Universität Ulm | 89069 Ulm | Germany Faculty of
Engineering, Computer
Science and Psychology
Databases and Information
Systems Department
Investigating the Effects of
Modularization in Business Process
Models on the Cognitive Complexity of
Humans
Master’s thesis at Ulm University
Submitted by:
Julia Baß
Reviewer:
Prof. Dr. Manfred Reichert
Dr. Rüdiger Pryss
Supervisor:
Michael Zimoch
2019
Version from January 21, 2019
c
2019 Julia Baß
This work is licensed under the Creative Commons. Attribution-NonCommercial-ShareAlike 3.0
License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/3.0/de/
or send a letter to Creative Commons, 543 Howard Street, 5th Floor, San Francisco, California,
94105, USA.
Composition: PDF-L
A
TEX2ε
Abstract
Process models enable a better understanding of processes in enterprises as the
processes are visualized step by step. This is due to the fact that process models
provide precise knowledge with respect to the process through sequential linking of
activities. Understanding of the process model is essential for both modelers and
readers of a process. The complexity of process models has a direct effect on how
comprehensible they are. Modularization is an approach for reducing the complexity.
It can be applied in three different ways. Depending on the intended purpose, the
design varies. Furthermore, the various modularization approaches for business process
models can have implications on the cognitive complexity for humans as individuals.
This thesis is based on psychological and neuroscientific cognitive concepts to gather
findings through studies.
A survey and an eye tracking study were designed and conducted to obtain insights
in terms of the cognitive load when individuals with little experience read different
modularized business process models. The survey focuses on the cognitive load and
understandability while reading business process models. As a single factor between
subject study design was applied, a subdivision into three groups (one group for each
modularization approach) was utilized. In contrast to the survey, the eye tracking
study, with its 3x3 within subject design, provides insights into the performance success
(number of correct answers, required time). Further, design variants of the single
modularization approach are comparable.
The results of the survey and the eye tracking study indicate one significant difference.
This difference is based on the intrinsic cognitive load measured in the survey. Vertical
modularization provides a significantly higher intrinsic cognitive load compared to the
other approaches in business process modeling. However, as no further differences
arose, the utilization of modularization approaches in business process modeling has no
impact in terms of understandability of the process model.
iii
Acknowledgment
At this point I would like to thank all the persons that helped me in creating this thesis.
First, I want to thank the reviewers Prof. Dr. Manfred Reichert and Dr. Rüdiger Pryss,
through whose consent the realization of this thesis was possible.
Second, I would like to thank Michael Zimoch who was a great supervisor and always
supported me.
A special thanks goes to my family and friends who are always by my side and supported
me whenever they could.
Finally, I would like to thank all participants of the survey and study.
v
Contents
1 Introduction 1
1.1 Motivation and Problem Statement . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Structure of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Modularization of Business Process Models 5
2.1 Business Process Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Modularization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 Vertical Modularization . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.2 Horizontal Modularization . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.3 Orthogonal Modularization . . . . . . . . . . . . . . . . . . . . . . 12
3 Fundamentals of Cognitive Complexity 17
3.1 Human Cognitive Complexity . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2 Cognitive Load Theory (CLT) . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.3 Relation of Eye Behavior and Cognitive Load . . . . . . . . . . . . . . . . 22
4 Study Planning and Definition 25
4.1 Goal Definition and Context Selection . . . . . . . . . . . . . . . . . . . . 25
4.2 Hypotheses Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.3 Study Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.4 Study Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.5 Risk Analysis and Migrations . . . . . . . . . . . . . . . . . . . . . . . . . 43
5 Study Operation 49
5.1 Study Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.2 Study Execution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.3 Data Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
6 Study Analysis and Interpretation 57
6.1 Data Set Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
vii
Contents
6.2 Hypotheses Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6.3 Summary and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
7 Related Work 77
8 Future Work and Conclusion 81
8.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
8.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
A Survey 99
B Study 133
B.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
B.2 Image 1 (P1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
B.3 Image 2 (P2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
B.4 Image 3 (P3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
B.5 Image 4 (P4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
B.6 Image 5 (P5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
B.7 Image 6 (P6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
B.8 Image 7 (P7) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
B.9 Image 8 (P8) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
B.10 Image 9 (P9) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
B.11 Questionnaires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
B.12 Declaration of Consent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
viii
1
Introduction
Section 1.1 addresses the motivation that led to the preparation of this thesis. Further,
the problem is stated. In Section 1.2, the objectives of this thesis are explained. Finally,
the structure of this thesis is described in Section 1.3.
1.1 Motivation and Problem Statement
Business process models, e.g. in terms of incident management, can be complex [
1
,
2
,
3
],
even more complex than their underlying business processes [
4
]. As stated by [
4
],
’Complexity has undesirable effects on, among others, the correctness, maintainability,
and understandability of business process models’
. Furthermore, the size of the process
model and its understandability are related [
5
,
6
]. A large process model in terms of
size causes difficulties in reading. This is, for example, caused by errors that appear
more frequently in larger business process models [
6
]. Minimizing the size shouldn’t
be realized through omission of relevant process parts [
7
,
6
]. Therefore, the creation
of process models that are understandable, complete with respect to the relevant
parts, and maintainable, has to be covered by modularization. Modularization exists to
decompose an element into smaller units. It provides flexibility in terms of its parts as
each unit is intrinsically complete [
8
] and interchangeable [
9
]. Most research regarding
modularization of business process models has a conceptual nature [
10
,
11
]. By
contrast, the objective of this thesis is to obtain insights through eye tracking with
respect to the resulting cognitive load while reading business process models based
on different modularization approaches. Measuring cognitive load of business process
models using eye tracking technology is a research subject of the DBIS (Databases and
1
1 Introduction
Information Systems Department) at Ulm University. In their research, business process
models were, for example, compared with regard to their basic activities. However,
modularization was not considered under this aspect. Even though a lot of research
about modularization of business process models is already established [
11
], various
problems still exist. On the one hand, business process modelers and/or users aren’t
inevitably experienced in modularization. On the other hand, a lot of scientific research
is missing in certain research fields. As an example, different activities can be applied to
model the same matters; however, extensive research in terms of their impact on the
models’ understandability is missing. In terms of modularizing business process models,
a paper from [
1
] compares the usability of three different modularization approaches.
The participants that assessed the usability were experts. Consequently, novices were
not included in the evaluation. Therefore, basic understandability of the subdivision and
statement quality aren’t given, as it requires the inclusion of novices. Hence, instead of
usability, the cognitive load is measured in this thesis.
1.2 Objective
The objective of this thesis is to compare horizontal, vertical, and orthogonal modulariza-
tion, the three modularization approaches extracted from [
1
]. Therefore, the graphical
representation, predominantly based on the examples of business process models avail-
able in [
1
], will be discussed. Additionally, representations that are stated in [
1
] will be
part of the discussion. The representation differs regarding the activities that can be ap-
plied for modularization in business process models. As an example, different events that
vary with respect to the icons can be utilized. The various representations could cause
effects on the understandability and the cognitive load. Instead of evaluating the usability
of modularized business process models, cognitive load will be measured through a
survey and an eye tracking study. Both times the participant has to answer question-
naires about the cognitive load while reading BPMN 2.0 process models. Through the
study, insights are given that offer evidence indicating the difficulties and challenges of
each modularization approach. Furthermore, hints advising the modeler and reader
of process models could be prepared based on these results. Further, possible icon
2
1.3 Structure of the Thesis
changes are prepared that could cause an easier understandability and navigation in
modularization, next to linking.
1.3 Structure of the Thesis
Chapter 2
focuses on modularization in business process models. An explanation
regarding business process models is provided. Besides, it specifies how complexity,
redundancy, and further aspects can be handled. Therefore, different modularization
approaches are presented. In
Chapter 3
, insights into the cognition with respect to
complexity are provided. Thereby, the cognitive complexity and cognitive load theory
are described. Further, the relation between eye behavior and cognitive load is given.
Superficially these factors are part of the measurements in the studies. With respect to
the survey and the eye tracking study, the studies are planned and designed properly
in
Chapter 4
. The goal of this chapter is the acquisition of insights based on mea-
surements that were gathered while participants read modularized business process
models. Moreover, the formulating of hypotheses, setup and design of the study, risk
analysis, and the avoidance of risks are addressed. Afterwards, the study operation is
presented in
Chapter 5
. First, the study preparation is reported. It includes pilot studies
and the recruitment process of participants. Then, the structure for running the studies
is described. Finally, this chapter provides first insights regarding the evaluated data.
This data is comprised of independent variables and attributes of the participant like
the age. In
Chapter 6
, the study is analyzed and interpreted. First, the used data set
reduction is provided. Then, the explained hypotheses are tested with reference to the
results received from the studies. A discussion based on the obtained results is carried
out. Thereby, the reasons that led to the results are discussed.
Chapter 7
examined
related work. Finally, Chapter 8 presents further research and summarizes this thesis.
3
2
Modularization of Business Process
Models
The thesis considers modularization in business process models. Therefore, it is neces-
sary to examine modularization and business process models theoretically and through
exemplifications. Section 2.1 introduces business process models. Its modularization is
described in Section 2.2.
2.1 Business Process Models
A process is a set of activities [
12
,
13
]. Activities are elements like events, gateways, and
tasks with a defined design and function. Further, they are composite units executed in a
workflow. Each process has a beginning and an end. In addition, it includes a structure
related to the ’how’ an organization accomplishes [
13
] and a defined goal [
14
]. This
shows that the management of operations and the structuring of work is necessary [
15
].
Therefore, concepts [16], methods, tools, and principles [15, 17] are applied.
Through optimized business processes, organizations can save time, money, and in-
crease customer satisfaction [
18
]. An approach for optimizing business processes can
be the BPM life-cycle [
2
]. Through its steps, process models are analyzed, designed,
implemented/reconfigured, enacted, and monitored [
2
,
19
]. This shows the need of
business process models which are the
’lion’s share of bpm (..) literature’
regarding to
[15].
In literature, thematic areas like similarity, complexity, and design are researched. Super-
ficially business process models are, for example, useful for the design and inspection
5
2 Modularization of Business Process Models
of Process Aware Information Systems (PAIS) [
20
,
21
], process automation [
16
], and
Service Oriented Architectures (SOA) [
22
,
23
,
24
]. PAIS are software systems that
handle processes through process models [
25
]. SOA support a unit composition of
interacting services with reference to the restructuring in infrastructures and applications
of software [
26
]. A business process model is a graphical representation of a process. It
provides insights into the process by linking single activities. Thereby, the text is reduced
to a minimum. Instead, a narrative text that is more difficult to understand [
27
] can be
utilized. Hence, it serves the purpose to be understood by different involved stakeholders
[
2
,
28
], and it offers an extensive understanding of a process [
12
,
29
]. Based on a study,
[
30
] discussed, among other things, the above mentioned benefits regarding the targets
of design and inspection, next to the understandability. However, they also refer to the
improvement and communication of process modeling.
While business process models are only useful if they are understood by people, an
explanation of the basic elements is provided in the business process modeling notation
BPMN 2.0 [31].
Figure 2.1: Basic Elements of BPMN 2.0 Models
6
2.1 Business Process Models
Figure 2.1 shows a BPMN 2.0 business process model that is split up into two pools
(Pool 1 and Pool 2). A pool represents a responsible process-participant that handles
the process in its progress [
32
]. A process-participant can be a role, an organization, or
a system [
33
]. In a pool, a process model is displayed horizontally, and has a process
flow direction represented by arrows. Each process has a start event that triggers the
process. The start event is followed by an activity, in this case the task
Carry out A
. A
task is an element that includes the ’to do’ at this point. After task
Carry out A
,
Carry
out B
has to be executed. Afterwards, a decision has to be made. It is represented by
the rhombus that includes the ’X’. Based on the decision, it is possible to run the path
yes
or
no
. If
no
occurs, the process leads to an error end event that throws an error.
However, a crossover to the
Subprocess Event
is executed. The
Subprocess Event
is
situated in the dotted lined edging (located in Pool 1), and indicated by its name. After
running the subprocess, the whole process is completed. If instead of
no
,
yes
occurs,
the process runs a subprocess task. A subprocess task is represented by the ’+’ icon
in the bottom center of the task. Underneath this task, a further complete process is
given. After running the underlying process, the intermediate message event
Sent data
is executed. Using a sequence flow (the dotted arrow), this event calls the message start
event
receive data
located in Pool 2, so that the process in Pool 2 is triggered. After
the message event, the task
Carry Out C
is executed. After that, a gateway is given. A
gateway refers to a decision. Different gateways exist. The gateway in the process model
is an AND-Gateway represented by a rhombus with a ’+’ inside. The AND-Gateway
requires the execution of both paths; however, the execution does not have to run in
parallel. Hence, next to
Carry out D
,
Carry out E
has to occur, before the process of
Pool 2 is finished and the data can be sent to Pool 1. Back in Pool 1, the process is
finished after receiving the data. This is visualized through the end event.
Even though further business process modeling notations like EPC [
34
] and FlowChart
[
35
] exist, merely the BPMN 2.0 notation will be mentioned and presented in this thesis.
BPMN 2.0 is selected because it is the leading standard due to its frequent utilization in
business [
36
]. Therefore, it is suitable for this survey and eye tracking study. Indepen-
7
2 Modularization of Business Process Models
dent of the business process modeling notation, criteria for a good process model exist.
These criteria are comprised of, for example, the correctness [37, 38, 39] and design.
In terms of design, [
40
] focuses on the modeling of business processes. They defined
guidelines to achieve a process model that is well structured, understandable, and
correct. Therefore, the focus is, for example, given in decomposing models with more
than 50 elements. Decomposition means dividing a unit into subdivisions. As a large
model can be decomposed into single modules, the ensuing section provides insights
into the subject of modularization.
2.2 Modularization
Modularization is covered in literature since 1960 [
41
], and represents the decomposition
of a larger unit (e.g. a process model) into individual modules (e.g. multiple small
process models) [
8
]. Each module is an element that is independently manageable [
11
],
interchangeable [
9
], and intrinsically complete [
8
]. It interacts at particular interfaces
with other modules [
42
]. The concept of modularization can be found in different fields.
Next to products, programming, and systems, it is used in business processes [
1
,
43
]. In
Figure 2.2 a modular subdivision extended by modules is described. At the beginning,
the modular approach with four tiles is extended by three modules. This leads to eight
variants. Each variant stands for a possible usage. If the modular approach is extended
by another module (module four), eight additional variants can be generated. Therefore,
totally sixteen variants are possible. Through this example, it becomes clear that modules
can interact with each other, even though they are independent. Additionally, changes in
a module or the replacement of a module can be handled easily.
This thesis is focused on the utilization of modularization techniques in terms of business
process modeling. Modularization is utilized to handle complexity [
45
]. As stated by
[
46
],
’the degree of complexity management varies according to the ability of a notation
to represent information without overloading the human mind’
. Next to complexity, mod-
ularization is used to minimize the size [
47
], and to handle flexibility [
48
] of process
models. Therefore, modularized process models tend to be less error prone and more
8
2.2 Modularization
Figure 2.2: Modular Subdivision [44]
understandable [
4
]. Further, modularization supports reusability and maintainability [
1
].
All these items require successful modularization. Different aspects have to be consid-
ered in realizing a successful modularization. These aspects are operations, selections,
and prerequisites [
11
]. While [
11
] refer to the ’when’ (e.g. number of elements) and
’what’ (e.g. selection of parts of a model that can be modularized), models have to be
modularized through modularization selection and prerequisites. [
11
] mentions that basic
operators lead to a syntactical correctness with reference to operations. Syntactical
correctness bases on notational rules and, hence, the modeling syntax is understood
[14].
Next to the ’when’ and the ’what’ has to be modularized, additionally the ’how’ is of
interest. With respect to the ’how’, different possibilities of modularization exist. These
possibilities fulfill various purposes and can be combined in process models. In scientific
research, only vertical and horizontal modularization are mentioned. Therefore an
example is provided in the paper [
46
]. [
1
] states another approach of modularization:
the orthogonal modularization.
The three modularization approaches are presented below.
9
2 Modularization of Business Process Models
2.2.1 Vertical Modularization
Vertical modularization, also called
hierarchical structuring
[
49
], is the decomposition of
a process model into subprocesses. Corresponding to this foundation of modularization,
a hierarchical structure is given [
1
]. While the main process has a high abstraction level,
the underlying subprocesses are more refined [
50
]. This leads to the benefit that a single
module in form of a subprocess can be maintained [
1
,
51
], as well as continuously and
separately proved. Additionally, the possibility of reusage is given. Hence, redundancies
are improved [
1
]. By changing the activities of one subprocess, the change will be
addressed by all processes that call the subprocess. However, the reusage depends
on how the overlying process is subdivided. In scientific research, the differentiation
between embedded and independent subprocesses is considered [
52
,
53
,
54
]. An
embedded subprocess is part of the overlying process [
52
,
53
,
55
] and not reusable [
55
].
In contrast, the independent approach can be called by diverse processes [53, 54] and
is self-contained to the overlying process [
52
]. The realization of vertical modularization
can be handled in BPMN 2.0 through collapsed and expanded subprocesses [
1
,
47
]. An
example is shown in Figures 2.3 and 2.4. In both figures, a process model is divided
up into detailed processes. Figure 2.3 represents modularization through collapsed
subprocesses. Collapsed subprocesses are realized in a self-contained model and have
an information hiding quality. The calling task has an icon in the form of a ’+’ to address
a process given in another level. In the representation of collapsed subprocesses the
suggested number of levels by [
56
] is between three and seven, and should be observed.
In contrast to collapsed subprocesses, expanded subprocesses are groups in a process
model. Hence, they are positioned on the same level, as presented in Figure 2.4. [
47
]
compared the two forms, and reported that expanded subprocesses (with a solid line
between the parent and subprocess) provide a better understandability.
10
2.2 Modularization
Figure 2.3: Collapsed Subprocesses in Vertical Modularization
Figure 2.4: Expanded Subprocesses in Vertical Modularization
2.2.2 Horizontal Modularization
Horizontal modularization (i.e.
horizontal segmentation
[
57
]) has the objective to deduct
a model into various small models [
1
] with the same abstraction level [
58
]. It is utilized
to address benefits like reusability, and to decrease complexity by building models with
almost the same size [
1
]. Reusage is given when different processes interact with the
same process using the same interfaces.
In BPMN 2.0 it is realized through pools. The interaction between individual pools is given
by activities. An end or intermediate event of one pool calls a start or intermediate event
in another pool. Message, link, and signal events find use in horizontal modularization
[
1
]. The message event is presented in Figure 2.5, and the link event is depicted in
Figure 2.6. A message event is the sending of messages [
33
]. This is shown in Figure
2.5, and clarifies that an interaction between two pools occurs through a message flow.
The message end event of Pool 1 calls the message start event of Pool 2.
11
2 Modularization of Business Process Models
Figure 2.5: Message Event in Horizontal Modularization
In Figure 2.6, the link event is given multiple times, as each link event refers to a
subprocess. In this figure, link event
2
is focused because it refers from Pool 1 to Pool 2.
Figure 2.6: Link Event in Horizontal Modularization
2.2.3 Orthogonal Modularization
The orthogonal modularization is based on exceptions and further crosscutting concerns
[
1
]. At crosscutting concerns, such as security aspects [
1
], the modular process-structure
12
2.2 Modularization
can not be observed. Hence, the elements placed on (different) processes have to be
cut across the structure [
59
]. Efficient modularization through other approaches would
be difficult [
43
]. To handle orthogonal modularization, various realization opportunities
exist. On the one hand, exceptions can be realized through event subprocesses [
1
]. As
stated by [52], they can be classified into:
•
An event inside the subprocess that triggers an event (e.g. end error event) of the
overlying process after reaching a special element
•
An exception triggered at any point in time caused by an external event (e.g.
timeout)
For exception handling, [
60
] lists end events such as:
error
,
cancel
, and
compensate
.
The mentioned events trigger a start event in another pool. An example is presented
in Figure 2.7. At the beginning, the process in Figure 2.7 is sequentially executed.
Then, the XOR-Gateway, referring to a decision between two paths, is executed. When
q?
is
ja
the process is finished. Otherwise, when
q?
is
nein
, an exception has to
be handled. The error end event triggers the process that is necessary for exception
handling. This process is placed in an event subprocess called
Event
in Figure 2.7. An
event subprocess is represented by a rectangle with a dashed border. The process that
has to be executed starts with an error start event process, as it is called by an error
event. Thereafter, the process is executed and finishes through the end event given in
the event subprocess.
A further form of presentation of exception exists. [
60
] refers to the fact that exceptions
can be thrown from a subprocess to the overlying process. As presented in Figure 2.8,
elements of the vertical modularization are given. First, the overlying process triggers
the subprocess. The activities of the subprocess are executed. When
q?
is
ja
, both
processes are finished. Otherwise, when
q?
is
nein
the error end event triggers the
intermediate error event that is attached onto the edge of the subprocess task. Then,
the following process is executed according to the workflow.
13
2 Modularization of Business Process Models
Figure 2.7:
Exception Handling through Interrupting Event I in Orthogonal Modularization
Figure 2.8:
Exception Handling through Interrupting Event II in Orthogonal
Modularization
Exceptions can be handled through interrupting or non-interrupting event subprocesses
[1]. In both examples, the interrupting event subprocess is shown.
On the other hand, it can be realized by aspect-oriented paradigms [
1
]. This approach
is based on the fundamentals of aspect-oriented programming [
1
,
43
]. It provides the
benefits such as handling security aspects [1, 61].
[
59
] utilize the term
aspect oriented process modeling language
that has a strategy
where all concerns are processed in the same way. Differences in processing only occur
in the relationships. As stated by [
43
], business processes can be divided, through
aspect-oriented modularization, into:
14
2.2 Modularization
•The basic BP (core process) which contains the essence of the BP
•
The aspect process which captures the crosscutting information cutting across the
core process
In BPMN 2.0 the composition of the aspect-oriented process modeling language is by ’
a
process model described using any language for modeling processes’ [59].
The process in executing aspect-oriented paradigm is explained to the example of Figure
2.9.
The process is executed and reaches the activity
H
that includes a pointcut, after
selecting the path
nein
of the XOR-Gateway
q?
. A pointcut provides the possibility to
select one or more join points [
61
]. In Figure 2.9 only one join point is referenced by
H
.
This join point is in the advice (a module that captures the crosscutting concern [
61
])
Event
. The join point in the advice is the start event
Trenn Punkt
. After the join point
Trenn Punkt
is triggered, the process inside the advice is executed. At the end of the
modularized process, a join point
Zusammenführungs Punkt
is given that leads to the
activity
J
of the main process. The representation is in the modeling notation BPMN
2.0. Contrary to the example, no symbols are given to represent the aspect-oriented
modularization.
Figure 2.9: Aspect-Oriented Approach in Orthogonal Modularization
15
3
Fundamentals of Cognitive Complexity
Understandability as well as an easy and efficient coping of the information presented
are important in dealing with business process models. Hence, an introduction of human
cognitive complexity is presented in Section 3.1. In Section 3.2, the cognitive load and
the sources of impact are given. Finally, cognitive load by the pupil, with reference to
eye tracking, is addressed in Section 3.3.
3.1 Human Cognitive Complexity
[
62
] stated the term
cognitive complexity
in 1955. Cognitive complexity has different
meanings, as there is no clear theoretical definition given [
63
]. Further, it is adopted in
many research fields. Descriptions of cognitive complexity exist, such as this definition
from 1962:
’Cognitive complexity is defined as the number of independent dimensions-worth of
concepts the individual brings to bear in describing a particular domain of phenomena;
it is assessed with a measure of information-yield based on an object-sorting task’
[
64
].
