2011). Building on the ideas of
Procedures handbook (Version 4.1). IES. others: An examination of the idea combination process. Journal of
Janssen, J., & Bodemer, D. (2013). Coordinated computer-supported col- Experimental Social Psychology, 47(3), 554–561. https://doi.org/10.
laborative learning: Awareness and awareness tools. Educational 1016/j.jesp.2011.01.004
Psychologist, 48(1), 40–55. https://doi.org/10.1080/00461520.2012.749153 Kolfschoten, G., French, S., & Brazier, F. (2014). A discussion of the
Janssen, J., & Kirschner, P. (2020). Applying collaborative cognitive load cognitive load in collaborative problem-solving: The decision-mak-
theory to computer-supported collaborative learning: Towards a ing phase. EURO Journal on Decision Processes, 2(3–4), 257–280.
research agenda. Educational Technology Research and Development. https://doi.org/10.1007/s40070-014-0034-9
68, 783–805. https://doi.org/10.1007/s11423-019-09729-5 Kollar, I., Fischer, F., & Hesse, F. (2006). Collaboration Scripts – A
Janssen, J., Erkens, G., & Kanselaar, G. (2007). Visualization of agree- conceptual analysis. Educational Psychology Review, 18(2), 159–185.
ment and discussion processes during computer-supported collab- https://doi.org/10.1007/s10648-006-9007-2
orative learning. Computers in Human Behavior, 23(3), 1105–1125. Korde, R., & Paulus, P. B. (2017). Alternating individual and group idea
https://doi.org/10.1016/j.chb.2006.10.005 generation: Finding the elusive synergy. Journal of Experimental Social
Janssen, J., Kirschner, F., Erkens, G., Kirschner, P. A., & Paas, F. Psychology, 70, 177–190. https://doi.org/10.1016/j.jesp.2016.11.002
(2010). Making the black box of collaborative learning transparent: Lam, R. (2019). What students do when encountering failure in collab-
Combining process-oriented and cognitive load approaches. orative tasks. Npj Science of Learning, 4(1), 1–11. https://doi.org/10.
Educational Psychology Review, 22(2), 139–154. https://doi.org/10. 1038/s41539-019-0045-1
1007/s10648-010-9131-x Lam, R., & Kapur, M. (2017). Preparation for future collaboration:
J€arvel€a, S., Kirschner, P. A., Panadero, E., Malmberg, J., Phielix, C., Cognitively preparing for learning from collaboration. Journal of
Jaspers, J., Koivuniemi, M., & J€arvenoja, H. (2015). Enhancing Experimental Education, 86(4), 546–559. https://doi.org/10.1080/
socially shared regulation in collaborative learning groups: 00220973.2017.1386156
EDUCATIONAL PSYCHOLOGIST 49

