Conference PaperPDF Available

It All Starts with a Good Idea: A New Coding System for Analyzing Idea Finding Interactions (AIFI)

Authors:

Abstract and Figures

In today's fast-changing world, teams need to develop a sound capacity for finding new ideas. However, we know little about the behavioral micro-dynamics that are at the core of creativity in teams. To overcome these shortcomings, we present a new behavioral coding system for analyzing idea finding interactions (AIFI). The AIFI system aims to help researchers study fine-grained creative team processes. In terms of practical application, the AIFI system can serve to visualize the patterns of Idea finding over time. The codes of the AIFI system were derived both inductively (analyzing videos of innovation teams) and deductively (consulting existing coding systems). A first application of the AIFI system showed moderate agreement among coders, speaking to its interrater reliability. Further, we examined distinct relationships between the codes of the AIFI system and (1) ratings of idea quality provided by external raters and (2) team members' perceived effectiveness. Please use this identifier to cite or link to this item: http://hdl.handle.net/10125/59471
Content may be subject to copyright.
It All Starts with a Good Idea: A New Coding System for Analyzing Idea
Finding Interactions (AIFI)
Paul C. Endrejat
Technische Universität
Braunschweig, Germany
p.endrejat@tu-bs.de
Annika L. Meinecke
Universität Hamburg, Germany
annika.luisa.meinecke@uni-
hamburg.de
Simone Kauffeld
Technische Universität
Braunschweig, Germany
s.kauffeld@tu-bs.de
Abstract
In today’s fast-changing world, teams need to
develop a sound capacity for finding new ideas.
However, we know little about the behavioral micro-
dynamics that are at the core of creativity in teams. To
overcome these shortcomings, we present a new
behavioral coding system for analyzing idea finding
interactions (AIFI). The AIFI system aims to help
researchers study fine-grained creative team
processes. In terms of practical application, the AIFI
system can serve to visualize the patterns of Idea
finding over time. The codes of the AIFI system were
derived both inductively (analyzing videos of
innovation teams) and deductively (consulting existing
coding systems). A first application of the AIFI system
showed moderate agreement among coders, speaking
to its interrater reliability. Further, we examined
distinct relationships between the codes of the AIFI
system and (1) ratings of idea quality provided by
external raters and (2) team members’ perceived
effectiveness.
1. Introduction
Due to an increasingly complex and competitive
environment, organizations need to be more adaptable
than ever to quickly respond to external challenges.
Since tasks become more interconnected, most of
these challenges cannot be solved by a single
individual but require that teams, ideally consisting of
members with diverse disciplinary backgrounds, come
together and pull their efforts to drive organizational
innovation [29]. As a result, team creativity and team
innovation are on the agenda for most organizations
and sparked a growing amount of academic research
[42]. These efforts cumulated in a growing consensus
that we need to gain a better understanding of the
creative processes unfolding in teams [8].
It has long been acknowledged that we cannot
simply put gifted individuals together in a group and
expect creative results [38]. One important framework
in this line of research is McGrath’s [23] input-
process-output model that emphasizes the central role
of team processes as the core mechanism by which
team members combine their individual resources,
such as their creative skills, to master team task
demands and generate (creative) output [16].
Focusing on (creative) team processes requires a
temporal perspective that is sensitive to the ebbs and
flows of idea generation over time. To tackle this
temporal perspective and to understand how team
interaction shapes team outcomes, more and more
team process research makes use of observational
techniques and specifically behavioral coding [12, 26,
34]. "Coding" in this context describes that behavioral
units (e.g. a sentence that comes up during a group
discussion) will be assigned to a behavioral category
(e.g. the code “new idea” or “question”). That way,
researchers can trace the moment-to-moment
dynamics of team interactions which is pivotal to
unravel the underlying behavioral mechanisms that
lead to more or less successful team interactions [1].
A focus on fine-grained team processes is
especially important for innovation teams whose
members typically have a diverse disciplinary
background. Due to different mindsets and working
modes, managing interdisciplinary teams is
challenging [29]. Moreover, teamscompared to
individuals working alonetend to create extreme
ideas, either in a positive way towards the best
outcome, or in a negative way towards the worst
outcome [40]. Thus, there is need to understand how
the forces that are at play during creative problem-
solving can be channeled in a desirable direction,
leading to novel and useful ideas [9, 21].
As a result, there is a continuous call in the team
literature to further expand our understanding of
dynamic team process not only by using survey
methodology, but also by applying real-time behavior
observation techniques [3, 18]. With this paper, we
want to address this call by introducing a new
behavioral coding system for analyzing idea finding
interactions (AIFI).
Proceedings of the 52nd Hawaii International Conference on System Sciences | 2019
URI: hps://hdl.handle.net/10125/59471
ISBN: 978-0-9981331-2-6
(CC BY-NC-ND 4.0) Page 305
2. Background
2.1. The creative problem-solving process
Much research on creativity in groups follows
Osborne’s well-known concept of creative problem-
solving [30]. To structure the creative problem-solving
process, Osborn [25] suggested several stages, which
accrue to three overarching phases: fact-finding, idea-
finding, and solution-finding. Fact-finding includes an
analysis and definition of the problem. Team members
are advised to pick out and identify the core problem,
to point out the characteristics of the problem, and to
gather and analyze data relating to a challenge at hand
[10]. After having established a shared understanding
of the main problem, the team transitions to the idea
finding phase of creative problem-solving, which is
the focus of the current research.
Osborn suggested that idea finding consist of two
separate stages. First, team members are advised to
generate many and manifold ideas (i.e. idea
production). This includes coming up with tentative,
and sometimes even wild, ideas. Second, team
members need to further develop these ideas which
can include adding to ideas, modifying ideas, and
combining ideas (i.e. idea development). The goal is
to pick the most promising ideas and proceed to the
solution phase.
During solution finding, the team evaluates the
tentative solution, for example, by testing and
integrating user feedback and applying it in other
contexts. Finally, the team adopts the final solution
and continues with the implementation of an idea.
