Two cases on behavioral operations management research

Behavioral operations management –

two studies

Andreas Größler

Radboud University Nijmegen, the Netherlands

Behavioral OM: Trying a definition…

Operations Management (OM) is concerned with the design and

management of transformation processes in organisations, in order

to achieve (societal and/or economic) value.

Behavioral Operations Management (BOM) is a multi-disciplinary

branch of OM that explicitly considers the effects of human

behaviour and cognition on the transformation processes, which

are influence by individual biases, social preferences, and cultural

norms.

(cf. Loch&Wu, 2007)

Various special issues (JOM, POM, MSOM), academic conferences,

sub-groups in academic associations, ...

Nijmegen

2000 years old (lat. noviomagus)

Today ca. 170,000 inhabitants

Radboud University Nijmegen

• Seven faculties (including university hospital): 19,000 students

• Nijmegen School of Management: 3,100 students

European Master in System Dynamics

Bergen

Lisbon

Nijmegen

Palermo

Erasmus Mundus

Label of Excellence of

European Commission

Basis

kart

e: dig

itale

-euro

pakart

e.d

e

European Master in System Dynamics:

Overall programme

Intelligence, Knowledge, Personality, and

Interests – Determinants of Individual

Inventory Management Performance?

Jürgen Strohhecker and Andreas Größler (2013). Intelligence, Personality,

Interest and Knowledge – The effect of personal traits on inventory management

performance, International Journal of Production Economics, 142(3), 37–50.

Inventory management issues are commonplace in daily life

and in business.

Spiegel.de n24de

There are three bodies of literature dealing with the issue of

inventory management failures:

1. Normative approaches, operational research (for reviews cf., Williams and

Tokar, 2008; Gino and Pisano, 2008)

2. Psychological research on complex problem solving (e.g., Dörner, 1980;

Sternberg and Frensch, 1991; Brehmer, 1992; Brehmer and Dörner, 1993;

Ackerman and Kanfer, 1993; Dörner et al., 1994; Frensch and Funke,

1995; Dörner, 1996; Wittmann and Hattrup, 2004)

3. System dynamics research on stock management behaviour (e.g.,

Sterman,1989; Booth Sweeny and Sterman, 2000; Ossimitz, 2002;

Sterman and Booth Sweeny, 2002; Croson and Donohue, 2003, 2006;

Pala and Vennix, 2005; Cronin et al., 2009; Sterman, 2010)

In operational research, optimal solutions to inventory

management problems are sought.

Search.com

Inventoryops.com

In psychology, complex problem solving deals with the

behaviour of people in simulated situations.

Dörner

In system dynamics, experiments show the difficulty of

people to understand/to control stocks and flows.

The research gap lies in the intersection of the three

literatures.

OR – domain

CPS – person factors

SD – experimental

design

Why is stock management relevant? A more substantial

example…

ww

w.n

ationalg

eogra

phic

.com

Ackerman‘s (1996) PPIK theory has been tested against

performance in an inventory management task.

Pugepo: Inventory management for a pump producer as

experimental task.

Experimental design and participants:

• N = 72 participants, 3rd year bachelor students from German business

school; elective course “Operations Management”, autumn 2010

• Tests: - Intelligence: BIS at start of study

- Knowledge: average of marks for related courses from studies

- Personality: NEO FFI at start of experiment

- Interests: AIST-R at start of experiment

- Performance: PUGEPO simulation as last part of experiment, total accumulated costs

as measure

• Financial incentive: max. 9.55 €, average achieved 4.55 €

About 50% of participants fail to beat simple benchmark

strategies.

replicating incoming orders

keeping initial orders

Intelligence and knowledge have a clear effect on performance;

entrepreneurial interests have an adverse effect. A B C D E

Constant -.005 .028 .004 .001 .005

(.111) (.117) (.112) (.108) (.106)

BIS-AI-S -.363*** -.326*** -.383*** -.335***

(.111) .109 (.118) (.110)

WIWI -.286*** -.234** -.161 -.244**

(.122) (.118) (.131) (.115)

AIST-R practical and technical -.084

(.142)

AIST-R intellectual and investigative -.170

(.140)

AIST-R artistic and linguistic .062

(.137)

AIST-R social -.052

(.137)

AIST-R entrepreneurial .313** .219**

(.153) (.107)

AIST-R organizational and administrational -.016

(.123)

NEO-FFI neuroticism -.105

(.132)

NEO-FFI extraversion -.167

(.149)

NEO-FFI openness to experience .159

(.137)

NEO-FFI agreeableness .200

(.143)

NEO-FFI conscientiousness -.057

(.126)

R-squared .132 .082 .186 .330 .233

Adjusted R-squared .120 .069 .162 .180 .199

No. observations 72 72 72 72 72

Intelligence and knowledge highly relevant; little evidence

for interest and personality factors.

• One significant interest dimension: entrepreneurial interests (negative!)

• Variance explained is limited

• Number of participants needs to be increased

• Other forms of statistical analysis (e.g. structural equation modelling) would

allow investigating inter-construct relationships

• Influence of task complexity/ demand uncertainty on performance?

