bpm cluster meeting 2014

48
1/31 www.janclaes.info FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION Jan Claes Teaching assistant : PhD 2009 – 2015 Supervisors UGent : Geert Poels & Frederik Gailly Supervisors TU/e : Paul Grefen & Irene Vanderfeesten Why do people struggle with the complexity of constructing a process model?

Upload: jan-claes

Post on 24-Dec-2014

50 views

Category:

Business


4 download

DESCRIPTION

Slides of my presentation at BPM Cluster meeting, 21 August 2014, Eindhoven, NL

TRANSCRIPT

Page 1: BPM Cluster Meeting 2014

1/31www.janclaes.info

FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION

Jan ClaesTeaching assistant : PhD 2009 – 2015

Supervisors UGent : Geert Poels & Frederik GaillySupervisors TU/e : Paul Grefen & Irene Vanderfeesten

Why do people struggle with the complexity of constructing a process model?

Page 2: BPM Cluster Meeting 2014

2/31www.janclaes.info

Business Process Modeling (BPM)Assumptions

Constructing process models is useful Constructing the models is cognitively challenging

Business Process Modeling (BPM) Process model quality

“what is a good process model?” Process of process modeling (PPM)

“how to make a good process model (effectiveness) cccccccc using a good approach (efficiency)?”

Page 3: BPM Cluster Meeting 2014

3/31www.janclaes.info

FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION

How do people construct process models?

Page 4: BPM Cluster Meeting 2014

4/31www.janclaes.info

PPMChart visualization

CREATE_ACTIVITY CREATE_START_EVENT CREATE_END_EVENT CREATE_AND CREATE_XOR CREATE_EDGE MOVE_ACTIVITY MOVE_START_EVENT MOVE_END_EVENT MOVE_AND MOVE_XOR DELETE_ACTIVITY DELETE_START_EVENT DELETE-END_EVENT DELETE_AND DELETE_XOR DELETE_EDGE NAME_ACTIVITY RENAME_ACTIVITY NAME_EDGE RENAME_EDGE

Start event Edge Activity

Gateway

Edge

Activity

Edge

Edge

Activity

Edge Gateway

Edge

123

5

7

9

4

6

8

1112

10

time

mod

el e

lem

ents

Page 5: BPM Cluster Meeting 2014

5/31www.janclaes.info

Flow-oriented modeling

Page 6: BPM Cluster Meeting 2014

6/31www.janclaes.info

Aspect-oriented modeling

Page 7: BPM Cluster Meeting 2014

7/31www.janclaes.info

Undirected modeling

Page 8: BPM Cluster Meeting 2014

8/31www.janclaes.info

Observations (objective)

Obs. 1. Majority used serialized approachObs. 2. Large group flow-oriented modelingObs. 3. Small group aspect-oriented

modelingObs. 4. Large group combine FO and AOObs. 5. Small group undirected modelingObs. 6. Remainder is uncategorizedObs. 7. Undirected sessions lasted longer

Page 9: BPM Cluster Meeting 2014

9/31www.janclaes.info

FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION

What about the quality of the corresponding models?

Page 10: BPM Cluster Meeting 2014

10/31www.janclaes.info

Impressions (subjective)

Impr. 1. Modelers need serializationImpr. 2. Structured serialization

helps avoiding ‘mistakes’Impr. 3. Structured serialization

does not support every modeler to the same extent

Page 11: BPM Cluster Meeting 2014

11/31www.janclaes.info

FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION

Is there a rational for these observations and impressions?

Page 12: BPM Cluster Meeting 2014

12/31www.janclaes.info

Kinds of human memory

input Sensory memory

Workingmemory

Long term memory

performanceOBSERVATION SELECTION

ORGANIZATION

RETRIEVALTRANSFER

Workingmemory

Page 13: BPM Cluster Meeting 2014

13/31www.janclaes.info

Types of cognitive load

Time

Inst

anta

neou

s lo

ad

overall load

intrinsicload

extraneousload

germane load

freecapacity

assumed capacity

instantaneous load

overload

Page 14: BPM Cluster Meeting 2014

14/31www.janclaes.info

Cognitive Load Theory (CLT)

case complexity prior knowledgeinput material

representation fitworking memory

capacity

extraneous cognitive load

germane cognitive load

schema construction

process model quality

A B The more A, the more B A B The more A, the less B A B The more A, the more B on the long term

overall construction effort

cognitive overload

intrinsic cognitive load

++

+

+

+++

+ +

– – –

Page 15: BPM Cluster Meeting 2014

15/31www.janclaes.info

Cognitive Fit Theory (CFT)

Need for structure

Learning styleField dependencySensory

Visual

Inductive

Active

Sequential

Intuitive

Auditory

Deductive

Reflective

Global

I ♥︎� chaos I #$@&! chaos

DON’T SEE THE FOREST THROUGH THE TREES

Page 16: BPM Cluster Meeting 2014

16/31www.janclaes.info

FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION

What can we do with this knowledge?