As stated by [
65
], the main factor of cognitive complexity is the information processing
through individual personality structures, that are based on experiences [
66
] of individu-
als, for example. Human information processing is based on external input. Therefore,
different models with respect to information processing exist. One of the first models
is from [
67
]. They designed a model with interacting items. These items are:
sensory
register,short-term store, and long-term store.
17
3 Fundamentals of Cognitive Complexity
Figure 3.1: Structure of the Memory System [67] (simplified)
As shown in Figure 3.1, external input is recorded by a human. This is realized by
attentioning, recognizing, and searching [
68
]. After acquisition of the external input,
the acquired information is transferred to the short-term store in form of code or the
interpretation of external input [
69
]. The information may be forgotten after a while since
the capacities of the short-term store are limited [
70
]. This limitation is not necessarily
given [
71
]. However, information is temporarily stored [
72
]. Furthermore, the information
from short-term store can be transferred in form of a copy to the long-term store. The
likelihood that information will be transferred increases with its longer presence in the
short-term store [69].
So far, the ’how’ in information processing was presented. Further, the complexity is
relevant in cognitive complexity. The complexity is distinguished through three items
[
73
]. These items are
integrity
,
sophistication
, and
discrimination
. [
73
] argues that
discrimination and sophistication are based on the capacity regarding the areas of
assessment and distinction. On the other hand, [
73
] shows that the integrity picks
out an overall evaluation on the basis of isolated information as a central theme. To
measure the complexity, metrics in form of cognitive weights can find use [
74
]. They
capture the efforts that are necessary for successful performance. An example is
offered in Table 3.1. The table contains weightings that relate to the understandability of
business process modeling elements. A low weighting shows that the pattern is easily
understandable and less complex while a high weighting points out less comprehension.
While a sequence in the business process modeling notation YAWL is well understood,
an OR-split leads to a reduction of comprehension. Therefore, the authors of [
74
] argue
that AND/XOR are significantly higher in comprehension than OR. To measure cognitive
complexity, inter alia, complex units are used with regard to cognitive weighting, difficulty
18
3.2 Cognitive Load Theory (CLT)
Workflow Pattern [3] BPM control structure corresponding software con-
trol structure
W
Sequence consecutive steps in a workflow sequence 1
Exclusive Choice XOR-split (exactly one of two branches is
chosen) with corresponding XOR-join ...
branching if-then ... 2
XOR-split (exactly one of >= 3 branches is
chosen) with corresponding XOR-join
branching with case (with an
arbitrary number of selectable
cases)
3
Parallel Split and Syn-
chronization
An AND-split activates all outgoing links in
parallel, a corresponding AND-join synchro-
nizes the lows of control
execution of control flows in
parallel
4
Multiple choice and
Synchronizing Merge
OR-split (a number of branches is chosen
from 2 or more possible branches) with cor-
responding OR-join
branching with case, followed
by parallel execution
7
(none) Composite task (subtask, can be used for de-
composing a BPM into modules)
call of a user-defined function 2
Multiple Instances
Patterns
Multiple Instance Activity ( allows multiple in-
stances of an activity to run concurrently
branching, followed by parallel
execution
6
Table 3.1: Cognitive Weights W for BPM Elements [74]
of understanding, or the load that is a result of the limited capacities in information
processing. This is, for example, given in the paper [
3
]. The authors argue that high
cognitive effort is necessary to understand the elements of, and their relation in, process
models. Additionally, they point out that they set up business process modeling metrics
to determine the understandability for measuring necessary factors of cognitive load.
Further, they show why the cognitive load theory is suitable in this context. Therefore,
they argue
’that understanding of a fact in a BPM becomes more difficult if the number
of model elements that need to be attended to increases. This is backed by the work
on Cognitive Load Theory’
[
3
]. Even [
75
] points out that the cognitive load theory deals
with understanding, learning, and complexity.
Next, the cognitive load theory is presented below, as this thesis deals with the cognitive
capacity in terms of comprehensibility while reading business process models.
3.2 Cognitive Load Theory (CLT)
A requirement for coping with business process models is that they are understandable
and easy to process. In this context, the human working memory system has to handle
19
3 Fundamentals of Cognitive Complexity
the perceived information. Hence, the cognitive load theory of [
76
] can be applied. This
theory was generated to support the activities in learning through guidelines on how
information can be presented in order to support a better intellectual effort [77].
’Cognitive load theory (CLT) is a learning and instruction theory that describes instruc-
tional design implications of a model of human cognitive architecture based on a perma-
nent knowledge base in long-term memory (LTM) and a temporary conscious processor
of information in working memory’ [78].
The capacity of the human memory system is limited [
79
]. This is one of the assumptions
on which the CLT is based. [
80
] generated a graphical representation that is provided
in Figure 3.2. This representation includes the scheme construction of central learning
processes, next to the already mentioned limited capacity of the working memory.
Figure 3.2: Theoretical basis of the Cognitive Load Theory of [80]
The CLT addresses the topic cognitive load, which can be divided into the mental sources
of impact:
intrinsic
,
extraneous
, and
germane cognitive load
, according to [
77
]. They
are discussed in detail below.
•Intrinsic Cognitive Load (ICL)
This source of impact deals with the intrinsic information complexity [
81
]. Their
focus is on the complexity, difficulty, and scope of a task [
80
]. In literature, two
factors with an effect on intrinsic cognitive load are mentioned: element interactivity
and the prior knowledge of the learner. An example for element interactivity exists
20
3.2 Cognitive Load Theory (CLT)
in language learning. [
82
] points out that low element interactivity is given when a
person has to repeat single vocabularies. Therefore, they argue that element inter-
activity is higher by using the right grammar in generating sentences. It becomes
obvious that element interactivity should be low. Low element interactivity is better
understandable and separately learnable [
83
], as the learners don’t have to link
all of the elements with other elements [
84
]. However, high element interactivity
causes high cognitive load [
84
]. In the following, an example in context of business
process models is presented. Single parts of a process can be independently
understood.
Prior knowledge of learning is the experience that a learner has during the interac-
tion with materials that have to be learned. As an example, an expert in business
process modeling and reading can construct chunks. Hence, there’s no need to
memorize the single elements.
•Extraneous Cognitive Load (ECL)
This source of impact deals with the design of materials provided to the learner
[
82
,
85
]. It is reduced when progresses are simplified by the design. In busi-
ness process models it is dealt with using symbols of the modeling notation, the
execution direction, and the complexity.
•Germane Cognitive Load (GCL)
This source of impact deals with knowledge acquirement [
80
] and the necessary
consumption of cognitive resources for appropriation of learning content [
86
]. To
receive high germane cognitive load, learners have to be dedicated, and lead
their mental model to the progress of learning [
82
]. Therefore, the behavior while
learning is necessary to receive good results while reading a business process
model.
The composite load of the three mentioned types can’t transcend the memory resources,
because they are additive [
87
]. This means that the loads relate to each other and that a
load capacity exists. [
88
] points out that there is a need to know the level of the learners
knowledge. Hence, the intrinsic, germane, and extrinsic cognitive load can lead to the
21
3 Fundamentals of Cognitive Complexity
right learning outcomes. Further, the germane cognitive load should be high, as it is the
only load that has positive effects on learning.
It is obvious that the design of business process models and modularization can have
an impact on the cognitive load. In the next section, the relation of eye behavior and
cognitive load is presented. The neural behavior of the eye is influenced, among other
things, by cognitive load. Further, the eye can provide insights regarding the occurrence
of problems in evaluating the presented learning materials.
3.3 Relation of Eye Behavior and Cognitive Load
The eye has a complex structure [
89
] and behavior. It’s behavior depends, on the one
hand, on two muscles that control changes in pupil size. According to [
89
] these muscles
are, as presented in Figure 3.3, the dilator and sphincter pupillae muscle. The change in
size is caused by stimuli, impulses [
90
], cognitive processes/control [
91
], and emotions
[
92
]. On the other hand, eye movement is present. It is caused by switching between
different points of interest. Further, it provides input about attention and demands while
processing [93].
Figure 3.3: Cognitive Pupillometry [90]
22
3.3 Relation of Eye Behavior and Cognitive Load
A well known approach to measure cognitive load, based on certain cognitive processes
is given by eye tracking [93]. For this purpose, the following behaviors are recorded:
•Pupil dilation
Pupil size is influenced by mental effort [
91
,
94
]. Next to mental effort, [
94
] refers
to the subjective difficulty of the task and neural gain. Due to high parasympathetic
activity (i.e. mental effort), an innervation of pupil results (see Figure 3.3). An
example in reading business process models is the innervation of the pupil when
the learner has difficulties in understanding the shown circumstances.
•Fixation
The eye is looking at a stable point [
95
] of a visual stimuli for a period of 200-300
ms [
96
]. The level of experience in a task has an impact on the duration of fixation
[97]. Therefore, duration is a reference of information processing. When a reader
of a business process model fixates a process activity, the duration is measured
by the eye tracker. This duration gives information about the necessary time to
process the information presented. If the learner needs a long time, then the
cognitive load is high.
•Saccades
Between individual fixations, eye movement occurs. The time of eye movement
is a saccade [
95
] and executed rapidly [
98
,
96
]. A reduction of saccades is good
for efficiency [
94
], because no information is recorded during eye movement [
96
].
An example for a saccade is the movement of the eye from fixating one activity to
fixating another.
In case of measuring understandability in business process models, diverse research
exists. Eye tracking is also utilized by [
99
]. They set up an eye tracking study to receive
information about fixation time, etc. Hence, it becomes obvious whether the participant
only refers to a single section of the process model, and how cognitive capacities are
investigated. In the next section, a study is planned and defined.
23
4
Study Planning and Definition
The goal of the study is defined in Section 4.1. Section 4.2 considers the formalized
hypotheses. Section 4.3 refers to the setup, Section 4.4 picks up the study design.
Finally, the risks of analysis and migrations are discussed in Section 4.5.
All of these aspects are necessary to answer a research question. In order to re-
ceive meaningful findings, attention must be paid to the methodology of the monitoring
procedure. This requires detailed planning and definition.
4.1 Goal Definition and Context Selection
In regard to process modeling, a high amount of research represented by metrics [
100
],
guidelines [
40
], etc. exists. It provides the opportunity to generate adequate process
models of high quality. Next to the completeness and modeling quality, process models
have to be understood by all stakeholders. In this case, complexity caused by the limited
capacities in human information processing, is a main factor. To handle the complexity
of process models, three modularization approaches exist: horizontal, vertical, and
orthogonal [
1
]. They reference on ’how’ to modularize. Each modularization approach
takes up different subdivision approaches which differ in terms of design and the intended
use.
So far, a great amount of scientific research in terms of modularization was not addressed.
While the usability of the modularization approaches were compared, comparisons in
terms of the offered complexity are missing. This leads to the research question:
25
4 Study Planning and Definition
RQ1: Do different modularization approaches in business process modeling have an
impact on cognitive complexity of process readers?
Next to the cognitive complexity, it is of interest which modularization approach provides
a better overall produced result after performing the task. Hence, items that refer to
the performance success like the time needed for execution and accuracy measured by
tasks are of interest. This leads to the research question:
RQ2: Do different modularization approaches in business process modeling have an
impact with respect to a successful performance of process readers?
Finally, activities used in process modeling have a firmly defined syntactical meaning
for each business process modeling notation. Items could be misunderstood by the
stakeholders. Further, differences in the single modularization approach exist. For
vertical modularization, [
47
] points out that expanded subprocess provide a better
understandability than collapsed subprocesses. To receive insights whether the design of
a single modularization approach affects the cognitive complexity, the following research
question is placed:
RQ3: Do variations in representation of different modularization approaches in busi-
ness process modeling have an impact with respect to cognitive complexity of
process readers?
This results in the goal:
Obtaining insights in terms of the cognitive load when stakeholders read different modu-
larized business process models.
To address the research questions, a survey and an eye tracking study are applied.
Surveys
provide the benefit to obtain the opinion and attitude of a participant with little
expenditure [
101
]. Through
studies
, profound insights can be considered. This can be
realized through technical realizations such as
eye tracking
. With
eye tracking
cognitive
processes of the human can be measured. Such a measurement is based on visual
stimuli [
102
]. Hence, participants can read a business process model, and it becomes
obvious where problems in processing the model exist. An evaluation is comprised of,
for example, fixations and saccades.
26
4.2 Hypotheses Formulation
4.2 Hypotheses Formulation
The hypothesis is an unequivocal assumption concerning the studies’ result [
103
]. Two
types of hypotheses have to be taken into account. These are the
null hypothesis
, and
the alternative hypothesis.
The
null hypothesis H0
points out that no difference or effect between the considered
conditions exists [104]. Therefore, the goal of studies is to refute it.
The
alternative hypothesis H1
argues that a difference or effect between the considered
conditions occurs [104]. The goal of the studies is to prove it.
RQ1 is subdivided into the cognitive load and understandability. RQ2 addresses the
performance success, and RQ3 the design. Subsequently, hypotheses and their related
questions, that are utilized in this topic, are shown.
Cognitive Load
In Section 4.1, the main research question of the thesis is presented. It utilizes the
cognitive load - a composite formed by the intrinsic, germane, and extraneous cognitive
load - as the topic. Therefore, subquestions can be applied:
Q1.1: Do different modularization approaches influence the intrinsic cognitive load by
executing the task?
All conditions provide the same complexity in the studies with respect to their tasks.
Further, the study design (Section 4.4) was designed in a way to evade effects, such as
learning effects. Further, for data evaluation it is ensured that only slight deviations in
prior knowledge emerge.
H1.1 (a):
No significant difference is given between the different modularization ap-
proaches in terms of the intrinsic cognitive load.
H0:µV=µH=µO
Next to complexity and prior knowledge, the intrinsic cognitive load depends on element
interactivity. Therefore, the functionality of different activities has to be understood.
27
4 Study Planning and Definition
Caused by the differences in the modularization schema, for example, in terms of icons,
a difference in intrinsic cognitive complexity could result.
H1.1 (b):
A significant difference between the modularization approaches is given in
terms of the intrinsic cognitive load.
H1.1(b):µO> µH> µV
Q1.2: Do different modularization approaches influence the germane cognitive load by
executing the task?
A good design has positive effects on the efforts by executing the tasks. The performance
in carrying out the task could vary regarding differences in representation, for instance by
selecting different icons. Therefore, a difference between the modularization approaches
is expected.
H1.2 (a):
A significant difference is given between the modularization approaches in
terms of the germane cognitive load.
H1.2(a):µO> µH> µV
If no difference between the modularization approaches is obtained, the alternative hy-
pothesis has to be rejected and the null hypothesis occurs. Therefore, no modularization
approach leads to higher effort by executing the task.
H1.2 (b):
No significant difference is given between the modularization approaches in
terms of the germane cognitive load.
H0:µO=µH=µV
Q1.3: Do different modularization approaches influence the extraneous cognitive load
by executing the task?
Because of the different representations of the modularization approaches, for example
through icons, a difference in the extraneous cognitive load is expected.
H1.3 (a):
A significant difference is given between the modularization approaches in
terms of the extraneous cognitive load.
H1.3(a):µO> µH> µV
28
4.2 Hypotheses Formulation
However, the basic activities are the same, and only small differences between icons
exists. This could cause that no measurable difference is given and the null hypothesis
occurs.
H1.3 (b):
No significant difference between the modularization approaches in terms of
the extraneous cognitive load occurs.
H0:µO=µH=µV
Understandability
Next to the cognitive load, the understandability of process models is of interest. It leads
to the answer whether the represented modularization approaches in business process
models could be, for example, reproduced by the participants, or not.
Q2.1: Do different modularization approaches influence the perceived usefulness for
understandability?
An effect, caused by the different representations, in extraneous and germane cognitive
load is expected. The deviations in representation could have an impact regarding
the understandability. Hence, a difference between the modularization approaches, in
terms of the perceived usefulness for understandability, is expected. This leads to the
hypothesis:
H2.1 (a):
A significant difference between the modularization approaches in terms of
the perceived usefulness for understandability is given.
H2.1(a):µO> µH> µV
In case of little deviations in terms of the representation exist, it could be that no
significant difference will be observed and no effect is measurable. This would cause the
null hypothesis:
H2.1 (b):
A significant difference between the modularization approaches in terms of
the perceived usefulness for understandability is given.
H0:µO=µH=µV
29
4 Study Planning and Definition
Q2.2: Do different modularization approaches influence the perceived ease of under-
standing?
Perceived ease of understanding focuses on the reproduction and understandability of
the process models. The different modularization approaches could lead to different
results because of the deviations in representation. Therefore, the following hypothesis
is given:
H2.2 (a):
A significant difference is given between the modularization approach in terms
of the perceived ease of understanding.
H2.2(a):µO> µH> µV
In cases of minor deviations in representation, no difference in the perceived ease of
understanding is obtained. This would lead to the occurrence of the null hypothesis:
H2.2 (b):
No significant difference between the modularization approach in terms of the
perceived ease of understanding occurs.
H0:µO=µH=µV
Performance Success
So far, the cognitive load and understandability were considered. Further, the perfor-
mance success can be measured. The performance success is based on the achieved
performance while executing the task.
Q3: Do different modularization approaches influence the performance success?
Because of the expected differences in perceived usefulness for understandability,
perceived ease of understanding, and cognitive load, a further difference in terms of the
performance success is expected. Therefore, the following hypothesis is given:
H3 (a):
A significant difference between the modularization approaches in terms of the
performance success is given.
H3(a):µO> µH> µV
If the other measured objects would not provide a significant difference, no signifi-
cant difference for the performance success is expected. This would lead to the null
hypothesis:
30
4.2 Hypotheses Formulation
H3 (b):
No significant difference between the modularization approaches in terms of the
performance success occurs.
H0:µO=µH=µV
Design
Next to the differences regarding the modularization approaches, a difference with
respect to the single modularization approach is given. Therefore, next to the existing
designs, further icons were designed for each modularization approach. Hence, the
quality between the activities of a modularization approach can be compared with the
self designed icon.
Q4.1: Does the design of referencing the modularized area influence the performance
in horizontal modularization?
In horizontal modularization, only small differences with respect to the design are given.
Hence, no difference in cognitive load is expected.
H4.1 (a):
No significant difference with respect to the horizontal modularization in terms
of design and performance success occurs.
H0:µM=µL=µSD
As differences in the representation exist and additionally the complexity varies, the
alternative hypothesis could occur:
H4.1 (b):
A significant difference with respect to the horizontal modularization in terms
of design and performance success exists.
H4.2(2b):µM> µM> µSD
Q4.2: Does the design of referencing the modularized area influence the performance
in vertical modularization?
In vertical modularization, only small differences with respect to the design are given.
Therefore, no difference in cognitive load and performance success is expected.
31
4 Study Planning and Definition
H4.2 (a):
No significant difference with respect to the vertical modularization in terms of
design and performance success is given.
H0:µC=µE=µSD
[
47
] points out that differences for vertical modularization with respect to the understand-
ability exist. Therefore, a significant difference regarding the hypothesis is expected.
H4.2 (b):
A significant difference is given with respect to the vertical modularization in
terms of design and performance success.
H4.2(2b):µC> µSD > µE
Q4.3: Does the design of referencing the modularized area influence the performance
in orthogonal modularization?
In orthogonal modularization, differences regarding the design exist. Partially, symbols
exist that reference to modularization. Therefore, a difference in cognitive load and
performance success is expected.
H4.3 (a):
A significant difference with respect to the orthogonal modularization in terms
of design and performance success exists.
H4.3(2a):µAO > µSD > µE
The differences could be realized at least equivalently. This would lead to the null
hypothesis:
H4.3 (b):
No significant difference with respect to the orthogonal modularization in terms
of design and performance success exists.
H0:µAO =µSD =µE
4.3 Study Setup
Studies are the realization for measuring effects on dependent variables by variating
independent variables [101].
This section refers to the studies’ setups, subdivided into the research variants
survey
and
study
. Both are part of this thesis. This leads to the first subject-matter ’Selection of
32
4.3 Study Setup
Studies’. As the fundamentals of studies are attributed by the knowledge of response
variables, they are subsequently considered. Further, the selection of participants,
objects, questionnaires, and instrumentations are additional subject-matter.
Selection of Studies
Different types of studies exist. Survey and study are only two of them.
Survey:
A survey is applied to acquire findings such as opinions and attitudes [
101
].
If the survey is executed online, a large participatory number can be acquired with
little expenditure [
101
]. Hence, the participant is neither spacially bound, nor bound by
time (until the data collection has been completed). Summarized, a survey is useful for
superficial and fundamental considerations.
Study:
A study is chosen when profound insights have to be considered [
101
]. It provides
the benefit that confounding and environmental variables can be controlled, depending
on the study design [
105
]. Furthermore, the researcher is able to obtain data from the
participant, for instance by thinking aloud [
106
]. However, the recruitment process of
participants can be difficult [107].
In this thesis both study and survey are considered. The survey is applied to provide
insights in terms of the differences between modularization approaches. The data col-
lected should answer the research question RQ1 through measurements on the following
aspects: cognitive load, perceived usefulness for understandability, and perceived ease
of understanding. In addition, alternative representational designs can be questioned.
Hence, a utilization of design changes is possible in further research.
In the study, profound insights about the cognitive complexity - independent of the
intrinsic, extraneous, and germane cognitive load - can be measured. Therefore, the
utilization of eye tracking technologies is possible. Using an eye tracker, neuroscientific
information is measured providing insights regarding cognitive processing. It becomes
obvious where the differences in understanding come from, and where difficulties occur.
Hence, the study answers the research questions RQ2 and RQ3.
Definition of Response Variables
The studies that are utilized in this thesis were presented, so far. Next, the responsible
variables with respect to the single studies have to be defined. They are the basis for
33
4 Study Planning and Definition
each of the studies, and are important to reach their goal. Therefore, a differentiation
between the response variable types has to be considered. These types are:
•Independent Response Variable
The independent response variable is manipulated through the researcher in
studies [108]. Except of this variable, an equal treatment is of need [109].
•Dependent Response Variable
The dependent variables are measured [
108
] on participants [
109
] in studies. It
is measured to receive insights regarding the influence through the independent
variable on the dependent variable [
109
]. Hence, they provide a dependency to
the independent ones that exist.
The variables selected for survey and study are presented in detail below.
Survey:
The independent response variable is the modularization approach with the
levels: horizontal, vertical, and orthogonal. Further, prior knowledge and education relate
to the independent response variables. Cognitive load, perceived ease of understanding,
and perceived usefulness for understandability are chosen as dependent response
variables.
Study:
Independent response variables comprise of the prior knowledge, the education,
and the items of cognitive complexity. However, the independent response variable
contains the modularization approach with the levels: horizontal, vertical, and orthogonal.
Further, the designs of the modularization approach levels are measured. These are:
•Horizontal: link events, message events, self designed
•Vertical: collapsed subprocess, expanded subprocess, self designed
•Orthogonal: aspect-oriented, exception event, self designed
Dependent response variables in the study are: cognitive load, perceived ease of
understanding, perceived usefulness for understandability, and the performance success
(time, reached score). Further, dependent response variables exist. They result from
the measurement of the eye tracker. These are, for example: average fixation, fixation
duration, and fixation counts.
34
4.3 Study Setup
Next to the dependent response variables, control variables like the age, gender, and
knowledge are measured in both studies - survey and study.
Selection of Participants
A sample regarding the population is selected in studies. This is reasonable because
participants should reflect the population’s required characteristics [110].
Readers of business process models are not necessarily experienced in business
process models [
28
]. This is also the case for other stakeholders. Therefore, persons - in
form of students, research assistants, and alumni - that are not necessarily experienced
in modularized business process models will be selected as participants.
Survey: In the survey, the mentioned representative group is selected as participants.