Lam, R., & Muldner, K. (2017). Manipulating cognitive engagement in learning. Educational Psychology Review, 27(4), 645–656. https://doi.
preparation-to-collaborate tasks and the effects on learning. org/10.1007/s10648-015-9312-8
Learning and Instruction, 52, 90–101. https://doi.org/10.1016/j.lear- Noroozi, O., Teasley, S. D., Biemans, H. J. A., Weinberger, A., &
ninstruc.2017.05.002 Mulder, M. (2013). Facilitating learning in multidisciplinary groups
Lambiotte, J. G., Dansereau, D. F., Rocklin, T. R., Fletcher, B., with transactive CSCL scripts. International Journal of Computer-
Hythecker, V. I., Larson, C. O., & O’Donnell, A. M. (1987). Supported Collaborative Learning, 8(2), 189–223. https://doi.org/10.
Cooperative learning and test taking: Transfer of skills. 1007/s11412-012-9162-z
Contemporary Educational Psychology, 12(1), 52–61. https://doi.org/ O’Donnell, A., & O’Kelly, J. (1994). Learning from peers: Beyond the
10.1016/S0361-476X(87)80038-3 rhetoric of positive results. Educational Psychology Review, 6(4),
Liberati, A., Altman, D. G., Tetzlaff, J., Mulrow, C., Gøtzsche, P. C., 321–349. https://doi.org/10.1007/BF02213419
Ioannidis, J. P., Clarke, M., Devereaux, P. J., Kleijnen, J., & Moher, Paulus, P. B., Kohn, N. W., Arditti, L. E., & Korde, R. M. (2013).
D. (2009). The PRISMA statement for reporting systematic reviews Understanding the group size effect in electronic brainstorming.
and meta- analyses of studies that evaluate health care interventions: Small Group Research, 44(3), 332–352. https://doi.org/10.1177/
Explanation and elaboration. Journal of Clinical Epidemiology, 1046496413479674
62(10), E1–E34. https://doi.org/10.1016/j.jclinepi.2009.06.006 Piaget, J. (1926). The language and thought of the child (1st ed.).
Lipsey, M., & Wilson, D. (2001). Practical meta-analysis. Sage. Harcourt Brace.
Marion, S. B., & Thorley, C. (2016). A meta-analytic review of collab- Pociask, S., & Rajaram, S. (2014). The effects of collaborative practice
orative inhibition and postcollaborative memory: Testing the predic- on statistical problem solving: Benefits and boundaries. Journal of
tions of the retrieval strategy disruption hypothesis. Psychological Applied Research in Memory and Cognition, 3(4), 252–260. https://
Bulletin, 142(11), 1141–1164. https://doi.org/10.1037/bul0000071 doi.org/10.1016/j.jarmac.2014.06.005
McKenzie, J., & Brennan, S. (2019). Synthesizing and presenting find- Popay, J., Roberts, H., Sowden, A., Petticrew, M., Arai, L., & Rodgers,
ings using other methods. In J. Higgins, J. Thomas, J. Chandler, M. M. (2006). Guidance on the conduct of narrative synthesis in system-
Cumpston, T. Li, M. Page, & V. Welch (Eds.), Cochrane handbook atic reviews (Version 1). Economic and Social Research Council.
for systematic reviews of interventions version 6.0 (updated July Rajaram, S. (2011). Collaboration both hurts and helps memory: A
2019). Chochrane. cognitive perspective. Current Directions in Psychological Science,
McNamara, D. S. (2004). SERT: Self-explanation reading training. 20(2), 76–81. https://doi.org/10.1177/0963721411403251
Discourse Processes, 38(1), 1–30. https://doi.org/10.1207/s153269 Rajaram, S., & Pereira-Pasarin, L. P. (2007). Collaboration can improve
individual recognition memory: Evidence from immediate and
50dp3801_1
McNamara, D. S., & Magliano, J. (2009). Toward a comprehensive delayed tests. Psychonomic Bulletin & Review, 14(1), 95–100. https://
doi.org/10.3758/bf03194034
model of comprehension. Psychology of Learning and Motivation,
Rajaram, S., & Pereira-Pasarin, L. P. (2010). Collaborative memory:
51, 297–384. https://doi.org/10.1016/S0079-7421(09)51009-2
Cognitive research and theory. Perspectives on Psychological Science:
Mende, S., Proske, A., K€ orndle, H., & Narciss, S. (2017). Who benefits
a Journal of the Association for Psychological Science, 5(6), 649–663.
from a low versus high guidance CSCL script and why?
https://doi.org/10.1177/1745691610388763
Instructional Science, 45(4), 439–468. https://doi.org/10.1007/s11251-
Roschelle, J., & Teasley, S. D. (1995). The construction of shared know-
017-9411-7
ledge in collaborative problem solving. In C. O’Malley (Ed.),