2.2. Idea finding: More than brainstorming
In order to generate a lot of ideas during the first
stage of idea finding, Osborn [25] developed what has
become his legacy, the brainstorming technique. To
facilitate brainstorming, Osborn suggested several
guidelines which include focusing on quantity of
ideas, not criticizing or judging ideas, building on the
ideas of others, and freewheeling which describes the
spontaneous expression of (wild) ideas. Put
differently, it is easier to tame down than to think up
[25]. According to Osborn, the creative problem-
solving process is not tied to a fixed schedule but can
span across multiple meetings and iterations
depending on the specific task.
Despite its initial appeal and being widely used in
both research and practice, brainstorming has come
under a lot of scrutiny. Studies even suggests that, in
comparison to nominal groups, brainstorming groups
perform worse in terms of idea quality and idea
quantity [24]. Against this background, it seems to be
only a small advantage that brainstorming groups tend
to be more satisfied and pleased with the results they
come up with [27]. Accordingly, practitioners who
want to make use of the brainstorming technique
might experience the dilemma that they are not able to
offer their teams an effective and satisfying approach
towards creating new ideas.
Such a perspective however neglects that
brainstorming should be just the first step in the idea
finding phase [25]. After team members brought their
ideas to the table, they are tasked with further
developing these ideas. This second step goes beyond
mere brainstorming, as shown in Osborn’s model of
the creative problem-solving process. Unfortunately,
the idea development stage is often given less
attention, although developing ideas is inevitable in
order to move forward in the creative problem-solving
process. Previous research has shown that teams are
not likely to have problems producing a lot of original,
novel, and uncommon ideas [35]. Rather, it is critical
to evaluate these ideas for their practicability, since
they “may be unique or uncommon for good reason!
([32], p. 92).
This distinction between idea production, which
concerns the thinking up of ideas and can be aided by
brainstorming, and idea development, as the selection
and reprocessing of ideas, also aligns with Guilford’s
[7] concept of divergent (producing many ideas) and
convergent (channeling efforts towards one solution)
thinking. Following this paradigm, an ideal team
broadens its thinking and focus to produce a wealth of
ideas in the divergent phase, but then channels its
thinking and focus to agree on a specific idea in the
convergent phase [33]. Next to problems associated
with divergent thinking or brainstorming, it can also
be difficult for a team to stir successfully through the
convergent stage of idea development [30].
2.3. Understanding behavioral dynamics
during the idea finding process
We argue that more research is needed that zooms
in on how teams develop specific ideas in order to
examine idea finding, as the main part of creative
problem-solving, more fully [10, 36]. In other words,
idea finding is more than just brainstorming. The goal
of the new coding system we are about to introduce
therefore lies on including both idea production and
idea development to more fully reflect Osborn’s
original notions of idea finding and to also build on the
literature of divergent and convergent thinking. In
sum, we aim to introduce a theoretically-sound coding
system that supports the efforts of creativity and
innovation researchers.
Page 306
3. The AIFI system
3.1. Goals of the AIFI system
The AIFI system is aimed at analyzing the fine-
grained temporal interaction processes unfolding
during creative problem-solving. The AIFI system sets
out to capture the entire flow of conversational events
exchanged among team members in creativity or
innovation settings. As such, the behavioral codes of
the AIFI system are meant to be exhaustive and
mutually exclusive [1].
When developing the AIFI system, we wanted to
create an efficient coding system that is easy to use and
learn. At the same time, we wanted to arrive at a
coding system that has high external validity and can
be easily used for providing teams with feedback. In
particular, and inspired by previous research on
visualizing the temporal dynamics of change
conversations [14], we wanted the results of the AIFI
analysis to be easily visualized. The idea was to
capture whether teams are currently moving towards
new ideas (i.e. if ideas flow) or whether teams are
stuck in the idea finding process (i.e. inertia
momentum).
3.2. Steps in developing the AIFI system
The AIFI system was developed both inductively
(i.e. bottom-up) and deductively (i.e. top-down). In
doing so, we followed existing guidelines for
developing a new behavioral coding system [1, 41].
During the inductive phase, we observed video
recorded idea-generation teams to an initial
understanding of the team dynamics at play during
innovative tasks, and to decide on a unitizing rule to
segment the communication flow. For the deductive
phase, we aimed to systematize our observations by
building and expanding on existing interaction coding
systems.
3.2.1. Pilot data for developing the AIFI system
inductively. As part of our inductive approach, we
conducted a detailed task analysis. Initially, the Idea
finding interactions of twelve student teams
participating in a Design Thinking workshop were
video-recorded. All workshops were conducted by
facilitators with a professional background in Design
Thinking and lasted for three days. Ad-hoc teams of 4
to 5 students were taught the basics of Design
Thinking, and they applied their new knowledge to
tasks of project partners. The 12 teams consisted of a
total of 53 team members. On average, team members
were 25.50 years old (SD = 1.50), and most
participants were male (79.2% male).
On the second day of the workshops, the teams
were asked to generate ideas for their given
challenges. These team interaction episodes were
videotaped and discussed among the authors of the
current paper as well as student assistants. Later,
videos were annotated using Interact software [22],
and we decided on a unitizing rule for sequencing the
team interaction process into individual behavioral
units.
3.2.2. Literature research for developing the AIFI
system deductively. To relate our observations to
existing coding systems, we carried out a thorough
literature search covering previous literature and
empirical studies on team creativity and team
innovation [5, 27, 29, 42] as well as on team
interaction coding [18, 44]. We consolidated both
universal group interaction coding systems [2, 6, 13,
34] as well as more setting-specific coding systems for
idea generation [11, 15, 37]. For instance, the code
“blocking” was adapted from the Interaction
Dynamics Notation system [37] and relates to our
observation that some team members undermine the
creative process.
3.3. Unitizing
In line with previous research on coding group
interactions [2, 13], the unit of analysis is a thought
unit. A thought unit is the smallest meaningful
segment of behavior that constitutes an entire thought.
In practice, this is often a sentence or statement.
However, it can also be a single word (“Okay”) or an
incomplete sentence (e.g. when a team member calls
out a single idea during a brainstorming episode).
Moreover, a new behavioral event is parsed whenever
the main argument changes (e.g. several ideas are
voiced in a row) or when a team member states several
statements in a row (e.g. first stating a new idea and
then explaining an idea).