Study:
In the study, the restriction that only the mentioned research group will take part
is observed.
The acquisition of participants is represented in Section 5.1.
Selection of Materials
The materials are generated. The focus in preparing the objects is the knowledge and
expertise of the participants. Hence, process models were prepared with a simple and
understandable content, such as baking a pizza. Further, an adequate business process
modeling notation has to be selected. Not every process modeling notation supports
each modularization approach. Therefore, the widespread business process modeling
notation BPMN 2.0 [
31
] is applied. It is often utilized as an object of study in scientific
papers regarding business process models. In addition to its frequent use, it is also
selected due to the extensive scope of functions. Furthermore, it enables the utilization
of all three modularization approaches that have to be considered. The materials are
realized in German, as it is the common language in which the studies are executed.
The business process models were realized using the modeling tool ’Signavio’ [111].
Survey:
The survey is comprised of twelve process models as presented in Table 4.1.
For each modularization approach, four process models are prepared. Every process
model of the four can be mapped to a process model of the other approaches. Hence,
they are comparable.
35
4 Study Planning and Definition
With respect to the first process model ’Order’, modularization was represented by
message events, expanded subprocesses, and exceptions. In the process models ’Re-
fuel’, ’Pizza’, and ’Loan’, the modularization was realized through link events, collapsed
subprocesses, and aspect-oriented paradigm. The representation of the process models
is illustrated in Appendix A.
Process
Model
Horizontal Modularization Vertical Modularization Orthogonal Modularization
Order Message event Expanded subprocess Exception
Refuel Link event Collapsed subprocess Aspect-oriented
Pizza Link event Collapsed subprocess Aspect-oriented
Loan Link event Collapsed subprocess Aspect-oriented
Table 4.1: Process Models in the Survey and their Realization
Study:
The study picks up 81 process models, 9x9 as shown in Figure 4.1. While
nine processes differ regarding the content (P1 - P9), nine process models vary in the
realization of modularization. The realization of modularization is subdivided into three
modularization approaches. These are the orthogonal (O1 - O3), vertical (V1 - V3), and
horizontal (H1 - H3) modularization. Each approach is divided into three weightings:
changed by self designed (light gray), not changed (gray), and self designed (white).
Changed by self designed involves the modularization subdivisions presented in light
gray. These are:
aspect-oriented
,
collapsed subprocesses
, and
link events
. They are
not changed so far. The change is first utilized in the weighting self designed. In the
self designed weighting, items that refer to modularization were changed/supplemented.
However, the subdivisions represented in gray (
exception events
,
expanded subpro-
cesses
,
message events
) were utilized unchanged. Further, they were not changed in
successive subdivisions. As exemplified in Figure 4.1, the orthogonal modularization is
subdivided into aspect-oriented, exception event, and self designed. The aspect-oriented
and exception event are represented according to the figures in Section 2.2.3. In contrast,
self designed is based on small icon changes on the aspect-oriented approach. Hence,
for each of the presented modularization approaches, the self designed process model
(in Figure 4.1 represented through the 3) is based on the first weighting in Figure 4.1,
and provides minor adjustments of the icons.
36
4.3 Study Setup
Figure 4.1: Graphics in the Study and its Realization
Furthermore, the content of process model one (P1) is realized analogically for each
modularization approach and their subdivisions. Each graphic is comprised of four self
generated statements. The single statement can be answered with true or false. In
addition, the weighting between true and false is predominantly balanced. The process
models and the questions can be extracted from Appendix B.2 up to B.10. The design of
the eye tracking materials, created using the program ’Keynote’ [
112
], is presented in
Figure 4.2.
Gray is selected for the background because of the lower brightness reducing strain on
the eyes. Next, the positioning of the elements inside the material is explained. The
placement of the process model (yellow sector) is positioned in the upper center for each
image. The statement is placed bottom center. This statement has to be assessed with
true or false. Hence, the possibilities are presented to the participants. Underneath the
statement, true is placed on the left, while false is presented on the right. Arrangement
37
4 Study Planning and Definition
Figure 4.2: Structure of Eye Tracking Material
and presentation of the possibilities is decided by the input method of the assessment:
on a German standardized keyboard layout (QWERTZ). The character ’F’ has to be
selected for true, and ’J’ for false. For each correct answer the participant receives one
point. However, for an incorrect answer the participant won’t receive a point. All together,
36 points can be obtained.
Selection of Questionnaires
Cognitive load and the understandability are of relevance in cognitive complexity. This
leads to the usage of questionnaires that address these themes.
For cognitive load, a reliable questionnaire of the Department of Learning and Instruction
of the Ulm University is used [
113
]. It’s composed of three items to measure the
extraneous, two items to measure the intrinsic, and three items to measure the germane
cognitive load. In addition to the cognitive load, items that measure the motivation are
utilized. At each of these items, an initial- and endpointscore seven-point likert-scale is
applied. Table 4.2 show the items of the questionnaire.
Next to the cognitive load, understandability is of relevance. Therefore, two aspects will
be measured: perceived usefulness for understandability (PUU) and perceived ease of
38
4.3 Study Setup
Typ Original Question English Question German
ICL For this task, many things needed to be kept in
mind simultaneously.
Bei der Aufgabe musste man viele Dinge gle-
ichzeitig im Kopf bearbeiten.
ICL This task was very komplex. Diese Aufgabe war sehr komplex.
GCL I made an effort, not only to understand several
details, but to understand everything correct.
Sie haben sich angestrengt, sich nicht nur
einzelne Dinge zu merken, sondern auch den
Gesamtzusammenhang zu verstehen.
GCL My point while dealing with the task consisted of
elements supporting my comprehension of the
task.
Es ging Ihnen beim Bearbeiten der Lernein-
heiten darum, alles richtig zu verstehen.
GCL* The learning task consisted of elements support-
ing my comprehension of the task.
Die Lerneinheit enthielt Elemente, die Sie unter-
stützten, den Lernstoff besser zu verstehen.
ECL During this task, it was exhausting to find the im-
portant information.
Bei dieser Aufgabe ist es mühsam, die wichtig-
sten Informationen zu erkennen.
ECL The design of this task was very inconvenient for
learning.
Die Darstellung bei dieser Aufgabe ist ungünstig,
um wirklich etwas zu lernen.
ECL During this task, it was difficult to recognize and
link the crucial information.
Bei dieser Aufgabe ist es schwer, die zentralen
Inhalte miteinander in Verbindung zu bringen.
Table 4.2: Items to measure Cognitive Load
understanding (PEU). Both include four items. Each item utilizes a seven-point likert-
scale with initial- and endpointscore labeling. The questions are withdrawn by [
114
]
and find application with respect to papers from the Eindhoven University of Technology.
For this thesis, the questions were translated into German as the survey and study are
conducted in German. The items and their translation are presented in Table 4.3.
Next to the cognitive complexity, a questionnaire is of need for comparing the participants.
These comparisons can be handled through control variables. Hence, a demographic
questionnaire is utilized [115].
Survey:
Next to the demographic questionnaire that is extended by the item ’design’,
two questionnaires are used.
•Cognitive Load Theory
Changes in the survey: Terms, for example ’Aufgabe’ (task), were changed to
terms such as ’Prozessmodell’ (process model).
•
Perceived Usefulness for Understandability and Perceived Ease of Understanding
The questionnaire is integrated into the survey and can be extracted from Appendix A.
Study: Next to the demographic questionnaire, two questionnaires are utilized.
39
4 Study Planning and Definition
Typ Question English Question German
PUU Business process models represented in this
way would be difficult for users to understand.
Auf diese Weise repräsentierte Geschäft-
sprozessmodelle wären für die Benutzer schwer
zu verstehen.
PUU I think this presentation approach provides an
effective solution to the problem of representing
business process models.
Sie denken, dass dieser Präsentationsansatz
eine effektive Lösung für das Problem der
Darstellung von Geschäftsprozessmodellen bi-
etet.
PUU Using this type of process models would make
it more difficult to communicate business pro-
cesses to end-users.
Die Verwendung von Prozessmodellen dieser
Art würde die Kommunikation von Geschäft-
sprozessen an den Endbenutzer erschweren.
PUU Overall, I found the business process model in
this experiment to be useful.
Insgesamt haben Sie das Geschäftsprozess-
modell in diesem Experiment als nützlich er-
achtet.
PEU Learning to use this way of modelling business
processes would be easy for me.
Diese Art der Modellierung von Geschäft-
sprozessen zu erlernen, wäre für Sie einfach.
PEU I found the way the process is represented as
unclear and difficult to understand.
Sie halten die Darstellung des Prozesses für un-
klar und schwer verständlich.
PEU It would be easy for me to become skilful at using
this way of modelling business processes.
Es wäre leicht für Sie, diese Art der Modellierung
von Geschäftsprozessen zu beherrschen.
PEU Overall, I found this way of modelling business
processes difficult to use.
Insgesamt hat sich diese Art der Modellierung
von Geschäftsprozessen als schwierig erwiesen.
Table 4.3: Items to measure Perceived Usefulness for Understandability and Perceived
Ease of Understanding
•Cognitive Load Theory
Further questions are applied. These questions comprise the mental effort, diffi-
culty of executing the task, as well as motivation and enjoyment.
•
Perceived Usefulness for Understandability and Perceived Ease of Understanding
All questionnaires are demonstrated in Appendix B.11.
Selection of Technical Instrumentation
The instrumentation engages technologies used in the studies.
Survey:
As an online survey is utilized, a computer with access to the internet is
necessary to participate in the survey. The online survey was developed using the tool
’Google Formulare’ [
116
]. It is appropriate, as it is an open source tool, and easy to use.
Study:
In the study, an eye tracker is utilized (Section 4.5). The eye tracker used is a
stationary tower mounted eye tracking system, of the type SMI iView X Hi-Speed, from
SensoMotoric Instruments. It has a sampling rate of 240Hz. The benefits of such a
system are that changes in pupil size and fixation are measured frequently. Further,
40
4.4 Study Design
the eye is focused on the visual stimuli presented. As the head of the participant is
stabilized through the chin tray and the cranial is leaning against the headrest of the eye
tracker only little head movements are possible. The data measured during the study is
analyzed through the SMI BeGaze software.
4.4 Study Design
The study design relates to the structure given while running the study. The usage of an
inadequate structure could cause problems regarding the progress of the study and also
leads to its failure. Therefore, three things are considered:
1. First, the subject design. It includes two different approaches.
Between Subject Design
One group executes exactly one condition. For comparison, more participants are
of need. But no learning effects exist, as no participant can map the content of one
condition to the content of another condition.
Within Subject Design
Every participant has to conduct every condition. This leads to a better comparison
and the need of less participants. However, an impact of learning effects can occur,
because the participant can map the acquired knowledge from the first condition
to the ensuing conditions. This leads to a more successful performance in the sub-
sequent conditions. The eradication of learning effects can be realized as good as
possible through latin square. Based on latin square, after every accomplishment
the conditions are shifted by the factor one. Hence, a systematically change in the
experiment is executed.
2.
In addition to the different subject designs, four principles have to be considered:
balancing,randomizing,matching,blocking.
Balancing
Through balancing, (random) sequence effects like the intern validity and test
effects are controlled [117].
41
4 Study Planning and Definition
Randomizing
Through randomizing, differences between participants and the thereby arising
interferences are randomly distributed [118].
Matching
Through matching, differences between participants are obviate. This is realized
through distributing persons with similar characteristics to the different groups
[119].
Blocking
Through blocking, less participants are necessary. Blocks based on a character-
istic are selected and participants are assigned to the different units along this
characteristic [120].
3.
Next to the items mentioned, the number of factors is of interest in the experiment
design.
Single Factor
A single factor design is given when an experiment has exactly one independent
variable [121]. The number of levels is not of interest.
Multifactorial
A multifactorial design is given when an experiment has more than one independent
variable. If a representation like 2x4 is stated, two independent variables exist.
One of them has two and the other one four levels.
Survey:
In the survey, a single factor between subject design finds use. Therefore,
three online surveys, differentiated by the modularization approach, are designed. The
structure of the study is presented in Section 5, Figure 5.1.
Study:
In the study, a 3x3 within subject design finds use. The structure of the study is
represented in Figure 5.2.
42
4.5 Risk Analysis and Migrations
4.5 Risk Analysis and Migrations
In studies, different factors have to be taken into account to receive a reliable result. This
can be caused by the main quality criteria.
Reliability
It is missing when subsequent measurements have differences in their findings [
122
].
Hence, the research results are refuted as they do not correspond to the actual circum-
stances.
Objectivity
It is missing, when the research results are caused by aspects such as feelings and
beliefs [123] while executing the studies.
Validity
It is missing when the measurement doesn’t lead to the research result that has to
be investigated [
122
]. Types that have to be taken into account are, for example, the
internal (caused effects of treatment (A →B)), external (information about generalizing
elicited results), conclusion (relation of treatment and outcome), and construct (quality of
measurement) validity. In Figure 4.3, the relation between the mentioned validity types is
shown. In the following, the validities of the studies are considered.
•Internal Validity
Internal validity can be negative effected by the materials, technical instrumenta-
tions, and participation, next to further aspects. All of these aspects are given by
the reconciliation from the independent to the dependent variable.
The designed materials can lead to negative effects on internal validity. Therefore,
an extensive research with respect to modeling modularized business process
was executed. Furthermore, the designed business process models were kept
simple for the studies. This provides an adequate understanding. As an extension
of the eye tracking study, the position of the question was of interest. The details
represented on the particular unit of the study were carefully located.
43
4 Study Planning and Definition
Figure 4.3: Different types of Validity [124] adapted
Next to the designed materials, questionnaires could effect the internal validity.
For both studies, reliable, multiple tested questionnaires were selected. Further,
they consist of items that measure the necessary attributes to answer the research
question. Besides, they were inserted at well selected positions.
For both studies, the technical instrumentation can lead to negative effects on
internal validity. As the survey is executed online, the technical instrumentation
is of interest. A survey tool with a simple application was utilized. Therefore, the
necessary actions are obvious at each point in time. Further, the material was
subdivided into logical units. The technical instrumentation in the study is given by
using an eye tracker. For each participant the eye tracker is adapted accordingly.
To receive adequate data, a stationary eye tracker was utilized. This eye tracker
ensures only little head movements and has a high sampling rate. Therefore, data
loss is low. However, with respect to the eye tracker, problems could be caused
by pupil feedback. There are three problems according to [
125
]: data loss, big
amounts of data, and pupil diameter at baseline registration. All three problems
are known and considered accordingly in the evaluation.
44
4.5 Risk Analysis and Migrations
Finally, negative effects on internal validity can be caused by participants. Di-
vergence of experience, such as environmental variables should be as little as
possible. In the survey, environmental variables and an equal distribution of partici-
pants can’t be controlled. Furthermore, the experience of participants can effect
the internal validity. Hence, the experience should vary as little as possible through
equal distribution of the participants. Further, the design of the survey is crucial
for the impact of the experience. As a between subject design is utilized, a high
number of participants has to be selected for randomization. As an alternative,
matching would be an opportunity. In the study, the environmental variables such
as place and time are controlled. Further, a within subject design is utilized. Hence,
the individual divergences with respect to experience have little impact, as every
participant passes each condition. A possible effect is caused by the experience
of the individual participant, as he could be experienced in one modularization
approach. That would impact the other approaches. Therefore, a high number of
participants have to be selected.
•External Validity
External validity can be negatively effected by aspects such as the surroundings,
participants, and temporal aspects.
First, the surroundings can lead to negative effects on the external validity. Hence
environmental variables like loudness and light invasions are considered. During
data collection, these variables can’t be influenced as the participant can execute
the survey at any point in time and place. In the study, a labor study enables the
controlling of environmental variables. Therefore, each participant carries out the
study under the same conditions. Further, the study is carried out during the day.
Hence, temporal aspects are controlled.
Finally, the selection of participant is of interest. Business process model stake-
holders are not necessarily experienced. Especially the usage of all three modular-
ization approaches is not implicitly given. Therefore, participants with little to no
knowledge regarding modularization can be utilized to avoid differences based on
prior knowledge of individual approaches. Further, process models are realized
45
4 Study Planning and Definition
in the modeling notation BPMN 2.0. This is a well known modeling language that
supports all three modularization approaches.
•Conclusion Validity
Conclusion validity can be negatively affected by low statistical power, violated
assumptions, and non-normalized procedures.
The statistical power can lead to negative effects on the conclusion validity. There-
fore, the level of significance was kept by
p <
.05. Hence, the probability regarding
the null hypotheses was kept high.
Further, a violated assumption can lead to negative effects on the conclusion
validity. It is caused by the tests chosen for evaluation. Hence, the selection of
evaluational tests has to be done carefully and considering the obtained results.
Finally, a procedure that is not normalized could have negative effects on the con-
clusion validity. Therefore, the procedure is normalized throughout all participants.
•Construct Validity
The construct validity can have negative effects caused by the participant behavior
and measurement. Further, the study leaders can have an influence.
The construct validity can be negatively affected by participants. As the participant
is in a test situation, he could behave in a different way than normally. In the survey,
the participant is anonymous. This leads to the benefit that the participant does
not feel observed. Hence, he answers the questions honestly. In the eye tracking
study an interaction between the participant and the study leaders is present. As
an example, it is previously defined what information the participant receives in
the introduction. To avoid different behaviors towards the participants, the study
leaders have defined habits. Hence, no difference should arise. Furthermore, the
participants should be encouraged to behave as normally as possible with the help
of the study leaders’ behavior. Further, no indications in terms of the measured
response variables were provided to the participant. This is especially taken into
account in the introduction.
46
4.5 Risk Analysis and Migrations
Measurements were carefully selected. Further, the questionnaires were already
utilized in other empirical studies.
47
5
Study Operation
This chapter is subdivided into three sections. Sections 5.1 describes the preparation of
the study. Afterwards, Section 5.2 discusses the execution of the study. The last section
in this chapter, Section 5.3, entails data validation.
5.1 Study Preparation
After creating the materials and setting up the study, its preparation is required. In this
section, pilot studies and the recruitment process of participants are considered.
Survey:
Before starting the survey, two pilot studies were executed to obviate errors, to
increase quality, and to test technical functions such as data collection. Based on these
results the survey is adjusted.
The recruitment of participants was carried out during lectures of the Institute of
Databases and Information Systems (DBIS). As remuneration, participants of the survey
gained a bonus point of an exercise sheet in the lecture
Business Process Management
.
Study:
Two pilot studies were performed before executing the study. Hence, it was
possible to eliminate problems and mistakes. Further, the opportunity was given for
increasing the quality of the study.
Friends and acquaintances were acquired for participation through lettering. Further,
students and research assistants were asked to participate. As only one participant can
perform the study at a point in time, participants have to select a date via ’Doodle’ [
126
]
to avoid collision. Each participant receives chocolate for participating in the study.
49
5 Study Operation
5.2 Study Execution
The study execution addresses the structure that a participant carries out in the re-
spective study. Participating in one of the studies, as well as the participation at a pilot
progress, excludes the individual from further participation.
Survey:
After acquiring the participants through the lecture
Business Process Manage-
ment
, the participants have to progress the survey, as represented in Figure 5.1. For
execution, an internet-enabled terminal, such as a mobile phone or computer, is of need.
This is because participants access the studies via the learning management system
’Moodle’ [
127
]. After a participant clicks on the link provided in Moodle, he is randomly
assigned to one of the groups. The random assignment process is possible as it is
supported by the learning management system.
Figure 5.1: Survey Procedure
Next, the survey design is shown, as presented in Figure 5.2. After the participant is
assigned to one of the groups, the introduction is presented. The introduction covers
entities that contains information like the procedure, the note on voluntary, and the
declaration of substitution. Afterwards, a demographic questionnaire [
115
], as presented
in Section 4.4, is given. In the process model evaluation, the participant has to evaluate
a process model, before he has to answer questions regarding the cognitive load. This
is executed four times. Each time a different process model is shown as depicted in
Table 4.1. All three modularization approaches follow this repeated procedure. Further,
each group receives the same process models and in the same order of representa-
tion. After the execution is carried out four times, the evaluation of the modularization
50
5.2 Study Execution
approach, to which the participant is assigned to, ensues. The modularization approach
is graphically and textually exemplified. With further steps, the usefulness is questioned
using three self designed questions. Then, questionnaires regarding the cognitive load,
perceived usefulness for understandability, and perceived ease of understanding are
utilized. After measuring the items, the design of the single modularization approach is
considered. This is applied, as the design effects the understandability and cognitive
load. Questions regarding the quality of the basic element, such as the link event in
horizontal modularization, are asked, next to the icon design.
In the next step, all three modularization approaches are presented. For each approach
a designation is placed above the process model. The participant has to decide which
approach and combined approaches are adequate for the use in practice. In the next
step, he has to provide an assessment regarding the cognitive complexity that would
come up by reading all three approaches represented in one model. Before the survey
ends, the participant has the opportunity to provide feedback. Then, the participants
were evinced the study leaders gratitude. Finally, the used scientific literature was
shown to reference the sources of information. The materials that are submitted to the
participants can be obtained from Appendix A.
Study:
After contacting possible participants, they can choose a time and date in Doodle.
The following procedure is given for each participant:
The study is executed in an office room in the Ulm University - the same for each
iteration. Hence, the conditions during data collection remain equivalent. Before the
study begins, the participant is welcomed in the staircase of the DBIS and then led into
the study room. Then the experiment is executed. At the beginning of the experiment,
an introduction is given. The user receives an overview about the necessary basics
regarding the experiment. This is comprised of the study procedure and risks. Further,
the user is referred to the fact that he can stop the experiment at any point in time.
Then, the participant has to sign a declaration of consent. The signature entitles him
to take part in the study. Afterwards, the participant has to fill out the components of a
demographic questionnaire that is presented in Section 4.4. Next, the user is introduced
to the topic. As the study scrutinizes modularization, the introduction focuses on this
51
5 Study Operation
Figure 5.2: Survey Design
Figure 5.3: Study Procedure
52
5.2 Study Execution
topic. Further, an explanation about modularization and subareas with respect to the
experiment is provided. Then an example with regard to the study materials is presented
to the participant. Based on this example, a first insight in participation is provided. The
eye tracker is adapted to the participant. This is necessary because feedback from the
eye such as pupil size and corneal reflections have to be measured. In addition, the
eye tracker is calibrated. Then, the eye tracking study is executed. It is split into three
units, these are: modularization approach one, two, and three. They are executed in a
balanced manner as Figure 5.4 demonstrates.
The construction of the survey is as follows:
Each unit has a trial. A trial includes one process model. This process model is
differentiated four times regarding the presented questions. The procedure is executed
three times blocked in one unit. It is blocked through the weighting of the representations
in a modularization approach. In the example of Figure 5.5, a process model (P) is given
Figure 5.4: Study Design
53
5 Study Operation
for each column. Every process model varies with respect to the content. Regarding
the row, defined variants are used. These variants are composited in a set of three
modularization approaches. Each modularization approach includes: a later changed
through self designed (1), a not changed (2), and a self designed (3) representation
form.
Figure 5.5: Trials
The gap between ’O’, ’V’, and ’H’ is given by a calibration. It is necessary because
participants make small head movements. They can lead to failures regarding the
measurements and the evaluated results. Hence, a good tracking of the eye is possible.
This progress is done twice. After completion of the eye tracking study, the participant
is taken aside to answer the questionnaires regarding the cognitive load, perceived
usefulness for understandability, and ease of understanding.
The materials that are given to the participant are demonstrated in Appendix B.
54
5.3 Data Validation
5.3 Data Validation
Data validation gives an overview about the period of time the data was collected.
Next, collected data is considered in terms of the independent variables. For each
independent variable the mean (M), standard deviation (SD), and number of participants
(N) is presented.