Menekse, M., & Chi, M. T. H. (2019). The role of collaborative interac-
Computer supported collaborative learning. Springer.
tions versus individual construction on students’ learning of engin-
Roscoe, R. D. (2014). Self-monitoring and knowledge-building in learn-
eering concepts. European Journal of Engineering Education, 44(5),
ing by teaching. Instructional Science, 42(3), 327–351. https://doi.
702–725. https://doi.org/10.1080/03043797.2018.1538324 org/10.1007/s11251-013-9283-4
Miller, M., & Hadwin, A. (2015). Scripting and awareness tools for reg- Roscoe, R. D., & Chi, M. T. H. (2007). Understanding tutor learning:
ulating collaborative learning: Changing the landscape of support in Knowledge-building and knowledge-telling in peer tutors’ explana-
CSCL. Computers in Human Behavior, 52, 573–588. https://doi.org/ tions and questions. Review of Educational Research, 77(4), 534–574.
10.1016/j.chb.2015.01.050 https://doi.org/10.3102/0034654307309920
Moreland, R. L., & Myaskovsky, L. (2000). Exploring the performance Santanen, E. L., Briggs, R. O., & De Vreede, G. J. (2004). Causal rela-
benefits of group training: Transactive memory or improved com- tionships in creative problem solving: Comparing facilitation inter-
munication? Organizational Behavior and Human Decision Processes, ventions for ideation. Journal of Management Information Systems,
82(1), 117–133. https://doi.org/10.1006/obhd.2000.2891 20(4), 167–197. https://doi.org/10.1080/07421222.2004.11045783
Mullen, B., Johnson, C., & Salas, E. (1991). Productivity loss in brain- Scardamalia, M., & Bereiter, C. (2006). Knowledge building: Theory,
storming groups: A meta-analytic integration. Basic and Applied pedagogy, and technology. In R. K. Sawyer (Ed.), The Cambridge hand-
Social Psychology, 12(1), 3–24. https://doi.org/10.1207/s15324834 book of the learning sciences (pp. 97–115). Cambridge University Press.
basp1201_1 Sch€unemann, H., Vist, G., Higgins, J., Santesso, N., Deeks, J., Glasziou,
Nakagawa, S., & Cuthill, I. C. (2007). Effect size, confidence interval P., Akl, E., & Guyatt, G. (2019). Interpreting results and drawing
and statistical significance: A practical guide for biologists. Biological conclusions. In J. Higgins, J. Thomas, J. Chandler, M. Cumpston, T.
Reviews of the Cambridge Philosophical Society, 82(4), 591–605. Li, M. Page, & V. Welch (Eds.), Cochrane handbook for systematic
https://doi.org/10.1111/j.1469-185X.2007.00027.x reviews of interventions version 6.0 (updated July 2019). Cochrane.
Nijstad, B. A., & Stroebe, W. (2006). How the Group Affects the mind: 2019. www.training.cochrane.org/handbook
A cognitive model of idea generation in groups. Personality and Schwartz, D. L., Chase, C. C., Oppezzo, M. A., & Chin, D. B. (2011).
Social Psychology Review, 10(3), 186–213. https://doi.org/10.1207/ Practicing verus inventing with contrasting cases: The effects of tell-
s15327957pspr1003_1 ing first on learning and transfer. Journal of Educational Psychology,
Nijstad, B. A., Stroebe, W., & Lodewijkx, H. (2003). Production block- 103(4), 759–775. https://doi.org/10.1037/a0025140
ing and idea generation: Does blocking interfere with cognitive Schwartz, D. L., Sears, D., & Chang, J. (2007). Reconsidering prior
processes? Journal of Experimental Social Psychology, 39(6), 531–548. knowledge. In M. C. Lovett & P. Shah (Eds.), Thinking with data
https://doi.org/10.1016/S0022-1031(03)00040-4 (pp. 319–344). Routledge.
Nijstad, B. A., Stroebe, W., & Lodewijkx, H. F. M. (2002). Cognitive Slavin, R. E. (2011). Instruction based on cooperative learning. In R. E.
stimulation and interference in groups: Exposure effects in an idea Mayer & P. A. Alexander (Eds.), Handbook of research on learning
generation task. Journal of Experimental Social Psychology, 38(6), and instruction (pp. 344–360). Taylor & Francis.
535–544. https://doi.org/10.1016/S0022-1031(02)00500-0 Stephenson, G. M., & Wagner, W. (1989). Origins of the misplaced
Nokes-Malach, T. J., Richey, J. E., & Gadgil, S. (2015). When is it bet- confidence effect in collaborative recall. Applied Cognitive
ter to learn together? Insights from research on collaborative Psychology, 3(3), 227–236. https://doi.org/10.1002/acp.2350030303
50 S. MENDE ET AL.