Unitizing is carried out in a sequential fashion,
meaning the AIFI systems leads to a continuous
stream of coded behavior. This way, the patterns of
Idea finding over time can be analyzed.
3.4. Description of the AIFI system
An overview of the AIFI system is provided in
Table 1. The AIFI system categorizes behaviors
according to their function for the group discussion.
This means, that the AIFI systems does not provide
a qualitative content-analysis of the creative team
process but a quantitative analysis in terms of behavior
Page 307
counts. In line with our overall goal to develop a sound
coding system that is both suitable for research and
practice, we decided on a hierarchical structure.
Table 1: Overview of the AIFI system
On the most coarse-grained level, the AIFI system
differentiates two emergent team states: group flow
versus inertia momentum. The group flow category
describes that the team is currently moving in the
direction of generating and developing new ideas.
Thus, all behaviors that fall in this category are rated
+1 to indicate that they help towards generating an
idea. On the contrary, inertia momentum describes that
team members are moving away from finding new
ideas and are rated -1. Behaviors that we expect to
neither support nor hinder the finding of ideas are
coded as neutral behavior and receive a rating of zero.
For practitioners who are interested to use a (live)
coding system that is easy to apply but can still trace
the moment-to-moment dynamics, this might be the
level to focus on. We provide a suggestion for how to
visualize this dichotomy (i.e. idea flow versus inertia
momentum) in the discussion.
At the second level, the AIFI system differentiates
five behavioral team process categories: idea
facilitation versus idea inhibition, team spirit
facilitation versus team spirit inhibition, and again
neutral behaviors (which are not yet further
subdivided at this point in the coding process).
Zooming in even further, the AIFI system
differentiates 15 fine-grained behavioral categories.
3.4.1. Codes facilitating Idea finding (+1). Idea
expression: This code is used when a team member
comes up with a new idea but does not further
elaborate on the idea. Such statements are typically
very short (“How about using a different color
scheme?”).
Idea explanation: When team members do not just
state an idea but briefly explain or describe an idea,
this is coded as idea explanation (“Red is much more
eye-catching”).
Idea development: This code is applied when an
idea that has been mentioned before is further
developed by modifying, combining, comparing, or
prioritizing the idea [5]. Thus, new perspectives are
added to an existing idea (“How about adding a
flashing indicator?”).
Knowledge: When team members contribute
relevant (domain-specific) knowledge
(“Distinguishing between red and green is more
intuitively because you know it from traffic lights”) or
reference personal experience (e.g. “When I was in
school…”) that moves the idea finding process
forward, this is coded as knowledge. Providing
knowledge to a team discussion can often serve as the
basis to come up with new ideas and concepts [43].
3.4.2. Codes facilitating team spirit (+1). Support:
When team members express their explicit agreement
or appreciation with an idea, person, or the process,
this is coded as support (e.g. “That’s a great idea!”,
“How cool”, “I like that a lot”). Requesting an idea to
be further explained (e.g. “Could you give an
example?”) or asking for opinions is also coded as
support (“What do you think?”)
Humor: Humorous remarks, jokes, and laughter
are coded as humor. If team members laugh at
someone’s expense this is coded as relationship
conflict (see below).
3.4.3. Neutral codes (0). Process organization:
References to time (e.g. “We only have 5 minutes
left”), reading out task descriptions, mentioning the
overall task, or asking how to proceed (e.g. “Who
wants to go first?”) is coded as process organization.
Simultaneous talk: Even though coders are
encouraged to identify the speaker currently heading
the conversation, sometimes two or more team
members talk at the same time. The code simultaneous
talk is used to capture these events.
Other: Statements that do fit not into any of the
functional categories are coded as other behavior in
order to make the coding exhaustive.
Emergent
state
Process
Behavioral
codes
Group flow
(+1)
Idea facilitation
Idea expression
Idea explanation
Idea
development
Knowledge
Team spirit
facilitation
Support
Humor
Neutral (0)
Process
organization
Simultaneous
talk
Other
Inertia
momentum
(-1)
Idea inhibition
Off-topic
conversation
Loss in detail and
repetition
Silence
Blocking
Team spirit
inhibition
Relationship
conflict
Complaining
Page 308
Figure 1: Stream of idea finding interaction coded with the AIFI system using Interact software [22]
3.4.4. Codes inhibiting idea finding (-1). Off-topic
conversation: Statements not contributing to idea
finding and that signal a lack of interest are coded as
off-topic conversation (e.g. “Did you watch the game
last night?”, “What could we have for lunch today?”).
Loss in detail and repetition: Very long
explanations that do not provide new information are
coded as loss in detail and repetition. The code is also
used, if team members keep referring to their previous
ideas. This repetitive behavior is often shown, when a
team member cannot let go of his or her idea although
the other group members show no interest progressing
with this idea.
Silence: Interactions during idea finding are
typically fast-paced and energetic. If team members do
not say anything for more than 6 seconds, the code
silence is used.
Blocking: The code blocking is used when team
members disagree with other team members or their
ideas or refer to negative feelings (“I hate this idea”).
3.4.5. Codes inhibiting team spirit (-1). Relationship
conflict: If team members show aggression, personally
attack somebody else, make subliminal jokes at the
expense of another team member, or try to undercut
the authority or competence of a fellow team member
(e.g. “You are not more than a student”), this is coded
as relationship conflict.
Complaining: When team members express
disinterest or pessimism (“This sucks”), try to find a
scapegoat (“This is all XY fault”), try to end the
discussion early (“Can we stop already?”), this is
coded as complaining.
3.5. Technical requirements
Coding with the AIFI system does not have to be
software assisted. Nevertheless, we recommend a
software solution to minimize the coding effort and to
simplify data processing and analysis. We used
Interact software in our research [22]. A screenshot
illustrating how the annotation looks like using
Interact software is provided in Figure 1.