Survey: The data was collected from November the 19th to November the 26th.
95 persons participated in the survey. 48 were male, 45 female, and two participants
gave no information regarding their gender. The participants that indicated their age were
between 25 and 35, and younger. All participants were academics or trainees/students,
with an average year of apprenticeship of 17.34 years. The distribution in the examination
variants can be taken from Table 5.1.
Independent Variable Horizontal
Modularization
Vertical Modu-
larization
Orthogonal
Modularization
M SD N M SD N N SD N
Educational years 17.32 1.95 34 16.93 1.18 28 17.72 1.57 32
Modelling experience 0.80 0.40 35 0.571 0.50 28 0.59 0.49 32
BPMN knowledge 1.00 0.00 34 1.00 0.00 28 1.00 0.00 32
Prior BPMN knowledge 0.57 0.50 35 0.57 0.50 28 0.62 0.49 32
Design knowledge 0.80 0.40 35 0.57 0.50 28 0.53 0.49 32
Table 5.1: Independent Variables in the Survey
Study:
The execution of the study was carried out from November the 27th to December
the 5th.
In this period of time, 22 persons participated in the study. 21 were male and one was
female. After reducing the data set, as given in Section 6.1, only 19 participants remained.
All of them were male, and their age was between 25 and 35, and younger. Most
participants were academics or trainees/students. Only one participant has completed
his apprenticeship. The average year of apprenticeship was 19.79 years. Further results
of the independent variables can be taken from Table 5.2.
55
5 Study Operation
Independent Variable Study
M SD N
Educational years 19.79 2.44 19
Modelling experience 0.84 0.37 19
BPMN knowledge 0.68 0.47 19
Prior BPMN knowledge 0.42 0.50 19
ICL 2.97 1.17 19
GCL 2.57 0.98 19
ECL 3.85 1.01 19
PEU 4.43 0.98 19
PUU 4.81 0.73 19
Table 5.2: Independent Variables in the Study
56
6
Study Analysis and Interpretation
The evaluation of the data and the related discussion are presented in this chapter. Its
structure is given in the following. In the beginning, the criteria of data set reduction are
presented in Section 6.1. The hypotheses are tested in Section 6.2. Finally, a summary
and discussion is presented in Section 6.3.
Data is evaluated with SPSS. The ANOVA is applied for testing the control variables and
the hypotheses. Significant differences (
p <
.05) are visually marked by an ’
∗
’ and highly
significant differences (
p <
.01) by ’
∗∗
’. The results are represented with numbers with
up to two decimals after the comma. Remaining decimal numbers are cut off.
p
represents the probability that the
null hypothesis H
0 is reported. When a significant
difference with
p=.049
is obtained, the probability for the occurrence of the null
hypothesis is 4.9%. The probability for the
alternative hypothesis H
1 increases with a
lower
p
value. Further, the mean (M), standard deviation (SD), and number of participants
(N) is presented for hypotheses testing.
Furthermore, when the sphericity of data is not given (
p <
.05 in a Mauchly Test) the
degree of freedom has to be corrected. When
ε >
.75 the Huynh-Feldt-Correction is
selected, otherwise Greenhouse-Geisser.
The spelling for Modularization is abbreviated to M. in the table.
6.1 Data Set Reduction
When the following conditions are met, the data is excluded from evaluation.
57
6 Study Analysis and Interpretation
Survey:
•no data in terms of assessing process models is measured
In the survey, no data set reduction exists. Hence, the complete data set was considered
in the evaluation.
Study:
•it was not possible to track the eye
•divergence in environmental variables exists
Therefore, three data sets were removed in evaluation. Two times the data collection
with the eye tracker was not possible. One failure was caused by the glasses of a
participant. The pupil could not be captured. The second failure occurred, as it was
not possible to detect a coronal reflex of the pupil. The third data set reduction was
caused by environmental conditions. As the influence of light led to problems in data
collection, longer breaks between eye tracking occurred. As it could have an effect
on answering the statements, exclusion results. After the data set reduction, 19 data
sets were included in the evaluation. Next to the complete omission of data sets, single
values were removed for evaluation. This is caused by the values measured by the eye
tracker. As small head movements occurred while tracking, the data quality of single
tasks was sometimes too low to state the situation. Hence, when less then 50 % of the
data was measured for a process model, the data regarding the KPI was excluded in the
evaluation.
6.2 Hypotheses Testing
In this section, the hypotheses formalized in Section 4.3 are tested for significance.
Survey:
In the following, hypotheses in terms of cognitive load and understandability, the
items of RQ1, are tested.
Cognitive Load
Eight items were measured to receive insights into the cognitive load when reading
58
6.2 Hypotheses Testing
modularized business process models. These items can be differentiated through the
intrinsic, germane, and extraneous cognitive load. For each load, the four presented
process models are considered. Further, the mean of the four process models is
calculated. Finally, the load after acquiring knowledge based on the description of the
single modularization approach is outlined.
Q1.1: Intrinsic cognitive load
The results can be extracted from Table 6.1 and Table 6.2. Table 6.1 provides insights in
terms of the significance between all modularization approaches. Fuhrer, the means,
standard deviations, and number of participants are presented. Table 6.2 compares the
multiple comparisons among the single modularization approaches by the significances.
The ICL exhibits a significant difference (
F(2,92) =
4.10,
p=
.02) among the three
modularization approaches after merging the results of the four presented process
models. Significant differences (
p <
.05), obtained using the Bonferroni revealed that
vertical modularization increases the intrinsic cognitive load compared to horizontal
modularization (
p=
.04). Further, orthogonal modularization showed a significantly
lower intrinsic cognitive load than vertical modularization (p=.03).
ICL Horizontal M. Vertical M. Orthogonal M.
p M SD N M SD N M SD N
ICL P1 .13 1.82 0.82 35 2.19 0.89 28 1.79 0.81 32
ICL P2 .21 2.40 0.92 35 2.78 1.24 28 2.31 1.14 32
ICL P3 * .02 3.82 1.21 35 4.57 1.37 28 3.71 1.22 32
ICL P4 * .04 3.22 1.19 35 3.98 1.34 28 3.32 1.20 32
ICL P1-P4 * .02 2.82 0.80 35 3.37 0.95 28 2.78 0.90 32
ICL Described .29 3.70 1.22 35 4.00 1.29 28 3.50 1.18 32
Table 6.1: Intrinsic Cognitive Load in the Survey
ICL Horizontal M. Vertical M. Orthogonal M.
ICL P1-P4 Horizontal M. 1.00 .04 1.00
Vertical M. - 1.00 .03
Orthogonal M. - - 1.00
ICL Described Horizontal M. 1.00 1.00 1.00
Vertical M. - 1.00 .36
Orthogonal M. - - 1.00
Table 6.2: Multiple Comparisons on Intrinsic Cognitive Load in the Survey
59
6 Study Analysis and Interpretation
Furthermore, two significant differences are set out in Table 6.1 for the single process
models. One significant difference with respect to process model three (
F(2,92) =
3.96,
p=
.02), and another for process model four (
F(2,92) =
3.22,
p=
.04). Process model
three showed a significant higher intrinsic cognitive load for vertical modularization than
for orthogonal modularization (
p=
.03). In process model four, no significant difference
was measured for the intrinsic cognitive load among the modularization approaches (
p >
.05).
This leads to the conclusion that hypothesis H1.1 (a), with reference to the fact that
no significant differences exist, has to be rejected for the vertical modularization. In
addition, the assumption in hypothesis H1.1 (b) that the vertical modularization achieves
the lowest intrinsic cognitive load, does not agree with the statistical results. In summary,
the intrinsic cognitive load indicates:
V ertical > (Orthogonal =Horizontal)
. Vertical
modularization is significantly higher than the horizontal and orthogonal modularization.
Furthermore, no significant difference between orthogonal and horizontal modularization
is reported.
Q1.2: Germane Cognitive Load
No significant differences (
F(2,92) =
2.02,
p=
.23) can be reported for the four summa-
rized process models.
GCL Horizontal M. Vertical M. Orthogonal M.
p M SD N M SD N M SD N
GCL P1 .14 4.12 1.24 35 4.66 1.29 28 4.37 0.92 32
GCL P2 .46 4.30 1.35 35 4.66 1.29 28 4.34 0.98 32
GCL P3 .12 4.51 1.04 35 5.02 1.03 28 4.56 1.07 32
GCL P4 .19 4.39 1.20 35 4.88 1.01 28 4.66 0.95 32
GCL P1-P4 .23 4.34 1.06 35 4.81 1.00 28 4.48 0.71 32
GCL Described .48 4.59 1.04 35 4.86 1.02 28 4.83 0.95 32
Table 6.3: Germane Cognitive Load in the Survey
Further, no significant difference (
F(2,92) =
0.74,
p=
.48) was measured between the
modularization approaches for the GCL after their description. As presented in Table 6.3
and 6.4, no significant difference was measured on GCL. Neither through comparing all,
nor comparing the single modularization approaches a significant difference could be
60
6.2 Hypotheses Testing
GCL Horizontal M. Vertical M. Orthogonal M.
GCL P1-P4 Horizontal M. 1.00 .15 1.00
Vertical M. - 1.00 .53
Orthogonal M. - - 1.00
GCL Described Horizontal M. 1.00 .83 .98
Vertical M. - 1.00 1.00
Orthogonal M. - - 1.00
Table 6.4: Multiple Comparisons on Germane Cognitive Load in the Survey
observed. This leads to discarding the alternative hypothesis that significant differences
are given. Hence, the null hypothesis can be reported.
Q1.3: Extraneous Cognitive Load
Regarding the extraneous cognitive load, the mean of process models one through four
reports no significant difference (
F(2,92) =
1.50,
p=
.22) among the three modulariza-
tion approaches.
ECL Horizontal M. Vertical M. Orthogonal M.
p M SD N M SD N M SD N
ECL P1 .58 1.99 0.84 35 2.16 0.83 28 1.93 0.96 32
ECL P2 .39 2.35 1.32 35 2.60 1.43 28 2.81 1.35 32
ECL P3 .37 3.59 1.34 35 4.07 1.53 28 3.75 1.22 32
ECL P4 .16 2.85 1.23 35 3.48 1.48 28 3.22 1.22 32
ECL P1-P4 .22 2.70 0.89 35 3.08 0.85 28 2.93 0.91 32
ECL Described .07 2.97 1.02 35 3.55 1.34 28 2.94 1.13 32
Table 6.5: Extraneous Cognitive Load in the Survey
Furthermore, no significant difference (
F(2,92) =
2.62,
p=
.07) was measured for the
ECL of the modularization approaches, after the description was given to the participants.
Further, no significant difference was reported as presented in Table 6.4 and 6.5. This
leads to the rejection of the alternative hypothesis. Hence, no difference between the
modularization approaches exists in terms of the ECL.
Understandability
Understandability is relevant for a successful utilization of modularization. Therefore, two
content areas were evaluated: perceived usefulness for understandability, and perceived
61
6 Study Analysis and Interpretation
ECL Horizontal M. Vertical M. Orthogonal M.
ECL P1-P4 Horizontal M. 1.00 .27 .86
Vertical M. - 1.00 1.00
Orthogonal M. - - 1.00
ECL Described Horizontal M. 1.00 .14 1.00
Vertical M. - 1.00 .13
Orthogonal M. - - 1.00
Table 6.6: Multiple Comparisons on Extraneous Cognitive Load in the Survey
ease of understanding. The data of both areas was measured after the process model
evaluation and measuring the cognitive load in the modularization approach evaluation
(see Figure 5.2). Next, for each content area the individual items are considered. Further,
the mean of the items is presented for the individual content areas.
Q2.1: Perceived Usefulness for Understandability
The perceived usefulness for understandability is composed of four items. No item
reported a significant difference (
p >
.05). This is also given in the end result for
perceived usefulness for understandability (F(2,92) = 0.83, p=.43).
As no significant difference can be seen in Table 6.7 (for all modularization approaches)
and 6.8 (between the single approaches), the null hypothesis is reported.
PUU Horizontal M. Vertical M. Orthogonal M.
p M SD N M SD N M SD N
1Business process models repre-
sented in this way would be difficult
for users to understand.
.05 2.69 1.23 35 3.46 1.42 28 2.78 1.43 32
2I think this presentation approach
provides an effective solution to the
problem of representing business
process models.
.72 4.74 1.33 35 4.71 1.24 28 4.50 1.41 32
3Using this type of process mod-
els would make it more difficult to
communicate business processes to
end-users.
.24 2.83 1.09 35 3.39 1.49 28 3.03 1.40 32
4Overall, I found the business pro-
cess model in this experiment to be
useful.
.97 5.03 1.36 35 5.04 1.20 28 4.97 1.22 31
PUU
sum
codified: 1, 3 .43 4.56 1.02 35 4.22 1.11 28 4.40 1.00 32
Table 6.7: Perceived Usefulness for Understandability in the Survey
62
6.2 Hypotheses Testing
PUU Horizontal M. Vertical M. Orthogonal M.
PUU sum Horizontal M. 1.00 .60 1.00
Vertical M. - 1.00 1.00
Orthogonal M. - - 1.00
Table 6.8: Multiple Comparisons on PUU in the Survey
Q2.2: Perceived Ease of Understanding
As no significant difference (
F(2.92) =
0.43,
p=
.65) for perceived ease of understanding
occurs regarding the modularization approaches (see Table 6.9 and 6.10), the null
hypothesis can be reported.
PEU Horizontal M. Vertical M. Orthogonal M.
p M SD N M SD N M SD N
1Learning to use this way of mod-
elling business processes would be
easy for me.
.89 4.77 1.51 35 4.86 1.32 28 4.69 1.23 32
2I found the way the process is rep-
resented as unclear and difficult to
understand.
.31 2.46 1.22 35 2.96 1.50 28 2.59 1.29 32
3It would be easy for me to become
skilful at using this way of modelling
business processes.
.56 4.77 1.23 35 4.50 1.34 28 4.84 1.29 32
4Overall, I found this way of modelling
business processes difficult to use.
.70 2.83 1.17 35 3.07 1.35 28 3.03 1.25 31
PEU
sum
codified: 2, 4 .65 4.56 1.00 35 4.33 1.03 28 4.48 0.97 32
Table 6.9: Perceived Ease of Understanding in the Survey
PEU Horizontal M. Vertical M. Orthogonal M.
PEU sum Horizontal M. 1.00 1.00 1.00
Vertical M. - 1.00 1.00
Orthogonal M. - - 1.00
Table 6.10: Multiple Comparisons on PEU in the Survey
Study:
While the survey discusses the issue of RQ1, the study deals with RQ2 and RQ3.
Time units were adjusted. Times such as 01:12:23.01 were changed to 01:12:23.10, as
SPSS was not able to capture the original time.
Performance Success
The performance success is composed of the number of a correct answer and the
63
6 Study Analysis and Interpretation
time required for eye tracking. Further, the results received from the eye tracker such
as average fixation are utilized. A comparison with respect to different modularization
approaches is considered.
First, the correct answer is considered. It was measured using points. Each right answer
led to a point. Totally, 36 points were possible, twelve points for each modularization
approach. As Table 6.11 shows, the scores obtained show only little difference. This is
also supported by the results, analyzed through repeated measures. No measurable
difference (F(2,34) = 0.85, p=.43) occurred.
Second, the time measured while executing the task is of interest. A measurable
difference (
F(2,34) =
3.93,
p=
.02) with an observed power of 66.8% between the
modularization approaches exists when using repeated measures. However, further
evaluation as presented in Table 6.12 reported no difference between the conditions
(p > .05).
Performance
Success
Horizontal M. Vertical M. Orthogonal M.
p M SD N M SD N M SD N
Total score .43 10.33 1.57 18 10.94 1.16 18 10.61 1.33 18
Total time* .02 03:37.39 00:55.14 18 03:18.04 01:05.20 18 04:00.62 01:26.01 18
Table 6.11: Performance Success in the Study
Performance Success Horizontal M. Vertical M. Orthogonal M.
Total Score Horizontal M. 1.00 .39 1.00
Vertical M. - 1.00 1.00
Orthogonal M. - - 1.00
Total Time Horizontal M. 1.00 .27 .93
Vertical M. - 1.00 1.00
Orthogonal M. - - 1.00
Table 6.12: Multiple Comparisons on Performance Success in the Study
Finally, the key performance indicators (KPI) that were observed through the eye tracker
are analyzed via repeated measures. The evaluated results are presented in Table 6.13.
As most items of the KPI indicate no appreciable difference (
p >
.05), the number of
fixations varies (
F(1.41,23.97) =
5.39,
p=
.01) in terms of the modularization approach.
Next, it was verified which modularization approaches attribute to the difference. This
64
6.2 Hypotheses Testing
ensues through a One-Way ANOVA. With reference to the Bonferroni post-hoc analysis,
as presented in Table 6.14, no difference (
p >
.05) between the single modularization
approaches occurs.
KPI Horizontal M. Vertical M. Orthogonal M.
p M SD N M SD N M SD N
Dwell Time .49 92.96 7.28 18 94.49 2.91 18 91.82 9.87 18
Revisit .50 0.80 1.67 18 0.44 0.48 18 0.93 1.52 18
Revisits .89 0.35 0.32 18 0.33 0.32 18 0.37 0.40 18
Average Fixation .87 219.70 22.44 18 218.57 29.03 18 221.09 27.09 18
First Fixation .22 288.06 86.70 18 315.04 73.57 18 322.05 65.57 18
Fixation Count* .01 71.92 28.56 18 57.13 15.66 18 60.20 15.60 18
Table 6.13: Key Performance Indicator (KPI) in the Study
KPI Horizontal M. Vertical M. Orthogonal M.
Dwell Time Horizontal M. 1.00 1.00 1.00
Vertical M. - 1.00 .71
Orthogonal M. - - 1.00
Revisit Horizontal M. 1.00 1.00 1.00
Vertical M. - 1.00 .96
Orthogonal M. - - 1.00
Revisits Horizontal M. 1.00 1.00 1.00
Vertical M. - 1.00 1.00
Orthogonal M. - - 1.00
Average Fixation Horizontal M. 1.00 1.00 1.00
Vertical M. - 1.00 1.00
Orthogonal M. - - 1.00
First Fixation Horizontal M. 1.00 .78 .61
Vertical M. - 1.00 1.00
Orthogonal M. - - 1.00
Fixation Count Horizontal M. 1.00 .10 .28
Vertical M. - 1.00 1.00
Orthogonal M. - - 1.00
Table 6.14: Multiple Comparisons on KPI in the Study
As no meaningful difference in terms of the performance success occurs, the modular-
ization approaches don’t provide varieties. Therefore, the alternative hypothesis has to
be rejected. The null hypothesis is given.
Design
The design refers to the research question RQ3. Therefore, the different presentations in
65
6 Study Analysis and Interpretation
the individual modularization approaches are compared. The comparison is comprised
of measuring variables like the time, scores, and the KPI. Hence, it becomes obvious
which design of the single modularization approaches gain better performance success.
First, Q4.1 is considered. It points out how the representation influences horizontal
modularization.
Q4.1: Horizontal Modularization
Horizontal modularization was divided into:
link events
,
message events
, and
self de-
signed.
Horizontal M. Message Events Link Events Self Designed
p M SD N M SD N M SD N
Scores .91 3.47 0.77 19 3.42 0.76 19 3.37 0.89 19
Time .12 01:11.61 00:30.43 19 01:22.73 00:37.08 19 01:23.18 00:28:84 19
Dwell Time .06 94.53 3.59 16 95.87 2.12 16 93.97 3.78 16
Revisit .61 0.40 0.42 16 0.32 0.39 16 0.34 0.47 16
Revisits .54 0.32 0.29 16 0.26 0.29 16 0.26 0.33 16
Average Fixa-
tion
.42 222.27 25.33 16 225.69 22.00 16 228.07 24.79 16
First Fixation .30 312.49 93.44 16 292.35 85.10 16 325.62 75.56 16
Fixation Count .24 62.48 28.83 16 20.25 32.41 16 70.13 29.14 16
Table 6.15: Horizontal Modularization in the Study
Horizontal M. Message
Events
Link Events Self Designed
Scores Message
Events
1.00 1.00 1.00
Link Events - 1.00 1.00
Self Designed - - 1.00
Time Message
Events
1.00 .88 .82
Link Events - 1.00 1.00
Self Designed - - 1.00
Dwell Time Message
Events
1.00 1.00 .54
Link Events - 1.00 .63
Self Designed - - 1.00
Revisit Message
Events
1.00 .92 1.00
Link Events - 1.00 1.00
Self Designed - - 1.00
Revisits Message
Events
1.00 1.00 1.00
Link Events - 1.00 1.00
66
6.2 Hypotheses Testing
Self Designed - - 1.00
Average Fixation Message
Events
1.00 1.00 1.00
Link Events - 1.00 1.00
Self Designed - - 1.00
First Fixation Message
Events
1.00 .88 1.00
Link Events - 1.00 .92
Self Designed - - 1.00
Fixation Count Message
Events
1.00 .49 1.00
Link Events - 1.00 1.00
Self Designed - - 1.00
Table 6.16: Multiple Comparisons on Horizontal Modularization in the Study
As presented in Table 6.15 and 6.16, no difference occurred (
p >
.05). Hence, the null
hypothesis is reported.
Q4.2: Vertical Modularization
The effects of representation are considered in terms of vertical modularization. Vertical
modularization is composed of:
expanded subprocesses
,
collapsed subprocesses
, and
self designed.
With reference to Table 6.17 and 6.18, no recognizable difference (
p >
.05) between the vertical modularization representations can be reported. Hence, the
alternative hypothesis has to be discarded.
Vertical M. Expanded Subpro-
cess
Collapsed Subpro-
cess
Self Designed
p M SD N M SD N M SD N
Scores .95 3.67 0.59 18 3.61 0.77 18 3.67 0.48 18
Time .94 01:04.79 00:34.10 18 01:06.76 00:25.22 18 01:06.85 00:20:69 18
Dwell Time .10 95.07 2.61 17 95.64 2.16 17 92.61 6.95 17
Revisit .14 0.33 0.38 17 0.48 0.59 17 0.50 0.60 17
Revisits .34 0.30 0.34 17 0.28 0.32 17 0.36 0.39 17
Average Fixa-
tion
.29 217.47 33.48 17 213.17 31.60 17 220.84 29.19 17
First Fixation .51 219.36 73.38 17 318.13 139.53 17 321.71 87.84 17
Fixation Count .93 56.26 22.36 17 58.18 20.09 17 56.92 18.55 17
Table 6.17: Vertical Modularization in the Study
Q4.3: Orthogonal Modularization
The different representation approaches regarding to orthogonal modularization are
67
6 Study Analysis and Interpretation
Vertical M. Expanded Sub-
process
Collapsed Sub-
process
Self Designed
Scores Expanded Sub-
process
1.00 1.00 1.00
Collapsed Sub-
process
- 1.00 1.00
Self Designed - - 1.00
Time Expanded Sub-
process
1.00 1.00 1.00
Collapsed Sub-
process
- 1.00 1.00
Self Designed - - 1.00
Dwell Time Expanded Sub-
process
1.00 1.00 .38
Collapsed Sub-
process
- 1.00 .17
Self Designed - - 1.00
Revisit Expanded Sub-
process
1.00 1.00 1.00
Collapsed Sub-
process
- 1.00 1.00
Self Designed - - 1.00
Revisits Expanded Sub-
process
1.00 1.00 1.00
Collapsed Sub-
process
- 1.00 1.00
Self Designed - - 1.00
Average Fixation Expanded Sub-
process
1.00 1.00 1.00
Collapsed Sub-
process
- 1.00 1.00
Self Designed - - 1.00
First Fixation Expanded Sub-
process
1.00 1.00 1.00
Collapsed Sub-
process
- 1.00 1.00
Self Designed - - 1.00
Fixation Count Expanded Sub-
process
1.00 1.00 1.00
Collapsed Sub-
process
- 1.00 1.00
Self Designed - - 1.00
Table 6.18: Multiple Comparisons on Vertical Modularization in the Study
68
6.2 Hypotheses Testing
evaluated. As shown in Table 6.19, two differences with regard to repeated measures
occur. Difference one (
F(1.65,29.76) =
3.59,
p=
.04) refers to the dwell time (time
spent on fixating the graphic) and was corrected through the Huynh-Feldt correction.