Page 17: BPM Cluster Meeting 2014

17/31www.janclaes.info

ImpressionsCognitive theoryObservations

Theory building

Structured Process Modeling Theory

17/32www.janclaes.info

Page 18: BPM Cluster Meeting 2014

18/31www.janclaes.info

Structured Process Modeling Theory (SPMT)

A B A determines BA B The more A, the more B+ A B The more A, the less B– A B A translates into B

learning style

degree of serialization

adopted serialization style

need for structure

field-dependency

– +

intrinsic cognitive load for modeling phases

intrinsic cognitive load for aggregation phases

cognitive overload

intrinsic cognitive load for strategy building phases

+ + +

serialization style fitstructuredness of serialization

– – – –

1 2 3

Page 19: BPM Cluster Meeting 2014

19/31www.janclaes.info

FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION

How will the developed theory be tested?

Page 20: BPM Cluster Meeting 2014

20/31www.janclaes.info

Methodology1. Cognitive tests prior to the test session

Collect modeler data2. Observational modeling session

At TU/e in BPM lecture of September 26 Collect modeler, modeling and model data

3. Post-experiment interviews with selection of participants (e.g.,

outliers) Collect additional, qualitative data

Page 21: BPM Cluster Meeting 2014

21/31www.janclaes.info

Experiment set-upNo treatmentMeasure every involved construct of the theory

Modeler, modeling process, process modelCalculate (cor)relations of the theory

Page 22: BPM Cluster Meeting 2014

22/31www.janclaes.info

PRIOR TO THE EXPERIMENT SESSION

HOMEWORK

Page 23: BPM Cluster Meeting 2014

23/31www.janclaes.info

Experiment set-up1. Prior to the experiment session

Software to be installed on personal laptop ( = download from website and unzip)

5 cognitive tests (time range 5-30 min)• Reading span test (RSPAN)• Operation span test (OSPAN)• Counting span test (CSPAN)• Hidden figures test (HFT)• Need for structure (NFS)

Page 24: BPM Cluster Meeting 2014

24/31www.janclaes.info

Experiment set-up1. Prior to the experiment session

To be completed at home Before deadline Software collects data in one-zip file per test

which participants have to submit

Page 25: BPM Cluster Meeting 2014

25/31www.janclaes.info

Experiment set-up1. Prior to the experiment session

Students will receive their individual results as incentive to not cheat on these tests

Late responses or too much errors reveals cheating Student is not allowed to start actual experiment if

the test results are not submitted in time

Page 26: BPM Cluster Meeting 2014

26/31www.janclaes.info

AT THE EXPERIMENT SESSION

MODELING TASK

Page 27: BPM Cluster Meeting 2014

27/31www.janclaes.info

Experiment set-up2. At the experiment session

Modeling task and post-task survey In the same, already installed program

(using a different start-up code) Software automatically collects data

• Each operation in the tool ( = each activity in the modeling process)

• Constructed process model, number of operations, modeling time

in one-zip file which they have to submit

Page 28: BPM Cluster Meeting 2014

28/31www.janclaes.info

Experiment set-up2. At the experiment session

Student does not get credit if not all data (6 zip files) are submitted

Technical problems > replacement assignment(e.g., help me with coding the collected data)

Page 29: BPM Cluster Meeting 2014

29/31www.janclaes.info

AFTER THE EXPERIMENT SESSION

INTERVIEW

Page 30: BPM Cluster Meeting 2014

30/31www.janclaes.info

Experiment set-up3. After the experiment session

Data will be analyzed and outliners will be defined Selection of students will be invited for post-

experiment interview (voluntarily) Collect qualitative data

• Explain how they have modeled• Explain their cognitive profile• Explain which results we expected and why• Ask about their opinion: why did they (not) achieve the

predicted results?

Page 31: BPM Cluster Meeting 2014

31/31www.janclaes.info

Thanks for you attention!

> Do you have feedback on my research plans? <> set-up is useful for theory testing? <

> set-up is feasible (enough)? <

> other suggestions? <

Does someone has some time to test the software?

Jan [email protected]

http://www.janclaes.infoTwitter: @janclaesbelgium

Page 32: BPM Cluster Meeting 2014

32/31www.janclaes.info

FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION

How can the concepts of the theory be measured?