4. A first test of the AIFI system
4.1. Participants and procedure
Participants were 80 psychology students who
received course credit for their participation in an idea
finding experiment. In total 20 teams, each consisting
of four team members, worked on two tasks. One
challenge was to help the university becoming more
sustainable and environmentally friendly. The other
challenge was to create ideas concerning how the city
can become more appealing for its residents and
tourists. Each team worked on both challenges, one
time with a facilitator and one time without a
facilitator. Accordingly, the sample consists of 40 idea
finding sessions. Participants were on average 23.30
years old (SD = 5.18), and 77.5% of the participant
were female which is representative for psychology
students.
Each idea finding session took 30 minutes and
consisted of a divergent phase (idea generation; 15
minutes), a convergent phase (idea selection; 10
minutes), and a final elaboration phase (five minutes).
Page 309
During the elaboration phase, participants were asked
to specify their final idea and prepare a flipchart. Each
team was provided with a time-timer clock to support
their time management. Moreover, a student assistant
ensured that each team kept up with the indicated time.
The final idea was presented by one team member
using the flipchart. These presentations were no longer
than one minute and served to later asses the ideas by
external raters. All team interactions and the final
presentations were videotaped.
4.2. Variables
4.2.1. Idea quantity. In order to assess idea quantity,
we asked two student assistants to count all post-its
that were produced by the teams during the divergent
phase. To assess the level of agreement among both
raters, we calculated a two-way random intraclass
correlation (ICC) with absolute agreement (average
measure). There was nearly perfect agreement
between both observers (ICC = .96) so that both scores
were averaged into one score.
4.2.2. Idea quality. In order to rate the quality of an
idea, we developed a new measure that captures both
the feasibility of an idea (three items, e.g. Given the
technical resources, the implementation of the idea is
possible”; ICC = .62) and its user-centeredness (four
items, e.g. “The idea solves problems that people
encounter in their daily routines”; ICC = .38). Raters
were three staff members of the university’s campus
innovation team. Their mission is to spot promising
ideas to improve student life at the campus and help
teams realize their projects by providing expertise and
resources.
4.2.3. Perceived effectiveness. To include a measure
that also takes into account the team members’ own
experience and evaluation, we assessed their perceived
effectiveness with the idea finding process. We used a
scale by Lemieux-Charles and colleagues [19] (e.g.
My team’s overall performance met my
expectations”). Items were answered on a five-point
Likert scale (1 = strongly disagree; 5 = strongly
agree). Reliability for this scale was satisfactory
(Cronbach’s alpha (α) = .82 without facilitator and
Cronbach’s alpha α =.83 for sessions with a
facilitator).
4.2.4. Behavioral coding (AIFI system). Coders were
two student assistants familiar with using Interact
software who received training in using the AIFI
system. Specifically, they participated in a joined
coding workshops. The unitizing rules as well as the
definitions and examples of the different coding
categories were explained and discussed.
Interaction coding was carried out as described
above. Overall, we coded a total of 21.435 behavioral
events, which means that on average, we observed 714
events per 30-minute idea finding session. To ensure
adequate reliability of the AIFI system, seven videos
were double-coded to assess the point-by-point
agreement between both coders. Specifically, using
Interaction software, an alignment Kappa coefficient
was calculated that accounts for the fact two observers
cannot parse each behavioral event at the exact same
second. We specified two behavioral events to be a
match/mismatch if they overlapped at least 50%.
Moreover, we specified non overlapping events to be
a match/mismatch if their onset times were within a
tolerance window of 1 second.
Kappa values can range from −1 to 1, with higher
values indicating higher agreement. We followed
conventional cutoff criteria as proposed by Landis and
Koch [17]: .21.40 = fair, .41.60 = moderate, .61.80
= substantial, and .811.00 = almost perfect.
4.3. Results
Means, standard deviations, and correlations
between all study variables are shown in Table 2. To
visualize the temporal flow of the coded team
interaction, we plotted the coded data for one sample
team (see Figure 2).
4.3.1. Reliability. Interrater reliability for the AIFI’s
process codes (i.e. 2nd level) yielded a Kappa value of
.53 indicating moderate agreement among both
coders. Interrater reliability for the more fine-grained
15 individual codes (i.e. 3rd level) was .44, hence still
in the moderate range. Interestingly, we found that our
first coder followed a more detailed unitizing approach
and identified more behavioral events than the second
coder. Future coder training should place higher
emphasis on how to sequence the behavior stream into
behavioral units.
4.3.2. Validity. In line with our expectations, idea
facilitating behaviors were associated with perceived
effectiveness (r = .34; p = .03) and a higher number of
generated ideas (r = .31; p = .05). Contrary to what we
would expect, teams that produced more ideas,
perceived themselves as less effective (r = -.34; p =
.03). Findings also revealed that team spirit facilitating
interactions such as support and humor were related
team members’ perceived effectiveness (r =.56; p <
.001).
Page 310
Table 2: Means, standard deviations, and intercorrelations between study variables
Mean
SD
1
2
3
4
5
6
7
5.70
.57
4.60
.67
-.31
24.28
7.11
.27
-.17
3.95
.44
-.19
-.01
-.34*
153.93
28.30
.11
.03
.31
.34*
199.50
51.83
.09
-.19
-.07
.56**
.54**
57.50
15.40
.16
.05
.26
-.13
.28
.23
1.40
1.39
-.20
-.19
-.09
.08
-.01
.14
.22
N = 40 teams. p < .10; *p < .05; **p < .01
Figure 2: Time line chart of a 30-minute idea finding episode coded with the AIFI system
Moreover, idea facilitating and team spirit
facilitating behaviors were correlated (r = .54; p <
.001), suggesting that they complement one another.
Unexpectedly, idea facilitation and idea inhibitive
behaviors were also correlated (r = .28; p = .08).
Another unexpected finding, which is not the focus
of the current study but worth mentioning, is the fact
that feasibility and user-centeredness tend to
contradict one another (r = -.31; p = .05). This finding
suggests that realistic and feasible ideas are not the
ideas that meet users’ needs. Furthermore, the number
of ideas generated in the idea generation phase seems
to indicate that the team will produce an idea that is
rated as user-centered (r = .27; p = .094). This supports
Osborn’s [25] claim that during brainstorming, a team
should create a high number of ideas.