The second one showed a difference (
F(2,36) =
3.62,
p=
.03) regarding the average
fixation. By evaluating the two measured differences through a One-Way ANOVA, no
difference between the groups (
p >
.05) was measurable. This is presented in Table
6.20.
Hence, the null hypothesis occurs.
Orthogonal M. Error Events Aspect-Oriented Self Designed
p M SD N M SD N M SD N
Scores .64 3.42 1.07 19 3.68 0.67 19 3.53 0.69 19
Time .41 01:07.06 00:15.95 19 01:16.15 00:31.44 19 01:11.27 00:25:05 19
Dwell Time* .04 90.31 11.81 19 92.01 9.67 19 92.72 7.75 19
Revisit .42 0.88 1.47 19 0.81 1.41 19 0.95 1.65 19
Revisits .57 0.35 0.41 19 0.32 0.40 19 0.38 0.43 19
Average Fixa-
tion*
.03 216.42 29.30 19 219.94 29.44 19 227.75 26.63 19
First Fixation .65 309.44 77.95 19 325.10 76.50 19 320.31 82.54 19
Fixation Count .35 56.13 14.57 19 64.03 26.38 19 61.57 19.38 19
Table 6.19: Orthogonal Modularization in the Study
Orthogonal M. Error Events Aspect-
Oriented
Self Designed
Scores Error Events 1.00 1.00 1.00
Aspect-
Oriented
- 1.00 1.00
Self Designed - - 1.00
Time Error Events 1.00 .81 1.00
Aspect-
Oriented
- 1.00 1.00
Self Designed - - 1.00
Dwell Time Error Events 1.00 1.00 1.00
Aspect-
Oriented
- 1.00 1.00
Self Designed - - 1.00
Revisit Error Events 1.00 1.00 1.00
Aspect-
Oriented
- 1.00 1.00
Self Designed - - 1.00
Revisits Error Events 1.00 1.00 1.00
Aspect-
Oriented
- 1.00 1.00
69
6 Study Analysis and Interpretation
Self Designed - - 1.00
Average Fixation Error Events 1.00 1.00 .67
Aspect-
Oriented
- 1.00 1.00
Self Designed - - 1.00
First Fixation Error Events 1.00 1.00 1.00
Aspect-
Oriented
- 1.00 1.00
Self Designed - - 1.00
Fixation Count Error Events 1.00 .73 1.00
Aspect-
Oriented
- 1.00 1.00
Self Designed - - 1.00
Table 6.20: Multiple Comparisons on Orthogonal Modularization in the Study
Based on these results, the null hypothesis is given for each question regarding the
design. Some differences were measured in terms of the representations of single
modularization approaches. By comparing these representations through a One-Way
ANOVA, with reference to the Bonferroni post-hoc analysis, no differences were mea-
sured. Hence, the various modularization approaches in business process modeling
have no impact with respect to cognitive complexity of process model readers.
6.3 Summary and Discussion
In this section the results of Section 6.2 are discussed.
Cognitive Load
Intrinsic Cognitive Load:
A significant difference between vertical and the other modularization approaches is
shown. This could be caused by the different representations of the three modularization
approaches. The representation refers to activities like icons.
Intrinsic cognitive load has the factors element interactivity and prior knowledge of the
learner. Hence, these items have to be considered. No significant differences regarding
prior knowledge exists. This leads to the assumption that the significant difference is
caused by element interactivity.
70
6.3 Summary and Discussion
The significant difference of process model three and four could be caused by the growth
in complexity and the need of a subprocess. Hence, a further activity, the subprocess
activity, is required. It includes an icon in form of a plus. This icon can be misunderstood
by novices. In contexts like applets, a plus represents the adding of further content.
Hence, the participant could expect that he can add further content to the activity instead
of linking to an underlying process model.
Germane Cognitive Load:
In the survey, no significant difference was measured. This can be caused by the
motivation and the enjoyment while preparing the task. However, it can’t be assumed
that a difference between the groups in the survey exists, as only slight deviations are
caused. Next, the understandability doesn’t show a significant difference. The result in
GCL can be caused by the little differences, such as the quality and the realization of the
single modularization approaches.
Extraneous Cognitive Load:
No significant difference in ECL is shown. As the participants had to read the process
models and haven’t had to perform a task, it wasn’t measured if the participant under-
stood the process model. Further, it was not guaranteed that they dealt with the process
model. Participants, that did not deal with the process model, required less capacities of
the working memory. As no significant difference between the groups exists regarding
the prior knowledge and educational background, the same behavior was expected.
Therefore, it is assumed that no difference between the groups is given. Hence, the
difference in ECL can be caused by the quality or difficulty in representation of the
individual modularization approaches. As these are nearly identical, participants had to
apply the same capacities in information processing.
Understandability
Perceived Usefulness for Understandability:
In the survey, no significant difference was obtained regarding the perceived usefulness
for understandability. This leads to two assumptions:
•
the representation of the three modularization approaches is analogical. Hence,
no differences in understandability were possible.
71
6 Study Analysis and Interpretation
•
the representation of the three modularization approaches isn’t analogical, but the
quality in representation and understandability is the same.
With reference to the significant difference in intrinsic cognitive load that is caused by
element interactivity the representation of the three approaches is not analogical. This is
also supported by the different symbols caused by the modularization approaches.
Thus, assumption two that ’the representation of the three modularization approaches
isn’t analogical, but the quality in representation and understandability is the same’ turns
out to be correct. This assumption is supported by the results that were measured
regarding the extraneous and germane cognitive load. With respect to these observed
measurement variables, no significant difference occurred. The design and represen-
tation quality, such as the information processing, seems to be equivalent. Hence, the
perceived usefulness for understandability is also equivalent.
Perceived Ease of Understanding:
No significant difference regarding the perceived ease of understanding was reported.
This leads to the assumption that the process models provide an almost equivalent
understandability and reproducibility. This assumption is supported by the results
regarding the germane and extraneous cognitive load. Therefore, the representation
exhibit the same quality and leads to the need of analogical mental resources of the
individuals.
The element interactivity measured through the intrinsic cognitive load leads to two
possibilities:
•
Element interactivity exists without implication on perceived ease of understanding
•
Element interactivity exists and has implications on perceived ease of understand-
ing
If the first assumption occurs, the intrinsic cognitive load has no implications on the
perceived ease of understanding. Hence, elements such as the selected icons would not
have an effect. However, element interactivity is necessary for good understanding of the
content. Therefore, the second assumption seems to be possible. Element interactivity
points out that good results can be achieved when learners don’t have to link other
72
6.3 Summary and Discussion
elements [
84
]. After linking the elements and receiving insights into the design and topic,
the approaches can be understood by the participants without a measurable difference.
Performance Success
In the study, the performance success was measured without perceived differences.
This can be caused by different factors.
•Quality of materials
•Difficulty of materials
•Complexity of materials
•Little number of participants
The participants received in mean 31.26 points from 36 points. Hence, 86.83% of the
answers were correct. Therefore, the difficulty of materials could cause the effect. The
assessed contents and statements could be too simple for receiving meaningful results.
Further, the quality of materials could lead to the result. On the one hand, the chosen
examples could be self-descriptive and an understanding of the process model is
redundant. This assumption can be discarded as questions with respect to the usage
of modules and the need of activities in modularization were questioned. On the other
hand, the representation of the modularization approaches can cause effects regarding
the performance. In organizations, the representation of the process models varies in
terms of the presented process models. For example, in collapsed modularization, the
underlying process isn’t represented on the same page (except when it is opened) as
the overlying process. Hence, the actual difficulties aren’t necessarily eliminated.
Next, the complexity of the materials is quite low as this thesis focuses on the individual
elements’ modularization and the understandability of the modularization approaches.
The experimental design was presented in a way to receive meaningful results regarding
to the focus.
Finally, the small amount of participants could cause the effect. As only 19 persons
participated in the study, significant reporting was impossible. Further, the observed
power was quite often low. Hence, the study should be executed with a higher number
of participants to receive adequate results.
73
6 Study Analysis and Interpretation
Design
No difference caused by the design occurred with respect to the single modularization
approaches.
Possible effects on the design can be caused by:
•Process model complexity
•Items in the process model
In horizontal modularization, no measurable differences resulted. The design has only
slight variations, as merely changes regarding events/activities and sequenceflows exist.
The single process model designs don’t seem to provide higher complexity. Therefore,
the performance success can almost be equivalent.
In vertical modularization, no difference regarding to the design is provided. This is
in contrast to the results of [
47
]. [
47
] showed that collapsed subprocesses provide a
decrease in comprehension. As the focus of this research question was on the perfor-
mance success and not understandability, a different result can be obtained. However,
the understandability has effects on the performance success. Further, the fixation
count and the average fixation would differ through differences in understandability.
Therefore, the variation in representation and the process model complexity could cause
the difference. Next, the complexity of the individual representations varies. As one
representation approach is given in a sequence, the ensuing approach is composed
of further activities, next to subdivision. This could cause further cognitive load. As
already discussed, the complexity of the presented process model and the number of
modularization representations could lead to the result.
No difference in terms of the design in orthogonal modularization is reported. In or-
thogonal modularization, the activities and representations vary. In the aspect-oriented
approach, activities were chosen to reference the join point of the modularized section
through point cuts. With respect to exceptions, an interaction through events is given.
Hence, the similarity doesn’t lead to the missing difference. The similarity received in the
results of the eye tracker confirmed this assumption.
74
6.3 Summary and Discussion
Summary
In summary, the significant difference in intrinsic cognitive load seems to be caused
by misunderstanding the plus icon utilized in vertical modularization. As no further
significant difference occurred, a lot of intersection points exist, these are for example
based on:
•same quality/ realization of modularization approaches
•same capacities in information processing
•element interactivity
However, different factors can lead to effects regarding the results. The study was
executed with a low number of participants. As a result, the observed power was
quite low. Hence, the results of replicated studies may lead to different assumptions.
Further, a higher amount of participants would have effects on the validity. First of all
validity, in particular regarding the external validity, as the results would provide better
generalization. Therefore, more research is necessary to provide significant results
and clear statements. This can be realized through experimental reproduction and/or
additional studies. How further research could be realized is given in Section 8. Scientific
research, utilized in this thesis, is presented in the following section.
75
7
Related Work
The related work is comprised of two aspects. Aspect one is in terms of the differentiation
of the modularization approaches, next to their differences, while aspect two deals with
the selected materials.
In [
46
], vertical and horizontal modularization are mentioned regarding to modularization.
While many papers only refer to the horizontal and vertical modularization, [
1
] is cited
when all modularization approaches of business process models are object of interest.
The modularization approaches are comprised of horizontal, vertical, and orthogonal
modularization. Furthermore, [
1
] provides first insights in terms of the usability and
thus understandability of the single modularization approaches regarding to experts.
The authors point out, that orthogonal modularization provides a lower ease of use and
usefulness compared to the other two approaches. Further, [
1
] outlines the options of
the design regarding the single approaches.
The differences in a single modularization approach are object of investigation. With
respect to the vertical modularization, discerns exist. Expanded subprocesses provide
better understandability as opposed to collapsed subprocesses. This finding is due to
[
47
]. They point out that fully-flattened process models influence the understandability
positively. Thus they are better understandable than collapsed and expanded subpro-
cesses. Further, the authors argue that expanded subprocesses should preferably be
used in terms of understandability.
However, the design of the single modularization approach was considered. For vertical
modularization the design of expanded and collapsed subprocesses was researched.
The design of expanded subprocesses is based on the presentation in [
33
]. This poster
presents an open subprocess task in a collaboration diagram. Inside this task a self-
77
7 Related Work
contained process model is shown. For collapsed subprocess two different designs were
considered. For the survey the representation of [
1
] is selected and the illustration from
[
47
] is utilized for the study. While [
1
] uses dashed lines that open from the subprocess
task to the underlying process, [
47
] illustrated the subprocess outside the pool (indicating
the parent process) with a border around the subprocess.
The design of horizontal modularization contains message and link events. Both repre-
sentations were illustrated in [1].
In orthogonal modularization exceptions and aspect-oriented paradigm were considered.
First, activities to visualize exceptions had to be selected. Therefore, [
60
] referred to
different events and provided insights into the possibilities in representation. Hence, [
60
]
showed that exceptions can be handled in an event-subprocess or outside a subprocess.
For the aspect-oriented paradigm the illustration of [
1
] was utilized. Additional research
was applied to acquire knowledge regarding the procedure. In [
61
] single aspects and
the fundamentals of aspect-oriented paradigms were considered. Further they provide
insights into a possible implementation and tooling of aspect-oriented paradigms for
BPMN.
Next, the presentation medium makes no difference with respect to the understandability
[
47
]. Hence, computer or paper based materials can be selected for empirical research.
Materials of the eye tracking studies regarding business process models exist. Such
materials are given in diverse pre-studies of the Institute of Databases and Information
Systems (DBIS). In these pre-studies the main constructs of different business process
modeling notations are differentiated. This representation provides prospects of success
related to the previous findings, as significant differences are shown in the results. Mea-
suring variables that lead to the findings are, for example, the KPI and the performance
success.
Next to the materials that were prepared for measuring, the variables for comparison
were of interest. [
74
] set up cognitive weightings for BPM elements. The weightings
refer to the efforts that are necessary for understandability. An OR provides a lower
comprehension than an XOR. Furthermore, [
74
] pointed out that the understandability
is lower when the number of activities is low. The cognitive load theory supports this
78
result. Further, [
75
] addresses the fact that understandability and cognitive load theory
are related, as the cognitive load theory deals e.g. with understanding.
The presented related work focuses on the basic foundations on which this thesis is
based. Next, further research possibilities and a conclusion are provided.
79
8
Future Work and Conclusion
This chapter provides further insights regarding this thesis, such as additional research
possibilities. Section 8.1 presents a conclusion. Further, future work is discussed in
Section 8.2.
8.1 Conclusion
This thesis investigates the effects of modularization in business process models on the
cognitive complexity of humans. For this purpose, the developed research goal was to
obtain insights in terms of the cognitive load when individuals read different modularized
business process models. Based on further research, materials for two studies were
designed. Hence, a survey with 95 participants and a study with 18 participants were
executed.
In the survey, information regarding the cognitive complexity of the participants were
measured. The focus lied on answering questionnaires with respect to the cognitive
load, the PUU, and PEU after reading business process models. Each group (horizontal,
vertical, and orthogonal) received the same process models. Process models differing
in the representation were presented to the respective groups. These differences were
realized through changes in terms of the process model modularization approaches.
Hence, a participant of the group ’orthogonal modularization’ only received orthogonal
modularized process models. Then the participant had to answer questions according
to the CLT. Next, the participant had to answer questionnaires after looking at all three
modularization approaches. The results acquired using the survey showed a difference
81
8 Future Work and Conclusion
regarding the intrinsic cognitive load. Following, the vertical modularization leads to a
significantly higher ICL caused by element interactivity.
In addition to the survey, an eye tracking study was executed. Each participant had
to execute each approach. The focus was on the performance success. Therefore,
the participant had to assess statements with true or false while executing the eye
tracking study. Next to the accuracy of assessment, the time for executing the tasks
was measured. A further point of interest was the design variations in a modularization
approach. Therefore, the performance success of an individual approach was compared
with the others. No difference was reported in the results.
In conclusion, the results based on the survey and the eye tracking study reported no
difference. Hence, it is regardless which modularization approach is chosen. Only the
vertical modularization leads to higher intrinsic cognitive load. The understandability
reported, next to the germane and extraneous cognitive load, no significant difference.
However, more data has to be collected through analogical/further research. Based
on further results, guidelines for modularizing business process models and training
documents could be designed. This would lead to the benefit that a well understandable
modularization approach for process readers can be selected in process modeling.
Consequently, an optimal distribution of the cognitive capacities of readers is possible.
8.2 Future Work
Modularization is a comprehensive topic. The amount of further research possibilities is
enormous. Through the preliminary empirical results of this thesis it has been shown,
that it makes no difference which modularization approach is chosen. However, further
research is necessary to confirm these results, through replications and/or additional
studies.
Next, higher complexity of process models could be used for data evaluation. As
the survey results demonstrated, a small number of activities (little complexity) could
contribute to the fact that only small differences are given. Hence, more complexity
provides the opportunity for receiving significant differences between the modularization
82
8.2 Future Work
approaches. This assumption is supported by the results of the intrinsic cognitive load,
where process models with a higher number of elements led to a significantly higher
intrinsic cognitive load when reading vertical modularized business process models. In
addition to the results achieved, the usage of a higher complexity is also based on the
fact that business process models applied in practice have higher complexity than the
process models shown in the studies.
Further, the combination of modularization approaches can be considered, as the various
approaches are applicable in a business process model. The survey pointed out that
61.1% of the participants would prefer a combination of the horizontal and vertical
modularization, 28.4% the combination of horizontal and orthogonal modularization,
and 9.5% the combination of vertical and orthogonal modularization. No participant
preferred the combination of all three modularization approaches. A study could be, inter
alia, realized using an eye tracking study where insights, regarding the performance
success will become obvious, for example. Further, the participant could identify the
modularized parts by drawing into a process model. Hence, intuitive understanding
about modularization could be captured. Further, it becomes obvious if the participant is
aware of the term modularization and can transfer it into practice.
As the number of participants that would prefer a different design regarding the single
modularization approaches were less than 50%, a change of elements in modularization
is not of need.
However, the different representations of a single modularization approach can be mea-
sured. With respect to vertical modularization, [
47
] compared collapsed and expanded
subprocesses, such as the usage of non-modularized process models. They received
insights in terms of the understandability and therefore the usefulness when using the
modularization approach. Such findings could also be recorded for the horizontal and or-
thogonal modularization. With respect to the horizontal approach, the utilization of signal,
message, and link events can be compared. The orthogonal modularization includes
the comparable elements: exceptions and aspect-oriented modularization. In terms of
exception handling, various design possibilities can be compared. For aspect-oriented
modularization, different possibilities in representation could be evaluated, as no clear
83
8 Future Work and Conclusion
modeling representation exists. This would lead to the benefit that a standardization of
representation could be established for BPMN. Hence, literature and documentations
would have a unique and understandable design for representation.
Finally, further notations of business process modeling could be used for subsequent
studies. Furthermore, the different business process modeling notations could be
compared with respect to their modularization.
84
Bibliography
[1]
La Rosa, M., Wohed, P., Mendling, J., ter Hofstede, A.H.M., Reijers, H.A.,
van der Aalst, W.M.P.: Managing Process Model Complexity Via Abstract Syntax
Modifications. IEEE Transactions on Industrial Informatics 7(2011) 614–629
[2]
van der Aalst, W.M.P., ter Hofstede, A.H.M., Weske, M.: Business Process
Management:
A Survey. In van der Aalst, W.M.P., Weske, M., eds.: Business
Process Management, Springer (2003) 1–12
[3]
Figl, K., Laue, R.: Cognitive Complexity in Business Process Modeling. In:
International Conference on Advanced Information Systems Engineering, Springer
(2011) 452–466
[4]
Cardoso, J., Mendling, J., Neumann, G., Reijers, H.A.: A Discourse on Complexity
of Process Models. In Eder, J., Dustdar, S., eds.: Business Process Management
Workshops, Springer (2006) 117–128
[5]
Mendling, J., Reijers, H.A., Cardoso, J.: What Makes Process Models
Understandable?
In Alonso, G., Dadam, P., Rosemann, M., eds.: Business
Process Management, Springer (2007) 48–63
[6]
Reijers, H.A., Mendling, J.: A Study Into the Factors That Influence the
Understandability
of Business Process Models. IEEE Transactions on Systems,
Man, and Cybernetics - Part A: Systems and Humans 41 (2011) 449–462
[7]
Bobrik, R., Reichert, M., Bauer, T.: Parameterizable Views for Process
Visualization
. In Alonso, G., Dadam, P., Rosemann, M., eds.: BPM 2007. LNCS.
Volume 4714., Springer (2007) 88–95
[8]
Duden: Modularisierung. (Website) Online available at
https://www.duden.de/
rechtschreibung/Modularisierung; last accessed on 14. January 2019.
[9]
Hambach, S.: Modularisierung von Bildungsangeboten: Problembeschreibung
und Lösungsansatz, Gesellschaft für Informatik eV (2003) 183–189
85
Bibliography
[10]
Reijers, H.A., Mendling, J., Dijkman, R.M.: Human and automatic modularizations
of process models to enhance their comprehension. Information Systems
36
(2011) 881–897
[11]
Reijers, H., Mendling, J.: Modularity in Process Models: Review and Effects.
In Dumas, M., Reichert, M., Shan, M.C., eds.: Business Process Management.
Volume 5240., Springer (2008) 20–35
[12]
Aguilar-Saven, R.S.: Business process modelling: Review and framework.
International Journal of Production Economics 90 (2004) 129–149
[13]
Davenport, T.H.: Process Innovation: Reengineering Work Through Information
Technology. Harvard Business Press (1993)
[14]
Palleduhn, D.U., Neuendorf, H.: Geschäftsprozessmanagement und Integrierte
Informationsverarbeitung. Walter de Gruyter (2013)
[15]
van der Aalst, W.M.P., La Rosa, M., Santoro, F.M.: Business Process Management.
Business & Information Systems Engineering 58 (2016) 1–6
[16]
Dumas, M., La Rosa, M., Mendling, J., Reijers, H.A.: Fundamentals of Business
Process Management. Volume 2. Springer (2018)
[17]
van der Aalst, W.M.: Business Process Management: A personal view. Business
Process Management Journal 10 (2004)
[18]
Vanwersch, R.J.B., Vanderfeesten, I., Rietzschel, E., Reijers, H.A.: Improving
Business Processes: Does Anybody have an Idea? In Motahari-Nezhad, H.R.,
Recker, J., Weidlich, M., eds.: Business Process Management, Springer (2015)
3–18
[19]
van der Aalst, W., et al.: Process Mining Manifesto. In: Business Process
Management Workshops. BPM 2011, Springer (2012) 169–194
86
Bibliography
[20]
Mutschler, B., Builler, J., Reichert, M.: Towards an Evaluation Framework
for
Business
Process Integration and Management. In: Proceedings 2nd Int’l
Workshop
on Interoperability Research for Networked Enterprises Applications
and Software (
Workshop
held in conjunction with the 9th IEEE Intl. Conf. of
Enterprise Computing (EDOC’05)), IEEE Digital Library (2005)
[21]
Dumas, M., Van der Aalst, W.M., Ter Hofstede, A.H.: Process-Aware Information
Systems: Bridging People and Software Through Process Technology. John Wiley
& Sons (2005)
[22]
Emig, C., Weisser, J., Abeck, S.: Development of SOA-Based Software Systems
- an Evolutionary Programming Approach. In: Advanced Int’l Conference on
Telecommunications and Int’l Conference on Internet and Web Applications and
Services, IEEE (2006) 182
[23]
Rabhi, F.A., Yu, H., Dabous, F.T., Wu, S.Y.: A service-oriented architecture for
financial business processes. Information Systems and e-Business Management
5(2007) 185–200
[24]
Pant, K., Juric, M.B.: Business Process Driven SOA using BPMN and BPEL: From
Business Process Modeling to Orchestration and Service Oriented Architecture.