Stroebe, W., Nijstad, B. A., & Rietzschel, E. F. (2010). Beyond product- Vygotsky, L. S. (1978). Mind in society: the development of higher psy-
ivity loss in brainstorming groups: The evolution of a question. In chological processes. Harward University Press.
M. P. Zanna & J. M. Olson (Eds.), Advances in experimental social Wang, H., & Rose, C. P. (2007, August 1–4). A process analysis of idea
psychology (pp. 157–203). Academic Press. generation and failure [Paper presentation]. 29th Annual Meeting of
Sweller, J., & Sweller, S. (2006). Natural information processing sys- the Cognitive Science Society, Nashville, TN, USA.
tems. Evolutionary Psychology, 4(1), 434–458. https://doi.org/10. Webb, N. M. (1989). Peer interaction and learning in small groups.
1177/147470490600400135 International Journal of Educational Research, 13(1), 21–39. https://
Teasley, S. D. (1997). Talking about reasoning: How important is the doi.org/10.1016/0883-0355(89)90014-1
peer in peer collaboration? In L. B. Resnick, R. S€alj€ o, C. Pontecorvo, Webb, N. M. (2013). Information processing approaches to collabora-
& B. Burge (Eds.), Discourse, tools and reasoning: Essays on situated tive learning. In C. E. Hmelo-Silver, C. A. Chinn, C. K. K. Chan, A.
cognition (pp. 361–384). Springer. O’Donnell (Eds.), The international handbook of collaborative learn-
Tsovaltzi, D., Judele, R., Puhl, T., & Weinberger, A. (2015). Scripts, ing (pp. 19–40). Taylor & Francis.
individual preparation and group awareness support in the service Wegner, D. M. (1987). Transactive memory: A contemporary analysis
of learning in Facebook: How does CSCL compare to social net- of the group mind. In B. Mullen & G. R. Goethals (Eds.), Theories
working sites? Computers in Human Behavior, 53, 577–592. https:// of group behavior (pp. 185–208). Springer.
doi.org/10.1016/j.chb.2015.04.067 Weinberger, A. (2011). Principles of transactive computer-
van Boxtel, C., Van der Linden, J., & Kanselaar, G. (2000).
supported collaboration scripts. Journal of Digital Literacy, 6(3),
Collaborative learning tasks and the elaboration of conceptual know-
189–202.
ledge. Learning and Instruction, 10(4), 311–330. https://doi.org/10.
Weinberger, A., Stegmann, K., & Fischer, F. (2007). Knowledge conver-
1016/S0959-4752(00)00002-5
gence in collaborative learning: Concepts and assessment. Learning
Van Bruggen, J. M., Kirschner, P. A., & Jochems, W. (2002). External
representation of argumentation in CSCL and the management of and Instruction, 17(4), 416–426. https://doi.org/10.1016/j.learnin-
cognitive load. Learning and Instruction, 12(1), 121–138. https://doi. struc.2007.03.007
org/10.1016/S0959-4752(01)00019-6 Weldon, M. S., & Bellinger, K. D. (1997). Collective memory:
van Merri€enboer, J. J. G., & Sweller, J. (2005). Cognitive load theory Collaborative and individual processes in remembering. Journal of
and complex learning: Recent developments and future directions. Experimental Psychology. Learning, Memory, and Cognition, 23(5),
Educational Psychology Review, 17(2), 147–177. https://doi.org/10. 1160–1175. https://doi.org/10.1037/0278-7393.23.5.1160 https://doi.
1007/s10648-005-3951-0 org/10.1037//0278-7393.23.5.1160
Verbeek, J., Ruotsalainen, J., & Hoving, J. L. (2012). Synthesizing study Wentzel, K. R., & Watkins, D. E. (2002). Peer relationships and collab-
results in a systematic review. Scandinavian Journal of Work, Environment orative learning as contexts for academic enablers. School Psychology
& Health, 38(3), 282–290. https://doi.org/10.5271/sjweh.3201 Review, 31(3), 366–377. https://doi.org/10.1080/02796015.2002.
Vogel, F., Wecker, C., Kollar, I., & Fischer, F. (2017). Socio-cognitive 12086161
scaffolding with computer-supported collaboration scripts: A meta- Wittrock, M. C. (1989). Generative processes of comprehension.
analysis. Educational Psychology Review, 29(3), 477–511. https://doi. Educational Psychologist, 24(4), 345–376. https://doi.org/10.1207/
org/10.1007/s10648-016-9361-7 s15326985ep2404_2