5. Discussion
More than seven decades ago, Osborn [25]
articulated the timeless need to advance the
productivity of groups. He developed a creative
problem-solving process, with the idea finding phase
as the backbone of this process. However, previous
research questioned the effectiveness of the idea
finding phase, either due to problems concerning idea
generation [4] or idea evaluation [30].
Despite these problems, however, teams and
organizations are still required to continuously create
new solutions for emerging challenges. Thus, it is not
expected that idea finding teams will vanish from the
organizational context within the near future.
The aim of this paper was to enhance our
understanding of the fine-grained micro-dynamics
unfolding in creativity teams during the idea finding
phase. We thereby contribute towards recent calls to
study team members’ interaction process more closely
[10]. It does not suffice putting together a group of
highly creative individuals because team members’
potential to come up with ideas can be offset when
dysfunctional group dynamics emerge that undermine
finding new and feasible ideas [39]. We introduced the
AIFI system, a coding instrument for capturing the
moment-to-moment dynamics of creative team
interactions to better understand which specific team
behaviors characterize successful idea finding.
Findings from an experimental study provided initial
support for the criterion-related validity of the AIFI
coding system. In particular, findings disclosed that
team members perceive themselves as more effective
when they showed more idea facilitation (e.g.
explaining ideas) and team spirit facilitation (e.g.
humor) behaviors. In addition, our analyses also
revealed some unexpected associations such as a
correlation between idea facilitation behaviors and
idea inhibition (e.g. blocking) behaviors. One
explanation for this finding might be that those teams
who express and develop more ideas also have more
opportunities to block and criticize these ideas.
5.1. Theoretical implications
Thus far, stirring through the idea finding phase
has been a bit like cooking without knowing the exact
recipe. There are some ingredients (e.g. go for quantity
or defer judgement) that everyone would intuitively
agree on. However, knowledge about the exact amount
of these ingredients was missing. So far, we had no
reliable measurement instrument to quantify these
ingredients and understand how they interact to form
Page 311
a delicious outcome. Thus, to be able to replicate a
good recipe, this paper presented the AIFI system.
First, we examined natural occurring interactions
of innovation teams. The behavioral codes derived
from this first step have thus a high external validity.
Second, we built on previous research on group
interaction coding schemes to systemize our
observations and anchor the AIFI system in a diverse
and solid research base. We aimed to develop a tool
that allows to disclose temporal interaction processes
and emergent behavioral patterns, which in turn can
contribute to the theoretical framework that is
currently available for team process researchers. A
first application of the AIFI system revealed that
behaviors supporting an idea are more frequently
shown in teams that also show more behaviors
supporting the team spirit. Using computer software,
we were further able to map the temporal flow of idea
finding interactions (see Figure 2). Such visualizations
can yield additional insights and improve our
understanding of how team progress in finding new
ideas. Interpreting Figure 2, findings suggest that idea
inhibitive behaviors are more frequent in the divergent
phase (the first half, minutes 1-15) than in the
convergent phase (the second half, minute 16-25).
Finding sound explanations and testable hypotheses
for this finding could help to broaden the theoretical
basis of group creativity research.
5.2. Practical implications
The AIFI system is thought to be applicable across
a wider range of group creativity settings, including
traditional brainstorming sessions, group creativity
interactions during organizational meetings, and team
innovation tasks more generally. One requirement is
that team interaction is focused on generating new
ideas and solutions rather than on sharing information
only. The AIFI system can be applied to team data
collected in the laboratory and the field. A version for
real-time (ad-hoc) coding is currently under
development. The goal of the ad-hoc coding system is
that facilitators do not have to interrupt a team but can
still provide immediate feedback [28]. That is, a
stream of coded events could be displayed on a wall in
the background of the working space and provide
visual feedback about a team’s dynamic in real-time
[14]. To make this tool intuitively to use, we suggest
building on the dichotomy of group flow versus inertia
momentum to indicate whether a team moves away or
towards a new idea. Distinguishing only between three
codes (+1, 0, -1) should be suitable for ad hoc (real-
time) coding.
Figure 2. Illustration of the dynamics in an idea
finding team (adapted from [14], p. 423). Events
that help to progress an idea contribute towards
the group flow and are scored (+1), events that
hinder the process to find an idea and contribute
towards the inertia momentum and are scored
(-1). Events that are neither idea supportive, nor
inhibitive are scored as neutral (0).
Figure 2 shows a hypothetical team interaction
episode, consisting of 40 coded events. For this
illustration, we assume that the first half (events 1-20)
depicts the divergent thinking phase and the second
half (events 21-40) the convergent thinking phase.
At the beginning (events 1-10), the team started
with idea inhibitive behaviors, such as blocking
further idea development. However, in the second half
of the idea generation session, the team starts to come
up with ideas and team members tend to build on each
other’s ideas, which is indicated by an increase in the
curve during events 11-20. Similar to the divergent
thinking phase, the convergent thinking started with
idea and team spirit inhibitive behaviors, which
lowered the curve (events 21-30). As team members
might have used this visual feedback about their
dysfunctional behavior, they switched their working
mode and showed more behavior indicating that the
group flows towards a good solution (events 31-40).
Given this illustration, it becomes clear that the
development of such a group support system moves
forward the innovation management process. Such
system would provide immediate feedback to the
team, without interrupting its process. Using the AIFI
system in this way could help a team make use of its
creative potential more fully. In this regard, the role of
a facilitator is similar to that of a good teacher meaning
that he or she should only intervene when the team gets
off track [20].
5.3. Future research
5.3.1. Distinguish between divergent and
convergent thinking stages. Since the idea finding
phase consists of two distinct stages, it is important to
understand whether one could examine significant
differences between these thinking modes [10]. Thus,
Page 312
using the AIFI system to examine behavioral
dynamics and how these patterns might differ
according to the idea generation and idea evaluation
stage could further enlighten the understanding of
beneficial group interactions.
5.3.2. Sequence analysis. Thus far, we only used the
AIFI system to count behaviors. Another possible
application of the system would be to use it for
sequence analysis. Using sequence analysis allows
analyzing which behavior by one team member
increases the likelihood of another team member’s
behavior [1]. For instance, previous research [31] has
shown that ideas are often voiced after a short pause
rather than after periods of increased talk. This might
also be a promising avenue for facilitators who could
align their communication behavior accordingly and
support the idea finding process.