Packt Publishing Ltd (2008)
[25]
van der Aalst, W.M.P. In: Process-Aware Information Systems: Lessons to Be
Learned from Process Mining. Springer (2009) 1–26
[26]
Papazoglou, M.P.: Service-Oriented Computing: Concepts, Characteristics
and Directions. In: Web Information Systems Engineering, 2003. WISE 2003.
Proceedings of the Fourth International Conference on. (2003) 3–12
[27]
Inc., L.S.: What is Business Process Modeling Notation? (Website) Online
available at
https://www.lucidchart.com/pages/bpmn?a=0
; last accessed
on 13. January 2019.
[28]
Kolb, J., Reichert, M., Weber, B.: Using Concurrent Task Trees for Stakeholder-
centered Modeling and Visualization of Business Processes. In: S-BPM ONE
2012, Springer (2012) 237–251
87
Bibliography
[29]
Weske, M.: Business Process Management: Concepts, Languages, Architectures.
Volume 2. Springer (2012)
[30]
Indulska, M., Green, P., Recker, J., Rosemann, M.: Business Process
Modeling:
Perceived Benefits. In Laender, A.H.F., Castano, S., Dayal, U., Casati, F.,
de Oliveira, J.P.M., eds.: Conceptual Modeling - ER 2009, Springer (2009)
458–471
[31]
Group, O.M.: BPMN Specification - Business Process Model and Notation.
(
Website
) Online available at
http://www.bpmn.org
; last accessed on 13.
November 2018.
[32]
AG, G.S..C.: BIC Cloud – Prozessmodellierung BPMN 2.0. (Website) Online
available at
https://www.gbtec.de/template/elemente/310/BPMN_2.0_
Konventionen.pdf; last accessed on 03. October 2018.
[33]
Signavio, BPM Offensive Berlin, H.P.I.C.I., zu Berlin, H.U.: BPMN 2.0
- Business Process Model and Notation. (Website) Online available at
https://www.signavio.com/wp-content/uploads/2012/10/poster-
preview-bpmn-en.png; last accessed on 13. September 2018.
[34]
van der Aalst, W.M.P.: Formalization and Verification of Event-driven Process
Chains. Information and Software Technology 41 (1999) 639–650
[35]
Schultheiss, L.A., Heiliger, E.: Techniques of Flow-Charting. Clinic on Library
Applications of Data Processing (1963) 62–78
[36]
Allweyer, T.: BPMN 2.0: Introduction to the Standard for Business Process
Modeling. BoD–Books on Demand (2016)
[37]
Dehnert, J., Zimmermann, A.: On the Suitability of Correctness Criteria for
Business Process Models. In van der Aalst, W.M.P., Benatallah, B., Casati, F.,
Curbera, F., eds.: Business Process Management, Springer (2005) 386–391
[38]
Mendling, J.: Metrics for Process Models: Empirical Foundations of Verification,
Error Prediction, and Guidelines for Correctness. Volume 6. Springer (2008)
88
Bibliography
[39]
Laune, R., Mendling, J.: Structuredness and its significance for correctness of
process models. Information Systems and e-Business Management
8
(2010)
287–307
[40]
Mendling, J., Reijers, H.A., van der Aalst, W.M.: Seven Process Modeling
Guidelines (7PMG). Information and Software Technology 52 (2010) 127–136
[41]
Jose, A., Tollenaere, M.: Modular and platform methods for product family design:
literature analysis. Journal of Intelligent Manufacturing 16 (2005) 371–390
[42]
Klein, R.: Modellgestütztes Service Systems Engineering: Theorie und Technik
einer systemischen Entwicklung von Dienstleistungen. Springer (2007)
[43]
Cappelli, C., Santoro, F.M., Cesar Sampaio do Prado Leite, J., Batista, T.,
Lusia
Medeiros, A., Romeiro, C.S.: Reflections on the modularity of business process
models: The case for introducing the aspect-oriented paradigm. Business Process
Management Journal 16 (2010) 662–687
[44]
Jack Hu, S., Reid, J., Anderson, P.: Modularity and Scalable Manufacturing.
(Website) Online available at
https://www.ameslab.gov/sites/default/
files/Modular%20manufacturing_Hu.pdf
; last accessed on 19. January
2019.
[45]
Zugal, S., Pinggera, J., Weber, B., Mendling, J., Reijers, H.A.: Assessing the
Impact of Hierarchy on Model Understandability–A Cognitive Perspective. In
Kienzle, J., ed.: Models in Software Engineering, Springer (2011) 123–133
[46]
Genon, N., Heymans, P., Amyot, D.: Analysing the Cognitive Effectiveness of
the BPMN 2.0 Visual Notation. In Malloy, B., Staab, S., van den Brand, M., eds.:
International Conference on Software Language Engineering, Springer (2011)
377–396
[47]
Turetken, O., Rompen, T., Vanderfeesten, I., Dikici, A., van Moll, J.: The Effect
of Modularity Representation and Presentation Medium on the Understandability
of Business Process Models in BPMN. In La Rosa, M., Loos, P., Pastor, O., eds.:
Business Process Management, Springer (2016) 289–307
89
Bibliography
[48]
Bhat, J.M., Deshmukh, N.: Methods for Modeling Flexibility in Business Process.
Workshop on Business Process Modeling, Design and Support (BPMDS05),
Procedings of CAiSE05 Workshops (2005) 1–8
[49]
Turetken, O., Dikici, A., Vanderfeesten, I., Rompen, T., Demirors, O.: The
Influence
of Using Collapsed Sub-processes and Groups on the
Understandability
of
Business
Process Models. Business & Information Systems Engineering: The
International Journal of WIRTSCHAFTSINFORMATIK (2019)
[50]
De Masellis, R., Di Francescomarino, C., Ghidini, C., Maggi, F.M.: Declarative
Process Models: Different Ways to Be Hierarchical. In Sheng, Q.Z., Stroulia, E.,
Tata, S., Bhiri, S., eds.: Service-Oriented Computing, Springer (2016) 104–119
[51]
Radgui, M., Saidi, R., Mouline, S.: A Pattern for the Decomposition of Business
Processes. Special Issue of International Journal of Computer Applications
(0975-8887)
on Software Engiineering, Databases and Expert Systems-SEDEXS
(2012)
[52]
Decker, G., Dijman, R., Dumas, M., Garía-Bañuelos, L.: The Business Process
Modeling Notation. In Hofstede, A.H.M., Aalst, W.M.P., Adams, M., Russell, N.,
eds.: Modern Business Process Automation: YAWL and its Support Environment,
Springer (2010) 347–368
[53]
Dijkman, R.M., Dumas, M., Ouyang, C.: Semantics and Analysis of Business
Process Models in BPMN. Information and Software Technology
50
(2008)
1281–1294
[54]
Dijkman, R.M., Dumas, M., Ouyang, C.: Formal Semantics and Analysis of BPMN
Process Models using Petri Nets. Queensland University of Technology, Tech.
Rep (2007)
[55]
White, S.A.: BPMN Modeling and Reference Guide: Understanding and Using
BPMN. Future Strategies Inc. (2008)
[56]
Davis, R., Brabander, E.: ARIS Design Platform: Getting Started with BPM.
Springer (2007)
90
Bibliography
[57]
Davis, R.: Business Process Modelling with ARIS: A Practical Guide. Springer
(2001)
[58]
Johannsen, F., Leist, S.: Wand and Weber’s Decomposition Model in the Context
of Business Process Modeling. Business & Information Systems Engineering
4
(2012) 271–286
[59]
Capelli, C., Leite, J.C., Batista, T., Silva, L.: An Aspect-Oriented Approach to
Business Process Modeling. In: Proceedings of the 15th Workshop on Early
Aspects, ACM (2009) 7–12
[60]
Welke, R.: Session 5: Business Process Modeling (BPMN) Events. (Website)
Online aviailable at
http://www3.cis.gsu.edu/dtruex/courses/CIS4120/
Lectures-pdf/Day5-Events-Sp15.pdf
; last accessed on 08. November
2018.
[61]
Witteborg, H., Charfi, A., Collell, D.C., Mezini, M.: Weaving Aspects and Business
Processes through Model Transformation. In Villari M., Zimmermann W., L.K.K.,
ed.: Service-Oriented and Cloud Computing, Springer (2014) 47–61
[62]
Bieri, J.: Cognitive complexity-simplicity and predictive behavior. The Journal of
Abnormal and Social Psychology 51 (1955) 263–268
[63]
Streufert, S.: Zum Stand der kognitiven Komplexitätstheorie. In Mandel, H., Huber,
G., eds.: Kognitive Komplexität. Bedeutung, Weiterentwicklung, Anwendung,
Verlag für Psychologie, Hogrefe (1978) 85–96
[64]
Scott, W.A.: Cognitive Complexity and Cognitive Flexibility. Sociometry
25
(1962)
405–414
[65]
Seiler, T.B.: Überlegungen zu einer kognitionstheoretischen Fundierung des
Konstrukts der kognitiven Komplexität. In Mandel, H., Huber, G., eds.: Kognitive
Komplexität. Bedeutung, Weiterentwicklung, Anwendung, Verlag für Psychologie,
Hogrefe (1978) 111–141
[66]
Huber, H.D.: Der von Bildern umstelle Horizont. Systematischer Rahmenbau zu
Anna Oppermann. In Sökeland, J., Vorkoeper, U., eds.: Über Medialität - Anlass
Anna Oppermann. (2009) 97–105
91
Bibliography
[67]
Atkinson, R.C., Shiffrin, R.M.: Human Memory: A Proposed System and its Control
Processes. In: Psychology of Learning and Motivation. Volume 2. Academic
Press (1968) 89–195
[68]
Schneider, W., Shiffrin, R.M.: Controlled and Automatic Human Information
Processing: I. Detection, Search, and Attention. Psychological Review
84
(1977)
1–66
[69]
Lieberman, D.A.: Human Learning and Memory. Cambridge University Press
(2011)
[70]
Miller, G.A.: The Magical Number Seven, Plus or Minus Two: Some Limits on Our
Capacity for Processing Information. Psychological Review 63 (1956) 81–97
[71]
Cowan, N.: The Magical Number 4 in Short-Term Memory: A Reconsideration of
Mental Storage Capacity. Behavioral and Brain Sciences 24 (2000) 87–185
[72] Sweller, J.: Instructional Design. In: Australian Educational Review. (1999)
[73]
Spada, H.: Einige Anmerkungen zur Theorie der kognitiven Komplexität von
Harvey, Schroder und Mitarbeitern und zum Modell der kognitiven Sozialstation von
Oerter, Dreher und Dreher. In Mandel, H., Huber, G., eds.: Kognitive Komplexität.
Bedeutung, Weiterentwicklung, Anwendung, Verlag für Psychologie, Hogrefe
(1978) 157–168
[74]
Gruhn, V., Laue, R.: Adopting the Cognitive Complexity Measure for Business
Process Models. Volume 1. (2006) 236–241
[75]
Sweller, J.: How the Human Cognitive System Deals with Complexity. (2006)
13–25
[76]
Chandler, P., Sweller, J.: Cognitive Load Theory and the Format of Instruction.
Cognition and Instruction 8(1991) 293–332
[77]
Sweller, J., Van Merrienboer, J.J., Paas, F.G.: Cognitive Architecture and
Instructional Design. Educational Psychology Review 10 (1998) 251–296
[78]
Kalyuga, S.: Cognitive Load Theory: How Many Types of Load Does It Really
Need? Educational Psychology Review 23 (2011) 1–19
92
Bibliography
[79] Baddeley, A.: Working memory. Science 255 (1992) 556–559
[80]
Niegemann, H.M., Domagk, S., Hessel, S., Hein, A., Hupfer, M., Zobel, A.:
Kompendium multimediales Lernen. Springer (2008)
[81]
Sweller, J.: Element Interactivity and Intrinsic, Extraneous, and Germane Cognitive
Load. Educational Psychology Review 22 (2010) 123–138
[82]
Klepsch, M., Schmitz, F., Seufert, T.: Development and Validation of Two
Instruments
Measuring Intrinsic, Extraneous, and Germane Cognitive Load. Fron-
tiers in Psychology 8(2017)
[83]
Paas, F., Renkl, A., Sweller, J.: Cognitive Load Theory and Instructional Design:
Recent Developments. Educational Psychologist 38 (2003) 1–4
[84]
Sweller, J.: Visualisation and Instructional Design. In: Proceedings of the
International
Workshop on Dynamic Visualizations and Learning. Volume 18.
(2002) 1501–1510
[85]
Paas, F., Renkl, A., Sweller, J.: Cognitive Load Theory: Instructional
Implications
of the Interaction between Information Structures and Cognitve Architecture.
Instructional Science 32 (2004) 1–8
[86]
Seufert, T.: Lernen mit multiplen Repräsentationen: Gestaltungs- und
Verarbeitungsstrategien
. Lernchance Computer. Strategien für das Lernen mit
digitalien Medienverbünden 52 (2009) 45–66
[87]
Paas, F., Tuovinen, J.E., Tabbers, H., Van Gerven, P.W.: Cognitive Load
Measurement
as a Means to Advance Cognitive Load Theory. Educational
Psychologist 38 (2003) 63–71
[88]
Sweller, J., Ayres, P.L., Kalyuga, S., Chandler, P.: The Expertise Reversal Effect.
Educational Psychologist 38 (2003) 23–31
[89]
Marieb, E.N., Hoen, K.: Human Anatomy & Physiology. Pearson Education (2007)
[90]
Marieb, E.N.: Human Anatomy & Physiology. Volume 5. Benjamin-Cummings
Pub Co (2001)
93
Bibliography
[91]
Wierda, S.M., van Rijn, H., Taatgen, N., Martens, S.: Pupil dilation deconvolution
reveals the dynamics of attention at high temporal resolution. Proceedings of the
National Academy of Science 109 (2012) 8456–8460
[92]
Becker-Carus, C., Wendt, M.: Allgemeine Psychologie: Eine Einführung. Springer
(2017)
[93]
Amadieu, F., Van Gog, T., Paas, F., Tricot, A., Mariné, C.: Effects of Prior
Knowledge and Concept-Map Structure on Disorientation, Cognitive Load, and
Learning. Learning and Instruction 19 (2009) 376–386
[94]
Eckstein, M.K., Guerra-Carrillo, B., Miller, S.A.T., Bunge, S.A.: Beyond eye gaze:
What else can eyetracking reveal about cognition and cognitive development?
Development Cognitive Neuroscience 25 (2017) 69–91
[95]
Borys, M., Plechawska-Wójcik, M.: Eye-tracking metrics in perception and visual
attention research. EJMT 3(2017) 11–23
[96]
Rayner, K.: Eye Movements in Reading and Information Processing: 20 Years of
Research. Psychological Bulletin 124 (1998) 372–422
[97]
Van Gog, T., Scheiter, K.: Eye Tracking as a Tool to Study and Enhance Multimedia
Learning. Learning and Instruction 20 (2010) 95–99
[98]
Salvucci, D.D., Goldberg, J.H.: Identifying Fixations and Saccades in Eye-Tracking
Protocols. In: Proceedings of the 2000 Symposium on Eye Tracking Research &
Applications, ACM (2000) 71–78
[99]
Zimoch, M., Pryss, R., Schobel, J., Reichert, M.: Eye Tracking Experiments on
Process Model Comprehension: Lessons Learned. In: 18th Int’l Conference on
Business Process Modeling, Development, and Support (BPMDS 2017), Springer
(2017) 153–168
[100]
Gruhn, V., Laue, R.: Complexity Metrics for Business Process Models. In
Abramowicz, W., Mayr, H.C., eds.: 9th International Conference on Business
Information Systems (BIS 2006). Volume 85., Lecture Notes in Informatics (2006)
1–12
94
Bibliography
[101]
Lang, S.: Empirische Forschungsmethoden. (Website) Online
available
at
https://www.uni-trier.de/fileadmin/fb1/prof/PAD/SP2/
Allgemein/Lang_Skript_komplett.pdf
; last accessed on 07. November
2018.
[102]
Duchowski, A.T.: Eye Tracking Methodology: Theory and Practice. Volume 328.
Springer (2007)
[103]
Duden: Hypothese. (Website) Online available at
https://www.duden.de/
rechtschreibung/Hypothese; last accessed on 13. January 2019.
[104]
Bortz, Jürgen und Döring, N.: Forschungsmethoden und Evaluation für Human-
und Sozialwissenschaftler: Limitierte Sonderausgabe. Volume 4. Springer (2007)
[105]
Rettenbach, R., Christ, C.: Psychotherapie-Prüfung: Kompaktkurs zur
Vorbereitung
auf die Approbationsprüfung nach dem Psychotherapeutengesetz
mit Kommentar zum IMPP-Gegenstandskatalog. Volume 2004. Schattauer Verlag
(2016)
[106]
Nielsen, J.: Thinking Aloud: The # 1 Usability Tool. (Web-
site)
Online available
at
https://www.nngroup.com/articles/thinking-
aloud-the-1-usability-tool/; last accessed on 14. January 2019.
[107]
Patel, M.X., Doku, V., Tennakoon, L.: Challenges in recruitment of research
participants. Advances in Psychiatric Treatment 9(2003) 229–238
[108]
Rasch, B.: Methodenlehre Vorlesung 5. (Website) Online available
at
http://commonweb.unifr.ch/artsdean/pub/gestens/f/as/files/
4660/33580_113129.pdf; last accessed on 13. January 2019.
[109]
Aronson, E., Akert, R.M., Wilson, T.D.: Sozialpsychologie. Pearson Deutschland
GmbH (2010)
[110]
Anderson, A.: 3 Ways to Describe Populations and Samples in Business Statistics.
(Website) Online available at
https://www.dummies.com/education/
math/business-statistics/3-ways-to-describe-populations-
and-samples-in-business-statistics/
; last accessed on 14. January
2019.
95
Bibliography
[111]
Signavio: BPM ACADEMIC INITIATIVE. (Website) Online available at
https:
//academic.signavio.com/p/login; last accessed on 13. January 2019.
[112]
Inc., A.: Keynote. (Website) Online available at
https://www.apple.com/de/
keynote; last accessed on 13. January 2019.
[113]
Klepsch, M., Seufert, T.: Subjective Differentiated Measurement of Cognitive
Load, Paper presented at 5th International Cognitive Load Theory Conference,
Tallahassee (USA) (2012)
[114]
Turetken, O.: Process Models & Questionnaire. (Website)
Online available
at
http://is.ieis.tue.nl/staff/oturetken/index.php/process-
models/; last accessed on 14. November 2018.
[115]
Zimoch, M.: Studie - Verständlichkeit und Zugänglichkeit von Prozessmodellen.
(unpublished questionnaire)
[116]
Google: Professionelle Formulare leicht gemacht. (Website) Online available
at
https://www.google.de/intl/de/forms/about/
; last accessed on 13.
January 2019.
[117]
Dresden, T.U.: WIEDERHOLUNGSMESSUNG. (Website) Online
available
at
http://versuch.file2.wcms.tu-dresden.de/w/index.php/
Wiederholungsmessung?fbclid=IwAR2b9xcIgqcb7b8mA3P02DkRvezT_
7RHNryiYYfHpRHPWNQd3CcaerqLgmE; last accessed on 31. October 2018.
[118]
Dresden, T.U.: RANDOMISIERUNG. (Website) Online
available at
http://versuch.file2.wcms.tu-dresden.de/w/
index.php/Randomisierung?fbclid=IwAR1_gXWwnz6MVte-
19EXxVHTvVAbdjkg8JPZddGIA03YBtMP-qYU4Nx-zc0
; last accessed
on 31. October 2018.
[119]
Ludwig-Mayerhofer, W.: Willkommen zur Vorlesung Empirische
Methoden
I. (Website) Online available at
https://www.uni-siegen.de/
phil/sozialwissenschaften/soziologie/mitarbeiter/ludwig-
mayerhofer/methoden/methoden_downloads/methoden_i_4_
2013.pdf; last accessed on 07. November 2018.
96
Bibliography
[120]
Dresden, T.U.: BLOCKBILDUNG. (Website) Online
available at
http://versuch.file2.wcms.tu-dresden.de/
w/index.php/Blockbildung?fbclid=IwAR3qYe2gaCB0Y-
Dp9xsYJtPWvPtvQRmIXakqdbuXLOxk9J8ZIs1rUJWRY-M
; last accessed on
31. October 2018.
[121]
OpenCourseWare, M.: 9.63 Laboratory in Visual Cognition. (Website) Online
available at
https://ocw.mit.edu/courses/brain-and-cognitive-
sciences/9-63-laboratory-in-visual-cognition-fall-2009/
lecture-notes/MIT9_63F09_lec02.pdf
; last accessed on 11. December
2018.
[122]
Loy, T.: Gtütekriterien - Objektivität Reliabilität Validität. (Website) Online available
at
https://statistik-und-beratung.de/2016/03/2552
; last accessed
on 13. January 2019.
[123]
Press, C.U.: objectivity. (Website) Online available at
https:
//dictionary-cambridge-org.cdn.ampproject.org/v/s/
dictionary.cambridge.org/de/amp/englisch/objectivity?amp_
js_v=a2&_gsa=1&usqp=mq331AQECAFYAQ%3D%3D#referrer=
https%3A%2F%2Fwww.google.com&_tf=Von%20%251%24s&share=
https%3A%2F%2Fdictionary.cambridge.org%2Fde%2Fworterbuch%
2Fenglisch%2Fobjectivity; last accessed on 18. December 2018.
[124]
ERservices: Chapter 5 Research Design. (Website) Online available
at
https://courses.lumenlearning.com/suny-hccc-research-
methods/chapter/chapter-5-research-design/?fbclid=
IwAR0eNcjx66kPYSp8jqOuzfEgO6sFS2EEypyHzgd7w9keTXcW79FnN2VVSbk
;
last accessed on 11. November 2018.
[125]
Sirois, S., Brisson, J.: Pupillometry. Wiley Interdisciplinary Reviews: Cognitive
Science 5(2014) 679–692
[126]
Doodle: Meetings möglich machen. (Website) Online available at
https://
www.doodle.com/de/; last accessed on 15. January 2019.
97
A
Survey
This appendix contains the materials presented to the participants of the survey. The
participants were randomly assigned to one of the three groups. In the following,
the content of the questionnaires of the groups horizontal, vertical, and orthogonal
modularization are displayed.