Page 33: BPM Cluster Meeting 2014

33/31www.janclaes.info

Cognitive Load Theory (CLT)

case complexity prior knowledgeinput material

representation fitworking memory

capacity

extraneous cognitive load

germane cognitive load

schema construction

process model quality

A B The more A, the more B A B The more A, the less B A B The more A, the more B on the long term

overall construction effort

cognitive overload

intrinsic cognitive load

++

+

+

+++

+ +

– – –

Page 34: BPM Cluster Meeting 2014

34/31www.janclaes.info

overall construction effort

Cognitive Load Theory (CLT)

case complexity prior knowledgeinput material

representation fit

extraneous cognitive load

germane cognitive load

schema construction

process model quality

A B The more A, the more B A B The more A, the less B A B The more A, the more B on the long term

cognitive overload

intrinsic cognitive load

++

+

+

+++

+ +

– – –

working memory capacity

Reading span - Operation span - Counting span

Reading spanFor many years, my family and friends have been working on the farm. SPOT

Because the stuffy was room, Bob went air for some fresh outside. TRAILWe were fifty miles out at sea before we lost sight of the land. BAND

Counting span

Count the dark blue circles

Operation spanIS (8 / 4) – 1 = 1 ? BEAR

IS (6 X 2) – 2 = 10 ? BEANSIS (10 X 2) – 6 = 12 ? DAD

Page 35: BPM Cluster Meeting 2014

35/31www.janclaes.info

overall construction effort

Cognitive Load Theory (CLT)

case complexity prior knowledgeworking memory

capacity

extraneous cognitive load

germane cognitive load

schema construction

process model quality

A B The more A, the more B A B The more A, the less B A B The more A, the more B on the long term

cognitive overload

intrinsic cognitive load

++

+

+

+++

+ +

– – –

input material representation fit

Fit with modeling task = constantFit with modeler through questions

Example questionsI did experience problems using the modeling tool (5-point Likert)

Using the modeling language was not difficult (5-point Likert)I found it easy to understand the case description (5-point Likert)

Page 36: BPM Cluster Meeting 2014

36/31www.janclaes.info

Cognitive Load Theory (CLT)

prior knowledgeinput material

representation fitworking memory

capacity

extraneous cognitive load

germane cognitive load

schema construction

process model quality

A B The more A, the more B A B The more A, the less B A B The more A, the more B on the long term

overall construction effort

cognitive overload

intrinsic cognitive load

++

+

+

+++

+ +

– – –

case complexity

Case complexity = constant(number of activities and arrows in solution model can be used as proxy for number of concepts and relations)

Page 37: BPM Cluster Meeting 2014

37/31www.janclaes.info

overall construction effort

Cognitive Load Theory (CLT)

case complexityinput material representation fit

working memory capacity

extraneous cognitive load

germane cognitive load

schema construction

process model quality

A B The more A, the more B A B The more A, the less B A B The more A, the more B on the long term

cognitive overload

intrinsic cognitive load

++

+

+

+++

+ +

– – –

prior knowledge

Prior domain, modeling language and modeling method knowledgethrough questions

Example questionsI am familiar with the mortgage handling process (5 point-Likert)

How many process models did you ever create?I am experienced in process modeling (5-point-Likert)

Page 38: BPM Cluster Meeting 2014

38/31www.janclaes.info

overall construction effort

Cognitive Load Theory (CLT)

case complexity prior knowledgeinput material

representation fitworking memory

capacity

schema construction

process model quality

A B The more A, the more B A B The more A, the less B A B The more A, the more B on the long term

++

+

+

+++

+ +

– – –

extraneous cognitive load

germane cognitive load

cognitive overload

intrinsic cognitive load

Through questions

Example questionsI found it hard to complete the task (5-point-Likert)

Page 39: BPM Cluster Meeting 2014

39/31www.janclaes.info

Cognitive Load Theory (CLT)

case complexity prior knowledgeinput material

representation fitworking memory

capacity

extraneous cognitive load

germane cognitive load

process model quality

A B The more A, the more B A B The more A, the less B A B The more A, the more B on the long term

overall construction effort

cognitive overload

intrinsic cognitive load

++

+

+

+++

+ +

– – –

– schema construction

Page 40: BPM Cluster Meeting 2014

40/31www.janclaes.info

Cognitive Load Theory (CLT)

case complexity prior knowledgeinput material

representation fitworking memory

capacity

extraneous cognitive load

germane cognitive load

schema construction

A B The more A, the more B A B The more A, the less B A B The more A, the more B on the long term

overall construction effort

cognitive overload

intrinsic cognitive load

++

+

+

+++

+ +

– – –

process model quality

Syntactical quality: number of cognitive mistakesSemantic quality: assessment by experts using checklist (penalty for over- and underfit)