5.3.4 Refining the AIFI. Although the AIFI was
successfully applied to 40 idea finding sessions,
results of the application suggest room for
improvement. First, the interrater reliability is only
moderate, suggesting that more coder training is
necessary to achieve higher agreement among coders.
Other than that, one might also think about refining the
coding system. For instance, the codes “idea
explanation” and “idea development” have conceptual
overlap and it might be hard to distinguish between
these codes. A possible solution might be to merge
these two codes into one.
Another aspect that needs refining is our measure
of the quality of an idea. Especially the intraclass
correlation between raters regarding the user-
centeredness of an idea is rather week. This is
problematic since it means that much of the variance
is not explained by our predictors (i.e. the AIFI codes).
Accordingly, one might re-analyze the ideas that were
video-taped using more advanced measures that assess
the quality of an idea more fully.
6. Conclusion
By presenting a coding system that allows to
analyze a team’s interaction during idea finding, this
paper enhanced our understanding regarding
intragroup micro-dynamics that foster or undermine
creativity and innovation in teams. By its hierarchal
structure, the AIFI system is easy to apply for
practitioners and also allows researchers to carry out a
fine-grained analyses of team processes. Given this
flexible applicability, future research might refine and
advance the AIFI to increase its reliability and validity.
7. References
[1] Bakeman R. and V. Quera, Sequential Analysis and
Observational Methods for the Behavioral Sciences,
Cambridge University Press, Cambridge, MA, 2011.
[2] Bales, R.F., Interaction Process Analysis, Addison-
Wesley Press, Cambridge, MA, 1950.
[3] R.F. Baumeister, K.D. Vohs, and D.C. Funder,
Psychology as the Science of Self-Reports and Finger
Movements: Whatever Happened to Actual Behavior?,
Perspectives on Psychological Science, 2(4), 2007, pp. 396
403.
[4] M. Diehl and W. Stroebe, Productivity Loss in
Brainstorming Groups: Toward the Solution of a Riddle,
Journal of Personality and Social Psychology, 53(3), 1987,
pp. 497509.
[5] Eberle, R.F., Scamper: Games for Imagination
Development, D.O.K. Publishers, Buffalo, NY, 1971.
[6] G.C. Futoran, J.R. Kelly, and J.E. McGrath, TEMPO: A
Time-based System for Analysis of Group Interaction
Process, Basic and Applied Social Psychology, 10(3),
1989, pp. 211232.
[7] J.P. Guilford, The Structure of Intellect, Psychological
Bulletin, 53(4), 1956, pp. 267293.
[8] S. Harvey, Creative Synthesis: Exploring the Process of
Extraordinary Group Creativity, Academy of Management
Review, 39(3), 2014, pp. 324343.
[9] B.A. Hennessey and T.M. Amabile, Creativity, Annual
Review of Psychology, 61, 2010, pp. 569598.
[10] S.G. Isaksen and Treffinger, D.J., Celebrating 50
Years of Reflective Practice: Versions of Creative Problem
Solving, The Journal of Creative Behavior, 38(2), 2004,
pp. 75101.
[11] M.H. Jackson and M.S. Poole, Idea-Generation in
Naturally Occurring Contexts, Human Communication
Research, 29(4), 2003, pp. 560591.
[12] S. Kauffeld and N. Lehmann-Willenbrock, Meetings
Matter: Effects of Team Meetings on Team and
Organizational Success, Small Group Research, 43(2),
2012, pp. 130158.
[13] S. Kauffeld, N. Lehmann-Willenbrock, and A.L. &
Meinecke, The Advanced Interaction Analysis for Teams
(act4teams) Coding Scheme., in The Cambridge Handbook
of Group Interaction Analysis, E. Brauner, M. Boos, and M.
Kolbe, Editors. Cambridge University Press: Cambridge,
MA, 2018, pp. .422431.
[14] F. Klonek, H.F.K. Paulsen, and S. Kauffeld, They
Meet, They Talk but Nothing Changes: Meetings as a
Focal Context for Studying Change Processes in
Organizations, in The Cambridge Handbook of Meeting
Science, J.A. Allen, N. Lehmann-Willenbrock, and S.G.
Rogelberg, Editors. Cambridge University Press.:
Cambridge, MA, 2015, 413439.
[15] C.-Y. Kou and S. Harvey, Coding Scheme for Group
Creativity, in The Cambridge Handbook of Group
Interaction Analysis, E. Brauner, M. Boos, and M. Kolbe,
Page 313
Editors. Cambridge University Press: Cambridge, MA,
2018, 537546.
[16] S.W.J. Kozlowski, Advancing Research on Team
Process Dynamics, Organizational Psychology Review,
5(4), 2015, pp. 270299.
[17] J.R. Landis and G.G. Koch, The Measurement of
Observer Agreement for Categorical Data, Biometrics,
33(1), 1977, pp. 159174.
[18] N Lehmann-Willenbrock and J.A. Allen, Modeling
Temporal Interaction Dynamics in Organizational Settings,
Journal of Business and Psychology, 33, 2018, pp. 325344.
[19] L. Lemieux-Charles, M. Murray, G. Ross Baker, J.
Barnsley, K. Tasa, and S.A. Ibrahim, The Effects of Quality
Improvement Practices on Team Effectiveness: A
Mediational Model, Journal of Organizational Behavior,
23(5), 2002, pp. 533553.
[20] K. Lewin, Cultural Reconstruction, in Resolving
Social Conflicts: Selected papers on Group Dynamics,
Lewin, G, W., Editor. Harper & Row: New York, NY, 1948,
pp. 3442.
[21] Litchfield, R.C., Brainstorming Reconsidered: A
Goal-Based View, Academy of Management Review,
33(3), 2008, pp. 649668.
[22] Mangold, Interact Coding Software, 2015.
[23] McGrath, J.E., Social Psychology: A Brief
Introduction, Holt Rinehart and Winston, New York, NY,
1964.
[24] B. Mullen, C. Johnson, and E. Salas, Productivity Loss
in Brainstorming Groups: A Meta-Analytic Integration,
Basic and Applied Social Psychology, 12(1), 1991, pp. 3
23.