99
A Survey
Section 1 - Introduction
Figure A.1: Introduction
100
Section 2 - Demographic Questionnaire
Figure A.2: Section 2-1 of Demographic Questionnaire
101
A Survey
Figure A.3: Section 2-2 of Demographic Questionnaire
102
Figure A.4: Section 2-3 of Demographic Questionnaire
103
A Survey
Figure A.5: Section 2-4 of Demographic Questionnaire
104
Figure A.6: Section 2-5 of Demographic Questionnaire
105
A Survey
Section 3 - Process Models in the Modularization Approach
Figure A.7: Section 3-1 Questions
106
Figure A.8: Section 3-2 Questions
107
A Survey
Figure A.9: Section 3-3 Questions
Images in Horizontal Modularization (Iteration One up to Four)
Figure A.10: Image of First Iteration in Horizontal Modularization
108
Figure A.11: Image of Second Iteration in Horizontal Modularization
Figure A.12: Image of Third Iteration in Horizontal Modularization
Figure A.13: Image of Fourth Iteration in Horizontal Modularization
109
A Survey
Images in Vertical Modularization (Iteration One up to Four)
Figure A.14: Image of First Iteration in Vertical Modularization
Figure A.15: Image of Second Iteration in Vertical Modularization
Figure A.16: Image of Third Iteration in Vertical Modularization
110
Figure A.17: Image of Fourth Iteration in Vertical Modularization
Images in Orthogonal Modularization (Iteration One up to Four)
Figure A.18: Image of First Iteration in Orthogonal Modularization
111
A Survey
Figure A.19: Image of Second Iteration in Orthogonal Modularization
Figure A.20: Image of Third Iteration in Orthogonal Modularization
Figure A.21: Image of Fourth Iteration in Orthogonal Modularization
112
Section 4 - Modularization Approach
Figure A.22: Questions in Section 4-1
113
A Survey
Figure A.23: Questions in Section 4-2
114
Figure A.24: Questions in Section 4-3
115
A Survey
Figure A.25: Questions in Section 4-4
116
Figure A.26: Questions in Section 4-5
117
A Survey
Figure A.27: Questions in Section 4-6
118
Figure A.28: Questions in Section 4-7
119
A Survey
Image and Description in Horizontal Modularization
Figure A.29: Image and Description in Horizontal Modularization
120
Figure A.30: Suggested Icons in Horizontal Modularization
121
A Survey
Image and Description in Vertical Modularization
Figure A.31: Image and Description in Vertical Modularization
122
Figure A.32: Suggested Icons in Vertical Modularization
123
A Survey
Image and Description in Orthogonal Modularization
Figure A.33: Image and Description in Orthogonal Modularization
124
Figure A.34: Suggested Icons in Orthogonal Modularization
125
A Survey
Section 5 - Modularization Approaches
Figure A.35: Section 5-1
126
Figure A.36: Section 5-2
127
A Survey
Figure A.37: Section 5-3
128
Figure A.38: Section 5-4
129
A Survey
Figure A.39: Section 5-5
130
Section 6 - Final Input
Figure A.40: Feedback and Codeword
131
A Survey
Section 7- Literature
Figure A.41: Literature Presented
132
B
Study
The study is composed of different contents. As presented in Subsection 5.2, division
Study, an introduction is shown at the beginning of the study. The elements are given
in B.1. Further, images (P1-P9) with a specific order were represented in a gradual
visualization. Each image consists of a graphic and a statement as they are part of the
study. The elements of the images are presented in B.2 to B.10. Then, the questionnaires
are presented in B.11. Finally, the declaration of consent is shown in B.12.
B.1 Introduction
The introduction is divided into two sections. The subject-matter of the first section is the
content of the document submitted. The second subject-matter focuses on the content
of the oral report.
Content of the document submitted:
Studie – Verständlichkeit von Prozessmodellen
Bitte lesen Sie dieses Formular sorgfältig durch. Fragen Sie nach, wenn Sie etwas nicht
verstehen oder mehr Informationen benötigen.
Ziel der Studie:
In dieser Studie sollen Sie die Aussagen zu neun Prozessmodelle mittels Falsch/Wahr
beurteilen.
Die Studie wird unter der Leitung von Julia Baß (Masterstudentin, Universität Ulm)
durchgeführt und verfolgt rein wissenschaftliche Ziele.
133
B Study
Ablauf:
Die Studie findet für jeden Studienteilnehmer an genau einem Termin statt. Für die
Durchführung der Studie sind ca. 20 - 30 Minuten Zeit einzuplanen. Zu Beginn der
Studie erhalten Sie insgesamt zwei Papierbogen. Der erste Papierbogen enthält die
Studienbeschreibung, als auch einen demographischen Fragebogen. Der zweite Pa-
pierbogen enthält Fragen zum mentalen Aufwand während der Durchführung, sowie zur
Verständlichkeit der Prozessmodelle.
Beantworten und füllen Sie zuallererst den ersten Papierbogen aus. Im Anschluss
beginnt die Datenerhebung am Eye-Tracker. Zu Beginn erfolgt eine Kalibirerung. An-
schließend beurteilen Sie die neun Prozessmodelle der Reihe nach am Eye-Tracker.
Nach jeweils drei Prozessmodellen (12 Fragen) erfolgt erneut eine Kalibrierung. Nehmen
Sie sich zur Beantwortung der Aussagen Zeit. Sollten Sie auf eine Aussage keine Antwort
wissen, so klicken Sie die den Buchstaben der für Sie wahrscheinlicher ist. Nach der
Beurteilung der neun Prozessmodelle, füllen Sie bitte den Fragebogen zum mentalen
Aufwand und Verständlichkeit aus. Anschließend erhalten Sie Schokolade
Freiwilligkeit:
Sie nehmen an diesem Forschungsprojekt freiwillig teil. Ihr Einverständnis können
Sie jederzeit und ohne Angabe von Gründen widerrufen. Alle bis dahin studienbed-
ingt erhobenen Daten werden dann gelöscht. Dieser eventuelle Widerruf hat keine
Auswirkungen auf Ihre Person.
Risiken:
Es sind keine Risiken mit Ihrer Teilnahme an der Studie verbunden.
Erreichbarkeit des Studienleiters:
Sollten während des Verlaufes der Studie Fragen auftauchen, so können Sie jederzeit
Herrn Michael Zimoch unter der Telefonnummer erreichen: 0731 / 50 24 126.
Versicherung:
Während der Teilnahme an der Studie genießen Sie Versicherungsschutz. Die Univer-
sität Ulm und dessen an der Studie mitwirkende Mitarbeiter sind haftpflichtversichert für
den Fall, dass Sie durch deren Verschulden einen Schaden erleiden. Gleichzeitig weisen
wir darauf hin, dass Sie für die direkten Wege zum und vom Versuchsort nicht unfallver-
134
B.1 Introduction
sichert sind. Einen Schaden, der Ihrer Meinung nach auf diese Studie zurückzuführen
ist, melden Sie bitte unverzüglich dem Studienleiter.
Schweigepflicht und Datenschutz:
Alle Personen, welche Sie im Rahmen dieses Projektes betreuen, unterliegen der
Schweigepflicht und sind auf das Datengeheimnis verpflichtet. Die studienbezogenen
Untersuchungsergebnisse sollen in anonymisierter Form in wissenschaftlichen Veröf-
fentlichungen verwendet werden. Soweit es zur Kontrolle der korrekten Datenerhebung
erforderlich ist, dürfen autorisierte Personen (z.B.: des Auftraggebers, der Universität)
Einsicht in die studienrelevanten Teile der erhobenen Daten nehmen.
Content of the oral report:
This subdivision includes the notes realized for the introduction:
Modularisierung:
•Unterteilung eines großen Prozessmodells in kleinere Module
•Module haben ein Start und ein Ende
•Module sind unabhängig managebar
•Ein Modul interagiert an Schnittstellen mit anderen Modulen
Aufbau der Studie:
•Fragebogen, Einverständniserklärung
•Probelauf mit einem Image
•
Danach zur Studie: -> neun Prozessmodelle, zu jedem vier Aussagen -> Wahr/-
Falsch
•Nach 12 Aussagen (3 Prozessmodelle) kommt immer eine Kalibrierung
•Ende: Fragebogen zu Kognitive Belastung und Verständlichkeit
135
B Study
B.2 Image 1 (P1)
Image 1 consists of a sequence regarding handling a computer console, including the
steps for making changes requiring the entry of a process.
Aspect-Oriented
Figure B.1: Aspect-Ortiented in Orthogonal Modularization of P1
Exception Events
Figure B.2: Exception Events in Orthogonal Modularization of P1
136
B.2 Image 1 (P1)
Self Designed (Orthogonal)
Figure B.3: Self Designed in Orthogonal Modularization of P1
Collapsed Subprocesses
Figure B.4: Collapsed Subprocesses in Vertical Modularization of P1
Expanded Subprocesses
Figure B.5: Expanded Subprocesses in Vertical Modularization of P1
137
B Study
Self Designed (Vertical)
Figure B.6: Self Designed in Vertical Modularization of P1
Link Events
Figure B.7: Link Event in Horizontal Modularization of P1
138
B.2 Image 1 (P1)
Message Events
Figure B.8: Message Event in Horizontal Modularization of P1
Self Designed (Horizontal)
Figure B.9: Self Designed in Horizontal Modularization of P1
139
B Study
Set of Statements
Statement Horizontal Modular-
ization
Vertical Modulariza-
tion
Orthogonal Modu-
larization
ME LE SDH ES CS SDV EE AO SDO
Änderungen müssen durchgeführt werden true true true true true true true true true
Es sind keine Module in der Abbildung gegeben false false false false false false false false false
Module sind in der Abbildung gegeben true true true true true true true true true
Änderungen müssen nicht durchgeführt werden false false false false false false false false false
Table B.1: Statements of P1
B.3 Image 2 (P2)
Image 2 addresses an online purchase. It starts with filling a shopping cart and ends
with receiving a confirmation.
Aspect-Oriented
Figure B.10: Aspect-Ortiented in Orthogonal Modularization of P2
Exception Events
Figure B.11: Exception Events in Orthogonal Modularization of P2
140
B.3 Image 2 (P2)
Self Designed (Orthogonal)
Figure B.12: Self Designed in Orthogonal Modularization of P2
Collapsed Subprocesses
Figure B.13: Collapsed Subprocesses in Vertical Modularization of P2
Expanded Subprocesses
Figure B.14: Expanded Subprocesses in Vertical Modularization of P2
141
B Study
Self Designed (Vertical)
Figure B.15: Self Designed in Vertical Modularization of P2
Link Events
Figure B.16: Link Event in Horizontal Modularization of P2
142
B.3 Image 2 (P2)
Message Events
Figure B.17: Message Event in Horizontal Modularization of P2
Self Designed (Horizontal)
Figure B.18: Self Designed in Horizontal Modularization of P2
143
B Study
Set of Statements
Statement Horizontal Modular-
ization
Vertical Modulariza-
tion
Orthogonal Modu-
larization
ME LE SDH ES CS SDV EE AO SDO
Die Bestätigung muss erhalten werden true true true true true true true true true
Die Aktivität "Warenkorb befüllen" befindet sich in dem Modul,
welches das zweite aufgezeigte Modul aufruft
true true true true true true true true true
Bankverbindungsdaten müssen nicht angegeben werden false false false false false false false false false
Der Warenkorb kann, muss aber nicht befüllt werden false false false false false false false false false
Table B.2: Statements of P2
B.4 Image 3 (P3)
Image 3 comprises the topic baking pizza with a secret sauce.
Aspect-Oriented
Figure B.19: Aspect-Ortiented in Orthogonal Modularization of P3
144
B.4 Image 3 (P3)
Exception Events
Figure B.20: Exception Events in Orthogonal Modularization of P3
Self Designed (Orthogonal)
Figure B.21: Self Designed in Orthogonal Modularization of P3
Collapsed Subprocesses
Figure B.22: Collapsed Subprocesses in Vertical Modularization of P3
145
B Study
Expanded Subprocesses
Figure B.23: Expanded Subprocesses in Vertical Modularization of P3
Self Designed (Vertical)
Figure B.24: Self Designed in Vertical Modularization of P3
Link Events
Figure B.25: Link Event in Horizontal Modularization of P3
146
B.4 Image 3 (P3)
Message Events
Figure B.26: Message Event in Horizontal Modularization of P3
Self Designed (Horizontal)
Figure B.27: Self Designed in Horizontal Modularization of P3
147
B Study
Set of Statements
Statement Horizontal Modular-
ization
Vertical Modulariza-
tion
Orthogonal Modu-
larization
ME LE SDH ES CS SDV EE AO SDO
Der Belag muss immer hinzugefügt werden true true true true true true true true true
Die Zutaten müssen immer herangezogen werden true true true true true true true true true
Der Belag muss nicht hinzugefügt werden false false false false false false false false false
Module sind in der Abbildung gegeben true true true true true true true true true
Table B.3: Statements of P3
B.5 Image 4 (P4)
Image 4 focuses on refueling and the settlement of the chargeable amount.
Aspect-Oriented
Figure B.28: Aspect-Ortiented in Orthogonal Modularization of P4
Exception Events
Figure B.29: Exception Events in Orthogonal Modularization of P4
148
B.5 Image 4 (P4)
Self Designed (Orthogonal)
Figure B.30: Self Designed in Orthogonal Modularization of P4
Collapsed Subprocesses
Figure B.31: Collapsed Subprocesses in Vertical Modularization of P4
Expanded Subprocesses
Figure B.32: Expanded Subprocesses in Vertical Modularization of P4
149
B Study
Self Designed (Vertical)
Figure B.33: Self Designed in Vertical Modularization of P4
Link Events
Figure B.34: Link Event in Horizontal Modularization of P4
150
B.5 Image 4 (P4)
Message Events
Figure B.35: Message Event in Horizontal Modularization of P4
Self Designed (Horizontal)
Figure B.36: Self Designed in Horizontal Modularization of P4
151
B Study
Set of Statements
Statement Horizontal Modular-
ization
Vertical Modulariza-
tion
Orthogonal Modu-
larization
ME LE SDH ES CS SDV EE AO SDO
Die Geheimzahl muss nicht eingegeben werden, jedoch muss
die Buchung erfolgen
false false false false false false false false false
Direkt nachdem der Tankvorgang durchgeführt wurde, beginnt
ein neues Modul
false false false false false false false false false
Direkt nachdem der Tankvorgang durchgeführt wurde, beginnt
ein neues Modul
true true true true true true true true true
Die Aktivität "Tankstelle verlassen" muss ausgeführt werden true true true true true true true true true
Table B.4: Statements of P4
B.6 Image 5 (P5)
Image 5 takes up the topic of a supplier of goods. It begins with the acceptance of
the order. After an evaluation in terms of the availability of the goods, two options are
possible. If the goods are available they are sent and the process ends. Otherwise, the
process ends after running through the prepared solution.
Aspect-Oriented
Figure B.37: Aspect-Ortiented in Orthogonal Modularization of P5
152
B.6 Image 5 (P5)
Exception Events
Figure B.38: Exception Events in Orthogonal Modularization of P5
Self Designed (Orthogonal)
Figure B.39: Self Designed in Orthogonal Modularization of P5
153
B Study
Collapsed Subprocesses
Figure B.40: Collapsed Subprocesses in Vertical Modularization of P5
Expanded Subprocesses
Figure B.41: Expanded Subprocesses in Vertical Modularization of P5
154
B.6 Image 5 (P5)
Self Designed (Vertical)
Figure B.42: Self Designed in Vertical Modularization of P5
Link Events
Figure B.43: Link Event in Horizontal Modularization of P5
155
B Study
Message Events
Figure B.44: Message Event in Horizontal Modularization of P5
Self Designed (Horizontal)
Figure B.45: Self Designed in Horizontal Modularization of P5
156
B.7 Image 6 (P6)
Set of Statements
Statement Horizontal Modular-
ization
Vertical Modulariza-
tion
Orthogonal Modu-
larization
ME LE SDH ES CS SDV EE AO SDO
Kein Modul ist in der Abbildung gegeben false false false false false false false false false
Der Auftrag kann, muss aber nicht entgegengenommen werden false false false false false false false false false
Die letzte Aktivität bei der Ausführung, des aufgerufenen Moduls
ist "Gemäß ausgearbeiteter Lösung agieren"
true true true true true true true false false
Im aufgerufenen Modul ist die Aktivität "Lösung verhandeln"
gegeben
true true true true true true true true true
Table B.5: Statements of P5
B.7 Image 6 (P6)
Image 6 comprises the process of managing an order. There are two possibilities re-
garding the outcome of the process. First, the order can be sent. Second, customization
is carried out.
Aspect-Oriented
Figure B.46: Aspect-Ortiented in Orthogonal Modularization of P6
157
B Study
Exception Events
Figure B.47: Exception Events in Orthogonal Modularization of P6
Self Designed (Orthogonal)
Figure B.48: Self Designed in Orthogonal Modularization of P6
158
B.7 Image 6 (P6)
Collapsed Subprocesses
Figure B.49: Collapsed Subprocesses in Vertical Modularization of P6
Expanded Subprocesses
Figure B.50: Expanded Subprocesses in Vertical Modularization of P6
159
B Study
Self Designed (Vertical)
Figure B.51: Self Designed in Vertical Modularization of P6
Link Events
Figure B.52: Link Event in Horizontal Modularization of P6
160
B.7 Image 6 (P6)
Message Events
Figure B.53: Message Event in Horizontal Modularization of P6
Self Designed (Horizontal)
Figure B.54: Self Designed in Horizontal Modularization of P6
161
B Study
Set of Statements
Statement Horizontal Modular-
ization
Vertical Modulariza-
tion
Orthogonal Modu-
larization
ME LE SDH ES CS SDV EE AO SDO
Die Bestellung darf nicht versandbereit sein, um den Kunden zu
kontaktieren
true true true true true true true true true
In der Darstellung wird ein Modul aufgezeigt, das ausgeführt wird
wenn die Bestellung nicht versandbereit ist
true true true true true true true true true
Eine versandbereite Bestellung führt zur Modularisierung false false false false false false false false false
In der Darstellung wird ein Modul aufgezeigt, das nicht ausge-
führt wird wenn die Bestellung nicht versandbereit ist
false false false false false false false false false
Table B.6: Statements of P6
B.8 Image 7 (P7)
Image 7 comprises the topic PIN to a smartphone. If an incorrect pin is entered, the pin
has to be entered correctly to enable smartphone usage. Otherwise, if the pin is entered
correctly the smartphone is used.
Aspect-Oriented
Figure B.55: Aspect-Ortiented in Orthogonal Modularization of P7
162
B.8 Image 7 (P7)
Exception Events
Figure B.56: Exception Events in Orthogonal Modularization of P7
Self Designed (Orthogonal)
Figure B.57: Self Designed in Orthogonal Modularization of P7
163
B Study
Collapsed Subprocesses
Figure B.58: Collapsed Subprocesses in Vertical Modularization of P7
Expanded Subprocesses
Figure B.59: Expanded Subprocesses in Vertical Modularization of P7
Self Designed (Vertical)
Figure B.60: Self Designed in Vertical Modularization of P7
164
B.8 Image 7 (P7)
Link Events
Figure B.61: Link Event in Horizontal Modularization of P7
Message Events
Figure B.62: Message Event in Horizontal Modularization of P7
165
B Study
Self Designed (Horizontal)
Figure B.63: Self Designed in Horizontal Modularization of P7
Set of Statements
Statement Horizontal Modular-
ization
Vertical Modulariza-
tion
Orthogonal Modu-
larization
ME LE SDH ES CS SDV EE AO SDO
Der modularisierte Bereich muss nach dessen Aufruf nicht aus-
geführt werden
false false false false false false false false false
Der modularisierte Bereich muss nach dessen Aufruf ausgeführt
werden
true true true true true true true true true
Das Smartphone muss verwendet werden true true true true true true true true true
Wenn der Superpin notwendig ist, dann muss dieser richtig
eingegeben werden, damit das Smartphone verwendbar ist
false false false false false false false false false
Table B.7: Statements of P7
166
B Study
Self Designed (Orthogonal)
Figure B.66: Self Designed in Orthogonal Modularization of P8
Collapsed Subprocesses
Figure B.67: Collapsed Subprocesses in Vertical Modularization of P8
168
B.9 Image 8 (P8)
Expanded Subprocesses
Figure B.68: Expanded Subprocesses in Vertical Modularization of P8
Self Designed (Vertical)
Figure B.69: Self Designed in Vertical Modularization of P8
169
B Study
Link Events
Figure B.70: Link Event in Horizontal Modularization of P8
Message Events
Figure B.71: Message Event in Horizontal Modularization of P8
170
B.9 Image 8 (P8)
Self Designed (Horizontal)
Figure B.72: Self Designed in Horizontal Modularization of P8
Set of Statements
Statement Horizontal Modular-
ization
Vertical Modulariza-
tion
Orthogonal Modu-
larization
ME LE SDH ES CS SDV EE AO SDO
"Haftung hoch" ist der Pfad der eine Modularisierung beinhaltet true true true true true true true true true
Nach Genehmigung eines Darlehens, kann die Eingabe nicht
bestätigt werden
false false false false false false false false false
Der Kunde wird am Ende immer benachrichtigt false false false false false false false false false
Der Kunde wird am Ende nicht benachrichtigt true true true true true true true true true
Table B.8: Statements of P8
171
B Study
B.10 Image 9 (P9)
Image 9 comprises the theme card payment. The method of validation is chosen. Either
the entry of a pin is necessary, or a signature is required. The pin will be entered
incorrectly and the precess finishes with blocking the card.
Aspect-Oriented
Figure B.73: Aspect-Ortiented in Orthogonal Modularization of P9
Exception Events
Figure B.74: Exception Events in Orthogonal Modularization of P9
Self Designed (Orthogonal)
172
B.10 Image 9 (P9)
Figure B.75: Self Designed in Orthogonal Modularization of P9
Collapsed Subprocesses
Figure B.76: Collapsed Subprocesses in Vertical Modularization of P9
173
B Study
Expanded Subprocesses
Figure B.77: Expanded Subprocesses in Vertical Modularization of P9
Self Designed (Vertical)
Figure B.78: Self Designed in Vertical Modularization of P9
174
B.10 Image 9 (P9)
Link Events
Figure B.79: Link Event in Horizontal Modularization of P9
Message Events
Figure B.80: Message Event in Horizontal Modularization of P9
175
B Study
Self Designed (Horizontal)
Figure B.81: Self Designed in Horizontal Modularization of P9
Set of Statements
Statement Horizontal Modular-
ization
Vertical Modulariza-
tion
Orthogonal Modu-
larization
ME LE SDH ES CS SDV EE AO SDO
Die Eingabeform "Pin" führt zu einem weiteren Modul true true true true true true true true true
Der Beleg muss immer unterschrieben werden false false false false false false false false false
Die Aktivität "Pin eingeben" muss immer durchgeführt werden false false false false false false false false false
Modularisierung ist in der Abbildung gegeben true true true true true true true true true
Table B.9: Statements of P9
176
B.11 Questionnaires
B.11 Questionnaires
Three questionnaires are utilized in the study. First, the demographic questionnaire is
given. The questionnaire concerning cognitive load and understandability are given at
the end of the study.
Demographic Questionnaire
Seite 5 von 8
Persönlicher Code
Um die erhobenen Daten zu anonymisieren, verwenden wir für den weiteren Verlauf der Studie
einen Code. Erstellen Sie diesen Code nach dem folgenden Muster:
1. Zweite Ziffer des Tags Ihres Geburtsdatums
2. Erster Buchstabe Ihres Vornamens
3. Zweite Ziffer Ihres Alters
4. Letzter Buchstabe Ihres Familiennamens
5. Letzte Ziffer Ihres Geburtsjahres
6. Letzter Buchstabe des Vornamens
7. Erste Ziffer des Monats Ihres Geburtsdatums
8. Erster Buchstabe des Familiennamens
9. Berechnen Sie die Prüfsumme aller Ziffern
Beispiel:
Name: John Public
Alter: 29
Geburtstag: 19.05.1988
1. Zweite Ziffer des Tags Ihres Geburtsdatums: 19 9
2. Erster Buchstabe Ihres Vornamens: John J
3. Zweite Ziffer Ihres Alters: 29 9
4. Letzter Buchstabe Ihres Familiennamens: Public c
5. Letzte Ziffer Ihres Geburtsjahrs: 1988 8
6. Letzter Buchstabe des Vornamens: John n
7. Erste Ziffer des Monats Ihres Geburtsdatums: 05 0
8. Erster Buchstabe des Familiennamens: Public P
9. Berechnen Sie die Prüfsumme aller Ziffern: 9+9+8+0 26
Der in diesem Beispiel generierte Code lautet 9J9c8n0P26.