Appendix A. Characteristics of studies included in the review


Individual preparation for Immediate collaboration Analysis unit;
No. Reference; country collaboration condition (IC) condition (C) relevant measures Sample; Group size Topic; task GPT AIS
1 Allwood et al. (2003), Individual retrieval Collaborative Group; 80 university Crime scenario No No
experiment 1; Sweden practice þ confidence retrieval þ confidence Retrieval: group students; (video);
judgements followed by judgements recognition dyads retrieval
collaborative
retrieval þ confidence
judgements

2 Andersson and a) Individual creation of a) Collaborative creation of Group; 80 university Random words; Yes No
R€
onnberg (1997), single associations to single associations to Retrieval: cued group students; retrieval
experiment 1a; Sweden target words followed by target words followed by recall (semantic/ dyads
collaborative retrieval collaborative retrieval episodic cues)
(semantic cues) (semantic cues)
b) Individual creation of b) Collaborative creation of
single associations to single associations to
target words followed by target words followed by
collaborative retrieval (one collaborative retrieval
episodic cue per item) (episodic cues)
c) Individual creation of
single associations to
target words followed by
collaborative retrieval (two
episodic cues per item)

3 Barber et al. (2010), Individual creation of a) Collaborative creation of Group; 144 university Random words; Yes No
experiment 1a; USA sentences from target word sentences from target Retrieval: cued group students; retrieval
pairs followed by word pairs followed by recall (part-list cues) dyads
collaborative retrieval collaborative retrieval with
the same partner
b) Collaborative creation of
sentences from target
word pairs followed by
collaborative retrieval with
a different partner

4 Barber et al. Individual creation of Collaborative creation of Group; 90 university Random words; Yes No
(2012)a; USA sentences from target word sentences from target word Retrieval: open group students; retrieval
triads followed by triads followed by recall triads
collaborative retrieval collaborative retrieval Referencing: post-
collaborative
shared memory

(continued)
EDUCATIONAL PSYCHOLOGIST 51

Continued.
Individual preparation for Immediate collaboration Analysis unit;
No. Reference; country collaboration condition (IC) condition (C) relevant measures Sample; Group size Topic; task GPT AIS
5 Engelmann, Tergan et al. a) Individual concept Group; 90 university Ecology; Yes Yes (ICb)
(2009); Germany mapping followed by Inferencing: drawn students; concept mapping
collaborative elements in group triads
concept mapping concept map (nodes/
b) Individual concept relations)
mapping þ inspecting Referencing: inter-
future co-learners’ subjective uptakes/ post-
individually constructed collaborative knowledge
concept maps followed by awareness test
collaborative
concept mapping

6 Engelmann et al. a) Individual concept Group; 120 university Ecology; Yes Yes (ICb)
(2014); Germany mapping followed by Referencing: post- students; concept mapping
collaborative collaborative knowledge triads
concept mapping awareness test
b) Individual concept
mapping þ inspecting
future co-learners’
individually constructed
concept maps followed by
collaborative
concept mapping

7 Erkens et al. a) Individual essay writing Individual; 54 high school Geography; Yes Yes (ICb)
(2016); Germany followed by Inferencing: post- students; essay writing
collaborative discussion collaborative sum of dyads
b) Individual essay topic’s extent
writing þ inspecting future Referencing: post-
co-learners’ essay-related collaborative knowledge
topic-coverage followed convergence
by collaborative discussion

8 Gijlers et al. (2013)a; Individual drawing þ Collaborative drawing Group; 56 primary school Biology; Yes Yes
Netherlands inspecting future co-learners’ Inferencing: drawn students; drawing
individually generated elements in group dyads
drawings followed by drawing (concepts/
collaborative drawing processes)
Referencing: consensus-
building
(integration/conflict)

9 Huang et al. Individual answering of Collaborative answering of Group; 60 university Economics; Yes Yes
(2012); Taiwan concept application concept Inferencing: group students; concept application
questions þ inspecting future application questions concept application triads
co-learners’ individually performance
given answers followed by
collaborative answering of
the questions

10 Kablan (2014)a; Turkey a) Individual answering of Individual; 53 primary school Physics; Yes Yes (ICb)
thought-provoking Inferencing: post- students; concept application
questions followed by collaborative transfer quartets
collaborative answering of test (multiple choice/
the questions open questions)
b) Individual answering of
thought-provoking
questions þ peer
assessment followed by
collaborative answering of
the questions

11 Kohn et al. (2011), Individual idea Collaborative idea Group; 54 university Ways to improve Yes Yes
experiment 1a; USA generation þ inspection of generation þ inspection of Inferencing: total of idea students; university;
the future group members’ the group members’ combinations triads brainstorming and
individually generated ideas generated ideas followed by Referencing: idea idea combination
followed by collaborative collaborative idea combinations involving
idea combination combination others ideas