Tangible Stock/Flow Experiments −

Addressing Issues of Naturalistic

Decision Making

Jürgen Strohhecker and Andreas Größler (under review). Closer than expected −

(Missing) Differences between Tangible and Abstract Stock-Flow Task Performance,

Simulation&Gaming.

Human shortcomings in dynamic decision making

performance

Dynamic stock management largely fails!

Ample evidence from System Dynamics and psychological dynamic decision

making research:

Edwards 1962, Dörner 1980, Sterman 1989, Brehmer 1992, Dörner

1996, Paich & Sterman 1993, Moxnes 1998, Wittman & Hattrup 2004,

Moxnes & Jensen 2009

Classic example: controlling temperature (Reichert & Dörner 1988)

Period

Tem

pera

ture

in °

C

12.07.2013 .

Human deficits in understanding of accumulation (UoA)

processes

Average UoA (= understanding of stock/flow relations) is poor! Ample evidence from System Dynamics research:

Booth Sweeney and Sterman 2000, Ossimitz 2002, Sterman and Booth Sweeney 2002, 2007, Cronin and Gonzales 2007, Strohhecker 2009, Sterman 2010

Classic example: Bath tub paper/pencil task

Consider the bathtub shown below. Water flows in at a certain rate

and exits through the drain at another rate:

0

25

50

75

100

0 2 4 6 8 10 12 14 16

Flo

ws (

Litre

s p

er

Min

ute

)

Inflow

Outflow

0

50

100

150

200

0 2 4 6 8 10 12 14 16

Time (Minutes)

Wate

r in

Bath

tub

(Litre

s)

57.9 % correct

(Booth Sweeney & Sterman, 2000)

12.07.2013 .

Criticism from naturalistic decision making research

• “After all, people do manage to fill and drain their bathtubs… “ (Booth Sweeney & Sterman, 2000, p. 280)

• Proponents:

Zsambok & Klein, 1997, Lipshitz et al. 2001, Klein 2008, Lipshitz et al. 2006

• Main argument: Failures of humans to deal adequately with dynamic

complexity are apparent.

• However, these failures do not result from erroneous thinking but are artifacts

of the experimental method employed

• Our research objective: shedding light on two assumptions in this debate:

1. people are good decision makers in naturalistic situations when it comes to

stock/flow tasks and

2. performance in naturalistic and in more abstract tasks is related.

12.07.2013 .

Proposition 1:

Participants are able to achieve good performance (that is, a minor deviation

from the target in a reasonable short time span) when conducting a tangible

stock/flow experiment.

Proposition 2:

Participants are able to perform better in a tangible stock/flow experiment

than

a) in a similar simulator based experiment;

b) in paper-pencil stock/flow tasks.

Proposition 3:

Participants performing well in the tangible stock/flow task will also show a

good understanding of accumulation in a paper-pencil test and vice versa.

Our propositions

.

Research method

• Non-experimental (correlational/observational) with two observations:

1. Paper and pencil stock/flow inventory

2. Physical funnel and glass test

• Laboratory (controlled environment)

• 4th semester bachelor students

• Financial incentive of up to 20 € (linked to the performance in both tasks)

• Proposition 2a: comparison with results from literature

Instructions

Please look carefully at the following experiment setup:

Target

filling level

Beaker

Funnel

d = 3.0 mm

It is your task to pour the red liquid from the flask into the funnel so that it flows in the beaker

positioned below. You must by all means avoid that the liquid brims over the top of the funnel.

You are allowed two runs. Please achieve the following targets as best as you can:

1. Minimize the variance between the target filling level marked on the beaker and the

actual filling level (measured in milliliter)!

2. Minimize the filling time, which is measured as the time span between lifting the flask

from the table and posing it back onto the table!

For your participation in this experiment you are rewarded depending on the achievement of

these targets: The less the volume variance and the less time you need, the higher is the

monetary reward. It becomes zero if the liquid brims over the top of the funnel. The reward per

run is calculated exactly as follows (negative amounts are set equal to zero):

s

€2,0s5sTime Fillingml VarianceVolume

ml

€2,0€5wardRe

Propositions and results…

Proposition Measurement Result

Tangible SF

performance is good

Volume and time

deviation from

target/benchmark

Participants

significantly miss

target/benchmark

Tangible SF

performance is better

than abstract

Percentage deviation

from goals in tangible

task vs. in abstract

tasks

Participants do better

in abstract task than

in tangible task

Tangible and abstract

SF performance are

related

Correlation of

tangible and abstract

performance

measures

Participants’

performances are not

correlated

12.07.2013 .

Conclusions

• Naturalistic decision making proponents: people perform better in natural

situations

• Our goal: empirical investigation of this assumption

• Our (preliminary) result: Not one single of our propositions is supported by

empirical evidence

• Implication: the assumption of increased performance in natural decision

making situations does not hold for stock/flow tasks with a delay

• Two possible explanations:

1. No heuristic available “that makes us smart” regarding accumulation

processes involving a delay

2. Method shortcomings: although tangible still an artificial task in a

laboratory setting is used

Open issues

1. Classification of situations (according to structure or context?)

2. Giving information/cues in a way to support good decisions

3. Provide tools to acquire heuristics in new situations and to

support deliberate decision making