Ihr persönlicher Code: _______________________________________
Figure B.82: Declaration of Consent
177
B Study
Seite 6 von 8
Demographischer Fragebogen
1. Geben Sie Ihr Geschlecht an:
○ weiblich
○ männlich
○ keine Angabe
2. Geben Sie Ihr Alter an:
○ jünger als 25
○ 25 – 35
○ 36 – 45
○ 46 – 55
○ älter als 55
3. Geben Sie Ihren höchsten Bildungsabschluss an:
○ Ohne Abschluss
○ Hauptschulabschluss oder Volkshochschulabschluss
○ Realschulabschluss (Mittlere Reife)
○ Fachhochschulreife
○ Hochschulreife (Abitur)
○ Fachhochschulabschluss
○ Bachelor Hochschulabschluss
○ Master Hochschulabschluss
○ Sonstiger Abschluss, und zwar _____________________________
4. Wie hoch ist Ihre aktuelle Anzahl an Ausbildungsjahren (inkl. Grundschule)?
________________________________________________________
Figure B.83: Declaration of Consent
178
B.11 Questionnaires
Seite 7 von 8
5. Welche berufliche Ausbildung trifft am ehesten auf Sie zu?
○ Auszubildende(r) / Student(in)
○ Abgeschlossene Berufsausbildung
○ Abgeschlossene Ausbildung an einer Meister- oder Technikerschule
○ Akademiker(in)
○ Sonstiges, und zwar _____________________________________
6. Falls Sie studieren (oder studiert haben), geben Sie bitte Ihren Studiengang an:
________________________________________________________
7. Haben Sie bereits Erfahrung mit Prozessmodellen bzw. Prozessmodellierung?
○ ja
○ nein
8. Vor wie vielen Jahren haben Sie mit der Modellierung von Prozessmodellen begonnen?
________________________________________________________
9. Wie viele Prozessmodelle haben Sie innerhalb der letzten 12 Monate gelesen oder
analysiert?
________________________________________________________
10. Wie viele Prozessmodelle haben Sie innerhalb der letzten 12 Monate erstellt oder
bearbeitet?
________________________________________________________
Figure B.84: Declaration of Consent
179
B Study
Seite 8 von 8
11. Wie viele Aktivitäten hatten diese Modelle im Durchschnitt?
________________________________________________________
12. Wie viele Tage formale Ausbildung zum Thema „Prozessmodellierung“ haben Sie in den
letzten 12 Monaten erhalten?
Beispiel: Eine Vorlesung mit 2 Stunden / Woche ergibt ca. 8 Stunden Ausbildung pro Mo-
nat. Das entspricht einem Arbeitstag pro Monat. Eine Vorlesung inkl. Übungen ergibt folg-
lich 2 Arbeitstage pro Monat.
________________________________________________________
13. Wie viele Tage haben Sie in den letzten 12 Monaten mit dem Selbststudium zum Thema
„Prozessmodellierung“ verbracht? Beachten Sie die gleiche Beispielsrechnung wie vorher.
________________________________________________________
14. Bitte kreuzen Sie an, welche Notation zur Prozessmodellierung haben Sie während Ihrer
Ausbildung/Studium als erstes gelernt?
○ BPMN ○ Deklarativ ○ eGantt
○ EPK ○ Flow Chart ○ IDEF 3
○ Petri Netz ○ UML Aktivitätsd. ○ Andere
15. Mit welcher der in Frage 14 angegebenen Notationen zur Prozessmodellierung haben Sie
bisher die meiste Zeit verbracht?
________________________________________________________
Figure B.85: Declaration of Consent
180
B.11 Questionnaires
Cognitive Load and Understandability
Fragebogen
Ihr Code: _________________
Bitte beurteilen Sie die soeben ausgeführte Aufgabe:
sehr sehr
gering hoch
Bei der Aufgabe war Ihre mentale Anstrengung...
!
!
!
!
!
!
!
sehr sehr
leicht schwer
Wie leicht oder schwer war die Aufgabe zu lösen?
!
!
!
!
!
!
!
sehr sehr
wenig viel
Wieviel Spaß hatten Sie bei der Bearbeitung der Aufgabe?
!
!
!
!
!
!
!
stimmt stimmt
absolut völlig
nicht
Bei der Aufgabe musste man viele Dinge gleichzeitig im Kopf
bearbeiten.
Diese Aufgabe war sehr komplex.
Sie haben sich angestrengt, sich nicht nur einzelne Dinge zu merken,
sondern auch den Gesamtzusammenhang zu verstehen.
Es ging Ihnen beim Bearbeiten der Lerneinheiten darum, alles richtig
zu verstehen.
Die Lerneinheit enthielt Elemente, die Sie unterstützten, den Lernstoff
besser zu verstehen.
Bei dieser Aufgabe ist es mühsam, die wichtigsten Informationen zu
erkennen.
Die Darstellung bei dieser Aufgabe ist ungünstig, um wirklich etwas zu
lernen.
Bei dieser Aufgabe ist es schwer, die zentralen Inhalte miteinander in
Verbindung zu bringen.
Sie haben sich bei der Aufgabe angestrengt.
Die Aufgabe war anstrengend.
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
Figure B.86: Questionnaire Section of Cognitive Load
181
B Study
stimmt stimmt
absolut völlig
nicht
Auf diese Weise repräsentierte Geschäftsprozessmodelle wären für
die Nutzer schwer zu verstehen.
Sie denken, dass diese Präsentationsansätze eine effektive Lösung
für das Problem der Darstellung von Geschäftsprozessmodellen
bietet.
Die Verwendung von Prozessmodellen dieser Art würde die
Kommunikation von Geschäftsprozessen an den Endnutzer
erschweren.
Insgesamt erachten Sie die Geschäftsprozessmodelle in diesem
Experiment als nützlich.
Diese Art der Modellierung von Geschäftsprozessen zu erlernen, wäre
für Sie einfach.
Sie halten die Darstellung der Prozesse für unklar und schwer
verständlich.
Es wäre leicht für Sie, diese Art von Modellierung von
Geschäftsprozessen zu beherrschen.
Insgesamt hat sich die Modellierung dieser Art von
Geschäftsprozessen als schwierig erwiesen.
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
Figure B.87: Questionnaire Section of Understandability
182
B.11 Questionnaires
183
B Study
B.12 Declaration of Consent
Seite 4 von 8
Fakultät für Ingenieurwissenschaften, Informatik und
Psychologie
Institut für Datenbanken und Informationssysteme
Michael Zimoch
Albert-Einstein-Allee 11
89081 Ulm, Deutschland
Tel: +49 731 50-24230
michael.zimoch@uni-ulm.de
http://www.informatik.uni-ulm.de/dbis
Einwilligungserklärung:
Studie – Verständlichkeit von Prozessmodellen
Inhalt, Vorgehensweise, Risiken und Ziel des obengenannten Forschungsprojektes sowie die Befugnis zur
Einsichtnahme in die erhobenen Daten wurden ausreichend erklärt.
Ich hatte zusätzliche Fragen:
Ich hatte Gelegenheit Fragen zu stellen und habe hierauf Antwort erhalten. Ich hatte ausreichend Zeit, mich
für oder gegen die Teilnahme am Projekt zu entscheiden.
INFORMATION UND EINWILLIGUNGSERKLÄRUNG ZUM DATENSCHUTZ
Bei wissenschaftlichen Studien werden persönliche Daten über Sie erhoben. Die Speicherung, Auswer-
tung und Weitergabe dieser studienbezogenen Daten erfolgt nach gesetzlichen Bestimmungen und setzt
vor Teilnahme an der Studie folgende freiwillige Einwilligung voraus:
1. Ich erkläre mich damit einverstanden, dass im Rahmen dieser Studie erhobene Daten auf Frage-
bögen u. elektr. Datenträgern aufgezeichnet und ohne Namensnennung verarbeitet werden.
2. Außerdem erkläre ich mich damit einverstanden, dass eine autorisierte und zur Verschwiegen-
heit verpflichtete Person in meine erhobenen personenbezogenen Daten Einsicht nimmt, soweit
dies für die Überprüfung des Projektes notwendig ist.
Ich willige in die Teilnahme am Forschungsprojekt und die beschriebene Verwendung meiner Daten ein.
___________________ ______________________ ______________________________
Ort, Datum Name (Druckbuchstaben) Unterschrift (Teilnehmer)
Figure B.88: Declaration of Consent
184
List of Figures
2.1 Basic Elements of BPMN 2.0 Models . . . . . . . . . . . . . . . . . . . . . 6
2.2 Modular Subdivision [44] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Collapsed Subprocesses in Vertical Modularization . . . . . . . . . . . . . 11
2.4 Expanded Subprocesses in Vertical Modularization . . . . . . . . . . . . . 11
2.5 Message Event in Horizontal Modularization . . . . . . . . . . . . . . . . . 12
2.6 Link Event in Horizontal Modularization . . . . . . . . . . . . . . . . . . . 12
2.7
Exception Handling through Interrupting Event I in Orthogonal Modular-
ization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.8
Exception Handling through Interrupting Event II in Orthogonal Modular-
ization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.9 Aspect-Oriented Approach in Orthogonal Modularization . . . . . . . . . . 15
3.1 Structure of the Memory System [67] (simplified) . . . . . . . . . . . . . . 18
3.2 Theoretical basis of the Cognitive Load Theory of [80] . . . . . . . . . . . 20
3.3 Cognitive Pupillometry [90] . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.1 Graphics in the Study and its Realization . . . . . . . . . . . . . . . . . . 37
4.2 Structure of Eye Tracking Material . . . . . . . . . . . . . . . . . . . . . . 38
4.3 Different types of Validity [124] adapted . . . . . . . . . . . . . . . . . . . 44
5.1 Survey Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.2 Survey Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.3 Study Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.4 Study Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.5 Trials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
A.2 Section 2-1 of Demographic Questionnaire . . . . . . . . . . . . . . . . . 101
A.3 Section 2-2 of Demographic Questionnaire . . . . . . . . . . . . . . . . . 102
A.4 Section 2-3 of Demographic Questionnaire . . . . . . . . . . . . . . . . . 103
185
List of Figures
A.5 Section 2-4 of Demographic Questionnaire . . . . . . . . . . . . . . . . . 104
A.6 Section 2-5 of Demographic Questionnaire . . . . . . . . . . . . . . . . . 105
A.7 Section 3-1 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
A.8 Section 3-2 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
A.9 Section 3-3 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
A.10 Image of First Iteration in Horizontal Modularization . . . . . . . . . . . . . 108
A.11 Image of Second Iteration in Horizontal Modularization . . . . . . . . . . . 109
A.12 Image of Third Iteration in Horizontal Modularization . . . . . . . . . . . . 109
A.13 Image of Fourth Iteration in Horizontal Modularization . . . . . . . . . . . 109
A.14 Image of First Iteration in Vertical Modularization . . . . . . . . . . . . . . 110
A.15 Image of Second Iteration in Vertical Modularization . . . . . . . . . . . . 110
A.16 Image of Third Iteration in Vertical Modularization . . . . . . . . . . . . . . 110
A.17 Image of Fourth Iteration in Vertical Modularization . . . . . . . . . . . . . 111
A.18 Image of First Iteration in Orthogonal Modularization . . . . . . . . . . . . 111
A.19 Image of Second Iteration in Orthogonal Modularization . . . . . . . . . . 112
A.20 Image of Third Iteration in Orthogonal Modularization . . . . . . . . . . . 112
A.21 Image of Fourth Iteration in Orthogonal Modularization . . . . . . . . . . . 112
A.22 Questions in Section 4-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
A.23 Questions in Section 4-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
A.24 Questions in Section 4-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
A.25 Questions in Section 4-4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
A.26 Questions in Section 4-5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
A.27 Questions in Section 4-6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
A.28 Questions in Section 4-7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
A.29 Image and Description in Horizontal Modularization . . . . . . . . . . . . . 120
A.30 Suggested Icons in Horizontal Modularization . . . . . . . . . . . . . . . . 121
A.31 Image and Description in Vertical Modularization . . . . . . . . . . . . . . 122
A.32 Suggested Icons in Vertical Modularization . . . . . . . . . . . . . . . . . 123
A.33 Image and Description in Orthogonal Modularization . . . . . . . . . . . . 124
A.34 Suggested Icons in Orthogonal Modularization . . . . . . . . . . . . . . . 125
A.35 Section 5-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
186
List of Figures
A.36 Section 5-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
A.37 Section 5-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
A.38 Section 5-4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
A.39 Section 5-5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
A.40 Feedback and Codeword . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
A.41 Literature Presented . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
B.1 Aspect-Ortiented in Orthogonal Modularization of P1 . . . . . . . . . . . . 136
B.2 Exception Events in Orthogonal Modularization of P1 . . . . . . . . . . . 136
B.3 Self Designed in Orthogonal Modularization of P1 . . . . . . . . . . . . . 137
B.4 Collapsed Subprocesses in Vertical Modularization of P1 . . . . . . . . . 137
B.5 Expanded Subprocesses in Vertical Modularization of P1 . . . . . . . . . 137
B.6 Self Designed in Vertical Modularization of P1 . . . . . . . . . . . . . . . . 138
B.7 Link Event in Horizontal Modularization of P1 . . . . . . . . . . . . . . . . 138
B.8 Message Event in Horizontal Modularization of P1 . . . . . . . . . . . . . 139
B.9 Self Designed in Horizontal Modularization of P1 . . . . . . . . . . . . . . 139
B.10 Aspect-Ortiented in Orthogonal Modularization of P2 . . . . . . . . . . . . 140
B.11 Exception Events in Orthogonal Modularization of P2 . . . . . . . . . . . 140
B.12 Self Designed in Orthogonal Modularization of P2 . . . . . . . . . . . . . 141
B.13 Collapsed Subprocesses in Vertical Modularization of P2 . . . . . . . . . 141
B.14 Expanded Subprocesses in Vertical Modularization of P2 . . . . . . . . . 141
B.15 Self Designed in Vertical Modularization of P2 . . . . . . . . . . . . . . . . 142
B.16 Link Event in Horizontal Modularization of P2 . . . . . . . . . . . . . . . . 142
B.17 Message Event in Horizontal Modularization of P2 . . . . . . . . . . . . . 143
B.18 Self Designed in Horizontal Modularization of P2 . . . . . . . . . . . . . . 143
B.19 Aspect-Ortiented in Orthogonal Modularization of P3 . . . . . . . . . . . . 144
B.20 Exception Events in Orthogonal Modularization of P3 . . . . . . . . . . . 145
B.21 Self Designed in Orthogonal Modularization of P3 . . . . . . . . . . . . . 145
B.22 Collapsed Subprocesses in Vertical Modularization of P3 . . . . . . . . . 145
B.23 Expanded Subprocesses in Vertical Modularization of P3 . . . . . . . . . 146
B.24 Self Designed in Vertical Modularization of P3 . . . . . . . . . . . . . . . . 146
B.25 Link Event in Horizontal Modularization of P3 . . . . . . . . . . . . . . . . 146
187
List of Figures
B.26 Message Event in Horizontal Modularization of P3 . . . . . . . . . . . . . 147
B.27 Self Designed in Horizontal Modularization of P3 . . . . . . . . . . . . . . 147
B.28 Aspect-Ortiented in Orthogonal Modularization of P4 . . . . . . . . . . . . 148
B.29 Exception Events in Orthogonal Modularization of P4 . . . . . . . . . . . 148
B.30 Self Designed in Orthogonal Modularization of P4 . . . . . . . . . . . . . 149
B.31 Collapsed Subprocesses in Vertical Modularization of P4 . . . . . . . . . 149
B.32 Expanded Subprocesses in Vertical Modularization of P4 . . . . . . . . . 149
B.33 Self Designed in Vertical Modularization of P4 . . . . . . . . . . . . . . . . 150
B.34 Link Event in Horizontal Modularization of P4 . . . . . . . . . . . . . . . . 150
B.35 Message Event in Horizontal Modularization of P4 . . . . . . . . . . . . . 151
B.36 Self Designed in Horizontal Modularization of P4 . . . . . . . . . . . . . . 151
B.37 Aspect-Ortiented in Orthogonal Modularization of P5 . . . . . . . . . . . . 152
B.38 Exception Events in Orthogonal Modularization of P5 . . . . . . . . . . . 153
B.39 Self Designed in Orthogonal Modularization of P5 . . . . . . . . . . . . . 153
B.40 Collapsed Subprocesses in Vertical Modularization of P5 . . . . . . . . . 154
B.41 Expanded Subprocesses in Vertical Modularization of P5 . . . . . . . . . 154
B.42 Self Designed in Vertical Modularization of P5 . . . . . . . . . . . . . . . . 155
B.43 Link Event in Horizontal Modularization of P5 . . . . . . . . . . . . . . . . 155
B.44 Message Event in Horizontal Modularization of P5 . . . . . . . . . . . . . 156
B.45 Self Designed in Horizontal Modularization of P5 . . . . . . . . . . . . . . 156
B.46 Aspect-Ortiented in Orthogonal Modularization of P6 . . . . . . . . . . . . 157
B.47 Exception Events in Orthogonal Modularization of P6 . . . . . . . . . . . 158
B.48 Self Designed in Orthogonal Modularization of P6 . . . . . . . . . . . . . 158
B.49 Collapsed Subprocesses in Vertical Modularization of P6 . . . . . . . . . 159
B.50 Expanded Subprocesses in Vertical Modularization of P6 . . . . . . . . . 159
B.51 Self Designed in Vertical Modularization of P6 . . . . . . . . . . . . . . . . 160
B.52 Link Event in Horizontal Modularization of P6 . . . . . . . . . . . . . . . . 160
B.53 Message Event in Horizontal Modularization of P6 . . . . . . . . . . . . . 161
B.54 Self Designed in Horizontal Modularization of P6 . . . . . . . . . . . . . . 161
B.55 Aspect-Ortiented in Orthogonal Modularization of P7 . . . . . . . . . . . . 162
B.56 Exception Events in Orthogonal Modularization of P7 . . . . . . . . . . . 163
188
List of Figures
B.57 Self Designed in Orthogonal Modularization of P7 . . . . . . . . . . . . . 163
B.58 Collapsed Subprocesses in Vertical Modularization of P7 . . . . . . . . . 164
B.59 Expanded Subprocesses in Vertical Modularization of P7 . . . . . . . . . 164
B.60 Self Designed in Vertical Modularization of P7 . . . . . . . . . . . . . . . . 164
B.61 Link Event in Horizontal Modularization of P7 . . . . . . . . . . . . . . . . 165
B.62 Message Event in Horizontal Modularization of P7 . . . . . . . . . . . . . 165
B.63 Self Designed in Horizontal Modularization of P7 . . . . . . . . . . . . . . 166
B.64 Aspect-Ortiented in Orthogonal Modularization of P8 . . . . . . . . . . . . 167
B.65 Exception Events in Orthogonal Modularization of P8 . . . . . . . . . . . 167
B.66 Self Designed in Orthogonal Modularization of P8 . . . . . . . . . . . . . 168
B.67 Collapsed Subprocesses in Vertical Modularization of P8 . . . . . . . . . 168
B.68 Expanded Subprocesses in Vertical Modularization of P8 . . . . . . . . . 169
B.69 Self Designed in Vertical Modularization of P8 . . . . . . . . . . . . . . . . 169
B.70 Link Event in Horizontal Modularization of P8 . . . . . . . . . . . . . . . . 170
B.71 Message Event in Horizontal Modularization of P8 . . . . . . . . . . . . . 170
B.72 Self Designed in Horizontal Modularization of P8 . . . . . . . . . . . . . . 171
B.73 Aspect-Ortiented in Orthogonal Modularization of P9 . . . . . . . . . . . . 172
B.74 Exception Events in Orthogonal Modularization of P9 . . . . . . . . . . . 172
B.75 Self Designed in Orthogonal Modularization of P9 . . . . . . . . . . . . . 173
B.76 Collapsed Subprocesses in Vertical Modularization of P9 . . . . . . . . . 173
B.77 Expanded Subprocesses in Vertical Modularization of P9 . . . . . . . . . 174
B.78 Self Designed in Vertical Modularization of P9 . . . . . . . . . . . . . . . . 174
B.79 Link Event in Horizontal Modularization of P9 . . . . . . . . . . . . . . . . 175
B.80 Message Event in Horizontal Modularization of P9 . . . . . . . . . . . . . 175
B.81 Self Designed in Horizontal Modularization of P9 . . . . . . . . . . . . . . 176
B.82 Declaration of Consent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
B.83 Declaration of Consent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
B.84 Declaration of Consent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
B.85 Declaration of Consent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
B.86 Questionnaire Section of Cognitive Load . . . . . . . . . . . . . . . . . . . 181
B.87 Questionnaire Section of Understandability . . . . . . . . . . . . . . . . . 182
189
List of Tables
3.1 Cognitive Weights W for BPM Elements [74] . . . . . . . . . . . . . . . . . 19
4.1 Process Models in the Survey and their Realization . . . . . . . . . . . . . 36
4.2 Items to measure Cognitive Load . . . . . . . . . . . . . . . . . . . . . . . 39
4.3
Items to measure Perceived Usefulness for Understandability and Per-
ceived Ease of Understanding . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.1 Independent Variables in the Survey . . . . . . . . . . . . . . . . . . . . . 55
5.2 Independent Variables in the Study . . . . . . . . . . . . . . . . . . . . . . 56
6.1 Intrinsic Cognitive Load in the Survey . . . . . . . . . . . . . . . . . . . . 59
6.2 Multiple Comparisons on Intrinsic Cognitive Load in the Survey . . . . . . 59
6.3 Germane Cognitive Load in the Survey . . . . . . . . . . . . . . . . . . . 60
6.4 Multiple Comparisons on Germane Cognitive Load in the Survey . . . . . 61
6.5 Extraneous Cognitive Load in the Survey . . . . . . . . . . . . . . . . . . 61
6.6 Multiple Comparisons on Extraneous Cognitive Load in the Survey . . . . 62
6.7 Perceived Usefulness for Understandability in the Survey . . . . . . . . . 62
6.8 Multiple Comparisons on PUU in the Survey . . . . . . . . . . . . . . . . . 63
6.9 Perceived Ease of Understanding in the Survey . . . . . . . . . . . . . . . 63
6.10 Multiple Comparisons on PEU in the Survey . . . . . . . . . . . . . . . . . 63
6.11 Performance Success in the Study . . . . . . . . . . . . . . . . . . . . . . 64
6.12 Multiple Comparisons on Performance Success in the Study . . . . . . . 64
6.13 Key Performance Indicator (KPI) in the Study . . . . . . . . . . . . . . . . 65
6.14 Multiple Comparisons on KPI in the Study . . . . . . . . . . . . . . . . . . 65
6.15 Horizontal Modularization in the Study . . . . . . . . . . . . . . . . . . . . 66
6.16 Multiple Comparisons on Horizontal Modularization in the Study . . . . . 67
6.17 Vertical Modularization in the Study . . . . . . . . . . . . . . . . . . . . . . 67
6.18 Multiple Comparisons on Vertical Modularization in the Study . . . . . . . 68
6.19 Orthogonal Modularization in the Study . . . . . . . . . . . . . . . . . . . 69
6.20 Multiple Comparisons on Orthogonal Modularization in the Study . . . . . 70
191
List of Tables
B.1 Statements of P1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
B.2 Statements of P2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
B.3 Statements of P3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
B.4 Statements of P4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
B.5 Statements of P5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
B.6 Statements of P6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
B.7 Statements of P7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
B.8 Statements of P8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
B.9 Statements of P9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
192
Name: Julia Baß Matriculation number: 759965
Honesty disclaimer
I hereby affirm that I wrote this thesis independently and that I did not use any other
sources or tools than the ones specified.
Ulm,
.................... .............................................
Julia Baß