[25] Osborn, A.F., Applied Imagination: Principles and
Procedures of Creative Problem-solving., Scribner, New
York, NY, 1953.
[26] S.B.F., Paletz, C.D. Schunn, and K.H. Kim, Intragroup
Conflict Under the Microscope: Micro-Conflicts in
Naturalistic Team Discussions, Negotiation and Conflict
Management Research, 4(4), 2011, pp. 314351.
[27] P.B. Paulus, Groups, Teams, and Creativity: The
Creative Potential of Idea-generating Groups, Applied
Psychology, 49(2), 2000, pp. 237262.
[28] P.B. Paulus and M.T. Dzindolet, Social Influence
Processes in Group Brainstorming, Journal of Personality
and Social Psychology, 64(4), 1993, pp. 575586.
[29] R. Reiter-Palmon, T.d. Vreede, and de Vreede, G.J.,
Leading Interdisciplinary Creative Teams: Challenges and
Solutions, in Creativity and Leadership in Science,
Technology, and Innovation, B.R. Martin, C.M. Allwood,
M.D. Mumford, and S. Hemlin, Editors. 2013. Routledge:
New York, NY, pp. 240267.
[30] E.F. Rietzschel, B.A. Nijstad, and W. Stroebe, The
Selection of Creative Ideas after Individual Idea Generation:
Choosing between Creativity and Impact, British Journal of
Psychology 101, 2010, pp. 4768.
[31] R.B. Ruback, J.M. Dabbs, and C.H. Hopper, The
Process of Brainstorming: An Analysis with Individual and
Group Vocal Parameters, Journal of Personality and Social
Psychology, 47(3), 1984, pp. 558567.
[32] M.A. Runco and G.J. Jaeger, The Standard Definition
of Creativity, Creativity Research Journal, 24(1), 2012,
pp. 9296.
[33] Sawyer, R.K., Group Genius: The Creative Power of
Collaboration, Basic Books, New York, NY, 2007.
[34] C.C. Schermuly and W. Scholl, The Discussion
Coding System (DCS)A New Instrument for Analyzing
Communication Processes, Communication Methods and
Measures, 6(1), 2012, pp. 1240.
[35] I. Seeber, G.-J. de Vreede, R. Maier, and B. Weber,
Beyond Brainstorming: Exploring Convergence in Teams,
Journal of Management Information Systems, 34(4), 2017,
pp. 939969.
[36] I. Seeber, I., R. Maier, P. Ceravolo, and F. Frati,
"Tracing the Development of Ideas in Distributed, IT-
Supported Teams During Synchronous Collaboration", in
ECIS 2014 proceedings: 22th European Conference on
Information Systems ; Tel Aviv, Israel, June 9-11, 2014, M.
Avital, J.M. Leimeister, and U. Schultze, Editors. 2014: AIS
Electronic Library.
[37] N. Sonalkar, A. Mabogunje, and L. Leifer, Developing
a Visual Representation to Characterize Moment-to-moment
Concept Generation in Design Teams, International Journal
of Design Creativity and Innovation, 1(2), 2013, pp. 93108.
[38] S. Taggar, Individual Creativity and Group Ability to
Utilize Individual Creative Resources: A Multilevel Model,
Academy of Management Journal, 45(2), 2002, pp. 315
330.
[39] K. Tasa, S. Taggar, and G.H. Seijts, The Development
of Collective Efficacy in Teams: A Multilevel and
Longitudinal Perspective, Journal of Applied Psychology,
92(1), 2007, pp. 1727.
[40] A. Taylor and H.R. Greve, Superman or the Fantastic
Four? Knowledge Combination and Experience in
Innovative Teams, Academy of Management Journal,
49(4), 2006, pp. 723740.
[41] F. Tschan, J. Zimmermann, and N. & Semmer, Rules
for Coding Scheme Development, in The Cambridge
Handbook of Group Interaction Analysis, E. Brauner, M.
Boos, and M. Kolbe, Editors. Cambridge University Press:
Cambridge, MA, 2018, 191207.
[42] D. van Knippenberg, Team Innovation, Annual
Review of Organizational Psychology and Organizational
Behavior, 4(1), 2017, pp. 211233.
[43] T.d Vreede, I. Boughzala, G.-J. de Vreede, and R.
Reiter-Palmon, The Team Creativity Model: An
Exploratory Case Study, Journal of the Midwest
Association for Information Systems, 2017(1), 2017,
pp. 1934.
[44] M.J. Waller and S.A. Kaplan, Systematic Behavioral
Observation for Emergent Team Phenomena,
Organizational Research Methods, 21(2), 2016, pp. 500
515.
Page 314
... Behavior observation and coding analysis is a systematic method for unifying and coding continuous, naturally-occurring interaction behaviors, then making valid interpretations and inferences from the collected data. Coding means assigning a corresponding category of behavior (e.g., code Coming up with an idea) to a behavior unit (e.g., a member's statement in the discussion, "I think I could build a retractable clothes rack") (Endrejat et al., 2019). Waller and Kaplan (2016) summarized the observation process into four components: Data Collection Site, Coding Schemes and Intervals, Coder Selection and Training, and Analysis Focus. ...
... One way is to use the number of innovative solutions proposed by the team in the experiment. For example, Endrejat et al. (2019) tried to use the number of innovative solutions to represent team innovative performance. The other way is to evaluate teams' innovative achievements or performance by experts' scoring. ...
... Brainstorming focuses on the number of participants' ideas and asks participants not to judge others' ideas, but they can refer to what other people have expressed to come up with their unique opinions (Osborn, 1953). In recent years, as the research on team innovation has gradually attracted scholars' attention, Realistic Presented Problem (RPP) tasks that combined with brainstorming have been applied to the research on team innovation, which requires participants to propose as many solutions as possible based on a given realistic problem (Xue et al., 2018;Endrejat et al., 2019). ...