Page 41: BPM Cluster Meeting 2014

41/31www.janclaes.info

Cognitive Load Theory (CLT)

case complexity prior knowledgeinput material

representation fitworking memory

capacity

extraneous cognitive load

germane cognitive load

schema construction

process model quality

A B The more A, the more B A B The more A, the less B A B The more A, the more B on the long term

cognitive overload

intrinsic cognitive load

++

+

+

+++

+ +

– – –

overall construction effort

Time between start of reading case description and last operation in the toolNumber of operations

Page 42: BPM Cluster Meeting 2014

42/31www.janclaes.info

Structured Process Modeling Theory (SPMT)

A B A determines BA B The more A, the more B+ A B The more A, the less B– A B A translates into B

learning style

degree of serialization

adopted serialization style

need for structure

field-dependency

– +

intrinsic cognitive load for modeling phases

intrinsic cognitive load for aggregation phases

cognitive overload

intrinsic cognitive load for strategy building phases

+ + +

serialization style fitstructuredness of serialization

– – – –

Page 43: BPM Cluster Meeting 2014

43/31www.janclaes.info

Structured Process Modeling Theory (SPMT)

A B A determines BA B The more A, the more B+ A B The more A, the less B– A B A translates into B

learning style

degree of serialization

adopted serialization style

need for structure

field-dependency

– +

intrinsic cognitive load for modeling phases

intrinsic cognitive load for aggregation phases

cognitive overload

intrinsic cognitive load for strategy building phases

+ + +

serialization style fitstructuredness of serialization

– – – –

Through questions

Example questionsWhen I write a text:

• I usually work on content, structure, wording and format right from the start• I usually first decide on the content, then the structure, next I write sentences

and finally I work on the formatting

Page 44: BPM Cluster Meeting 2014

44/31www.janclaes.info

Structured Process Modeling Theory (SPMT)

A B A determines BA B The more A, the more B+ A B The more A, the less B– A B A translates into B

learning style

degree of serialization

adopted serialization style

need for structure

field-dependency

– +

intrinsic cognitive load for modeling phases

intrinsic cognitive load for aggregation phases

cognitive overload

intrinsic cognitive load for strategy building phases

+ + +

serialization style fitstructuredness of serialization

– – – –

Through questions

Need for Structure scale12 questions with 6-point Likert scale

Page 45: BPM Cluster Meeting 2014

45/31www.janclaes.info

Structured Process Modeling Theory (SPMT)

A B A determines BA B The more A, the more B+ A B The more A, the less B– A B A translates into B

learning style

degree of serialization

adopted serialization style

need for structure

field-dependency

– +

intrinsic cognitive load for modeling phases

intrinsic cognitive load for aggregation phases

cognitive overload

intrinsic cognitive load for strategy building phases

+ + +

serialization style fitstructuredness of serialization

– – – –

Hidden Figures Test

Page 46: BPM Cluster Meeting 2014

46/31www.janclaes.info

Structured Process Modeling Theory (SPMT)

A B A determines BA B The more A, the more B+ A B The more A, the less B– A B A translates into B

learning style

adopted serialization style

need for structure

field-dependency

– +

intrinsic cognitive load for modeling phases

intrinsic cognitive load for aggregation phases

cognitive overload

intrinsic cognitive load for strategy building phases

+ + +

serialization style fit– – – –

degree of serialization

structuredness of serialization

Assessment by experts

Flow-oriented – Aspect-oriented – CombinationHappy Path First – Undirected

Uncategorized

Page 47: BPM Cluster Meeting 2014

47/31www.janclaes.info

Structured Process Modeling Theory (SPMT)

A B A determines BA B The more A, the more B+ A B The more A, the less B– A B A translates into B

learning style

degree of serialization

adopted serialization style

need for structure

field-dependency

– +

intrinsic cognitive load for modeling phases

intrinsic cognitive load for aggregation phases

cognitive overload

intrinsic cognitive load for strategy building phases

+ + +

structuredness of serialization

– – – –

serialization style fit

Theoretical matchPost-experiment interview

Global learner = aspect-oriented, sequential = flow-orientedHigh need for structure = no undirected

Field-dependent = flow-oriented

Page 48: BPM Cluster Meeting 2014

48/31www.janclaes.info

Structured Process Modeling Theory (SPMT)

A B A determines BA B The more A, the more B+ A B The more A, the less B– A B A translates into B

learning style

degree of serialization

adopted serialization style

need for structure

field-dependency

– +

+ + +

serialization style fitstructuredness of serialization

– – – –

intrinsic cognitive load for modeling phases

intrinsic cognitive load for aggregation phases

cognitive overload

intrinsic cognitive load for strategy building phases

See earlier: through questions