12 Korde and Paulus (2017), Individual idea generation Collaborative idea generation Group; 90 university “Thumbs problem;” Yes No
experiment 1a,b, USA followed by collaborative followed by collaborative Inferencing: group students; brain-writing
idea generation idea generation ideation productivity triads

13 Korde and Paulus (2017), Individual idea generation Collaborative idea generation Group; 180 university “Thumbs problem;” Yes No
experiment 2a,b, USA followed by collaborative followed by collaborative Inferencing: group students; brain-writing
idea generation idea generation ideation productivity triads

14 Lam and Muldner a) Individual assignment of a) Collaborative assignment Individual; 90 university Psychology; Yes Yes
(2017); USA concepts to empirical of concepts to Inferencing: post- students; interpretation of
data þ inspecting future empirical data collaborative dyads empirical data
partner’s individually b) Collaborative invention of transfer test
made assignments concepts from
followed by collaborative empirical data
assignment of concepts to
empirical data
b) Individual invention of
concepts from empirical
data þ inspecting future
partner’s individually
generated inventions
followed by collaborative
invention of concepts
from empirical data
(continued)
52 S. MENDE ET AL.

Continued.
Individual preparation for Immediate collaboration Analysis unit;
No. Reference; country collaboration condition (IC) condition (C) relevant measures Sample; Group size Topic; task GPT AIS

15 Lambiotte et al. Individual application of a Collaborative application of a Individual; 39 university Geography; Applying Yes No
(1987)a; USA text learning strategy text learning strategy Retrieval: open group students; a learning and test
(MURDER-I) followed by (MURDER-I) followed by recall (total/accuracy) dyads taking strategy
collaborative retrieval collaborative retrieval

16 Moreland and a) Individual retrieval Collaborative retrieval Group; 189 university Device assembly No Yes (ICb)
Myaskovsky (2000), USA practice followed by practice followed by Retrieval: open students; procedure;
collaborative retrieval collaborative retrieval group recall triads performance
b) Individual retrieval and retrieval
practice þ receiving
information about future
group members’ expertise
followed by
collaborative retrieval

17 Rajaram and Pereira- a) Individual rating of the Individual; 30 university Random words/ Yes (ICb) No
Pasarin (2007), sharpness/graphical Retrieval: cued students; pictures;
experiment 1a,c; USA quality of the target items group recall triads retrieval
followed by
collaborative retrieval
b) Individual rating of the
pleasantness of the target
items followed by
collaborative retrieval

18 Stephenson and Wagner Individual retrieval practice Collaborative retrieval Group; 34 university Interrogation No No
(1989)a; GB followed by collaborative practice followed by Retrieval: group students; scenario (video);
retrieval collaborative retrieval recognition dyads retrieval

19 Tsovaltzi et al. (2015), a) Individual argument Collaborative argumentation Group; 128 university Psychology; Yes No
experiment 3; Germany construction followed by Collaborative argumentation Referencing: post- students; argumentation
collaborative (supported by a script) collaborative dyads
argumentation shared knowledge
b) Individual argument
construction followed by
collaborative
argumentation (both
supported by a script)

20 Van Boxtel et al. (2000); a) Individual design of a a) Collaborative Group; 40 secondary school Electricity; Yes No
Netherlands concept map followed by concept mapping Inferencing: reasoning students; concept mapping/
collaborative b) Collaborative episodes dyads poster generation
concept mapping poster generation Referencing: subject
b) Individual design of a matter related conflicts
poster followed by
collaborative
poster generation

21 Weldon and Bellinger a) Individual rating of the Group; 48 university Random words/ Yes (ICb) No
(1997), experiment sharpness/graphical Retrieval: open students; pictures;
1a,c; USA quality of the target items group recall triads retrieval
followed by collaborative
retrieval practice followed
by collaborative retrieval
b) Individual rating of the
pleasantness of the target
items followed by
collaborative retrieval
practice followed by
collaborative retrieval