Article
Full-text available
Despite a wealth of research on the interaction behavior patterns among team members from different angles, few studies focus on the combination of innovation management and innovation team. With the “Input-Process-Output” theoretical framework, this study takes the coding analysis to explore the differences in the interaction behavior patterns of members caused by the cognitive differences in the higher and lower innovative-performing teams. An innovation experiment was conducted in 12 innovation teams based on an experimental paradigm proposed for team innovation tasks. Subsequently, team members’ 1,754 behaviors were coded to analyze the similarities and differences in the interaction behavior patterns between higher and lower innovative-performing teams with lag sequential analysis. The results revealed that both higher and lower innovative-performing teams showed some same interaction behavior patterns. More specifically, the probability of idea facilitation behaviors being followed by team spirit facilitation behaviors was significantly higher than expected, while the probability of idea facilitation behaviors recurring was significantly lower than expected. However, in lower innovative-performing teams, there were some special interaction behavior patterns, such as “the probability of idea facilitation behaviors being followed by neutral interaction or idea inhibition behaviors was significantly lower than expected.” These phenomena may reflect some realistic situations in our life, such as “One echoes the other,” “Sitting on the sidelines” and “A gentleman is ready to die for his bosom friends” in the members’ interaction after cognitive differences happen. This paper provides opinions and suggestions for the research on the interaction behavior observation and coding analysis among members of innovation teams, as well as theoretical contributions to the research on the behavior observation of innovation teams.
Article
Although Kurt Lewin is a key figure in the history of organizational behavior, only a few scholars or practitioners can explain why he earned this status. In order to shine light on this dissonance, we elaborate on the reasons why Lewin's ideas are seldom discussed in today's curricula (e.g., Lewin's work is difficult to pin down due to his interdisciplinary interest as well as his convoluted writing style). To highlight the enduring applicability of Lewin's approaches, we link his ideas (e.g., psychological satiation) with more recent concepts, like burnout. To support the continuing relevance of Lewin's work, we outline how his ideas can inspire current epistemology by discussing his distinction between Aristotelian and Galilean thinking and his quest to find unexpected observations to generate new knowledge. Furthermore, as Lewin is regarded as the "practical theorist," revisiting his work will show how theoretical concepts can influence organizational practices. Specifically, we discuss how Lewin's ideas can improve design thinking, which is a modern team-based approach to generate innovations.
Chapter
Interaction Dynamics Notation (IDN), introduced by Sonalkar et al. in (Int J Des Creat Innov 1(2), 93–108, 2013), is a notation system for team collaboration. In the years since publication, it has garnered much interest and follow-up work. In this chapter, we consolidate much of the known work performed using IDN. We begin with an introduction to IDN, first abstractly through its description, origin, and symbol set, then concretely through short portions of team discussion. Then, we review research that references IDN, first detailing what is known about and through IDN, and second how IDN has been perceived and how it has inspired other work. We also report descriptive statistics of the symbols produced during collaboration, revealing patterns that help characterize the sequence of IDN symbols. Finally, we discuss two directions for future work.
Article
Full-text available
Collaborative brainstorming is often followed by a convergence activity where teams extract the most promising ideas on a useful level of detail from the brainstorming results. Contrary to the wealth of research on electronic brainstorming, there is a dearth of research on convergence. We used experimental methods for an in-depth exploration of two facilitation-based interventions in a convergence activity: attention guidance (focusing participants on procedures to execute a convergence task) and discussion encouragement (engaging participants in conversations to combine knowledge on ideas). Our findings show that both attention guidance and discussion encouragement are correlated with higher convergence quality. We argue that attention guidance’s contribution is in its support of coordination, information processing, and goal specification. Similar, we argue that discussion encouragement’s contribution is in its stimulation of idea clarification and idea combination. Contrary to past research, our findings further show that satisfaction was higher after convergence than after brainstorming.
Article
Full-text available
Most workplace phenomena take place in dynamic social settings and emerge over time, and scholars have repeatedly called for more research into the temporal dynamics of organizational behavior. One reason for this persistent research gap could be that organizational scholars are not aware of the methodological advances that are available today for modeling temporal interactions and detecting behavioral patterns that emerge over time. To facilitate such awareness, this Methods Corner contribution provides a hands-on tutorial for capturing and quantifying temporal behavioral patterns and for leveraging rich interaction data in organizational settings. We provide an overview of different approaches and methodologies for examining temporal interaction patterns, along with detailed information about the type of data that needs to be gathered in order to apply each method as well as the analytical steps (and available software options) involved in each method. Specifically, we discuss and illustrate lag sequential analysis, pattern analysis, statistical discourse analysis, and visualization methods for identifying temporal patterns in interaction data. We also provide key takeaways for integrating these methods more firmly in the field of organizational research and for moving interaction analytical research forward.
Article
Team innovation is of growing importance in research in organizational psychology and organizational behavior as well as organizational practice. I review the empirical literature in team innovation to draw integrative conclusions about the state of the science and to provide a research agenda to move the field forward. The review identifies two main perspectives in team innovation research, the knowledge integration perspective and the team climate perspective. Key conclusions focus on the need to integrate these perspectives to develop an integrative contingency model of the factors providing teams with diverse informational resources and the factors influencing the extent to which teams integrate these resources in a process of information exchange and integration. As part of these integrative efforts, construct consolidation efforts are important to reverse the tendency for proliferation of substantially overlapping moderators and mediators proposed. The review also identifies the contingencies of the relationship between idea development and idea implementation as the most important understudied issue in team innovation research. Expected final online publication date for the Annual Review of Organizational Psychology and Organizational Behavior Volume 4 is March 21, 2017. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.
Article
The use of moving images to generate data for behavioral analysis has long been a methodology available to organizational researchers. In this article, we draw from previous research in team dynamics to describe and discuss various methodological approaches to using video recorded behavior as a source of quantitative data. More specifically, we identify and examine key decision points for researchers and illustrate benefits and drawbacks to consider. The article concludes with suggestions for ways in which quantitative video-based approaches could be improved.
Article
In this paper I propose goal setting as a mechanism for linking brainstorming re- search to organizational creativity. Using the large body of literature on brainstorm- ing as a prototype, I develop a goal-based view of intervention in idea generation. A goal-based view provides new insight into the conduct and evaluation of brainstorm- ing research that improves its relevance to management, and it suggests a broader agenda for defining expectations for creative ideas and tailoring interventions to match these definitions.