22 Weldon and Bellinger Individual retrieval practice Collaborative retrieval Group; 48 university Narrative text; No No
(1997), experiment followed by practice followed by Retrieval: open students; retrieval
2a; USA collaborative retrieval collaborative retrieval group recall triads
Notes. GPT: generative preparation task: a task which requires the subject to create meaningful connections not already stated in the available instructional
material. Such connections could be drawn between different pieces of information from the instructional material and/or between the information from the
instructional material and preexisting knowledge (see Table 1); AIS: awareness induction support: an individual preparation is considered to be complemented
by awareness induction support if learners are explicitly provided with and instructed to inspect the preparation products of their (future) group members
before collaboration (see Table 1). If a study includes more than one IC and/or C condition, this is indicated by different letters at the beginning of the condi-
tion description. If two IC conditions (a) and (b) differ regarding GPT or AIS, this is indicated by “Yes (ICb)” in columns 7 or 8.
a
The design of this study contains more conditions than mentioned here. However, as this/these conditions are not relevant for our research questions, they are
omitted and the reported sample size is adapted accordingly. bFurther phases follow after the phases in each experimental condition described here. Since
these phases are not relevant concerning our research questions, they are omitted. cWithin-manipulation.
EDUCATIONAL PSYCHOLOGIST 53

Appendix B. Rules and procedures applied to the calculation of effect sizes for studies with multiple
measures for the same dependent variable, more than one experimental and/or control condition, and
missing summary data

Issue Concerned studies and applied rule/procedure


Study contains more than one individual preparation and/or Andersson and R€onnberg (1997, experiment 1); Barber et al. (2010, experiment 1); Moreland
immediate collaboration condition without reporting overall and Myaskovsky (2000); Tsovaltzi et al. (2015, experiment 3); Van Boxtel et al. (2000):
summary data. Collapse the mean values, standard deviations, and sample sizes of the respective conditions
This issue concerns studies used to answer RQ 1 and 2 only. using the method provided in Borenstein et al. (2009, pp. 221–222)
None of the studies used to answer RQ 3 include more than Moreland and Myaskovsky (2000): Three effect sizes are computed because the study
one condition for each individual preparation design variation. compares two conditions of individual preparation for collaboration—one with and one
without awareness induction support—to an immediate collaboration condition:

1. Comparison between immediate collaboration and the collapsed summary data of both
individual preparation for collaboration conditions (RQ 1)
2. Comparison between immediate collaboration and individual preparation for
collaboration without awareness induction support (RQ 2.1 and RQ 2.2)
3. Comparison between immediate collaboration and individual preparation for
collaboration with awareness induction support (RQ 2.2)

Study reports summary data for two or more measures Choose the most relevant measure:
indicative of one dependent variable and these measures do
not represent several sub-measures of the same  Engelmann, Tergan et al. (2009): the number of intersubjective uptakes during
overall measure collaboration reflects participants referencing during collaboration more directly than
post-collaborative knowledge awareness.
 Lambiotte et al. (1987): recall accuracy is more comparable to the retrieval measures
captured in the other collaborative memory studies than total recall performance

Study reports summary data for a) two or more measures Gijlers et al. (2013); Kablan (2014); Kohn et al. (2011, experiment 1):
representing sub-measures of the same overall measure or b) Compute effect sizes and standard errors for each sub-measure or time point and then
two or more measurements at different time points in the combine them, assuming a measure intercorrelation of 0.50 (cf. Borenstein et al., 2009,
collaboration phase and c) does not report summary data pp. 227–229).
averaged across all sub-measures or time points, respectively
Study does not report subsample sizes for the Tsovaltzi et al. (2015, experiment 3):
experimental conditions Impute equal distribution of the reported overall sample size over the single conditions
Study does not report the standard deviations necessary to Contact the authors. When without response:
compute Hedge’s g and the related confidence interval
 Engelmann, Tergan et al. (2009); Engelmann et al. (2014); Weldon and Bellinger (1997,
experiment 2): Compute effect sizes and confidence intervals based on test statistics
(F-values) reported for the comparisons of interest (Lipsey and Wilson, 2001, p. 199)
 Moreland and Myaskovsky (2000): Derive the standard deviations from p-value thresholds
(Higgins et al., 2019)
 Van Boxtel et al. (2000): Apply the formula of Lipsey and Wilson (2001, p. 181) to derive
the pooled standard deviation for the respective comparison from the overall sample
standard deviation reported for the measure of interest
 Rajaram and Pereira-Pasarin (2007, experiment 1); Stephenson and Wagner (1989);
Weldon and Bellinger (1997, experiment 1): Insert the square root of reported ANOVA
mean square errors which refer to factor effects including the comparisons of interest but
also irrelevant conditions as an approximate value for the pooled standard deviation (cf.
Fritz et al., 2012)

You might also like