ArticlePDF Available

Effort Aversion: Job choice and compensation decisions overweight effort

Authors:

Abstract and Figures

The current research proposes that people avoid choosing effortful work even when they predict that it will provide them with a better working experience, a phenomenon we call Effort Aversion. In each of the studies, we presented a choice between an effortless but boring job and an effortful but enjoyable job. Study 1 found that participants were willing to accept lower wages to work at the effortless job, but they preferred the effortful job. This preference reversal is explained by the greater consideration wage setters gave to effort. Study 2 is a consequential lab experiment, in which participants were assigned to work at a job based on the wage they set. Those whose wage demands led them to be assigned to the effortless job experienced lower enjoyment than those who were assigned to the effortful job. Study 3 demonstrates that preference reversal was not attenuated by drawing attention to the hedonic experience afforded by work.
Content may be subject to copyright.
Journal
of
Economic
Behavior
&
Organization
92 (2013) 152–
162
Contents
lists
available
at
SciVerse
ScienceDirect
Journal
of
Economic
Behavior
&
Organization
j
ourna
l
ho
me
pa
g
e:
www.elsevier.com/locate/jebo
Effort
Aversion:
Job
choice
and
compensation
decisions
overweight
effort
David
A.
Comerforda,b,c,,
Peter
A.
Ubela
aFuqua
School
of
Business,
Duke
University,
USA
bCentre
for
Behavioural
Science,
University
of
Stirling,
UK
cSchool
of
economics,
University
College
Dublin,
Ireland
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
7
August
2012
Received
in
revised
form
22
February
2013
Accepted
30
May
2013
Available online 13 June 2013
Keywords:
Job
choice
Preference
reversal
Effort
Aversion
a
b
s
t
r
a
c
t
The
current
research
proposes
that
people
avoid
choosing
effortful
work
even
when
they
predict
that
it
will
provide
them
with
a
better
working
experience,
a
phenomenon
we
call
Effort
Aversion.
In
each
of
the
studies,
we
presented
a
choice
between
an
effortless
but
boring
job
and
an
effortful
but
enjoyable
job.
Study
1
found
that
participants
were
willing
to
accept
lower
wages
to
work
at
the
effortless
job,
but
they
preferred
the
effortful
job.
This
preference
reversal
is
explained
by
the
greater
consideration
wage
setters
gave
to
effort.
Study
2
is
a
consequential
lab
experiment,
in
which
participants
were
assigned
to
work
at
a
job
based
on
the
wage
they
set.
Those
whose
wage
demands
led
them
to
be
assigned
to
the
effortless
job
experienced
lower
enjoyment
than
those
who
were
assigned
to
the
effortful
job.
Study
3
demonstrates
that
preference
reversal
was
not
attenuated
by
drawing
attention
to
the
hedonic
experience
afforded
by
work.
© 2013 Elsevier B.V. All rights reserved.
1.
Introduction
In
September
2010,
attendants
at
the
Irish
National
Gallery
issued
notice
that
they
would
strike.
Their
grievance
was
unconventional.
They
were
seeking
more
effortful
work
‘there’s
nothing
to
do
but
walk
up
and
down
and
watch
visitors’,
complained
one
(Carey,
2010).
The
current
research
tests
experimentally
whether
there
is
a
general
tendency
amongst
workers
to
choose
a
job
that
leads
them
to
suffer
underemployment.
Imagine
that
employees
at
the
Irish
National
Gallery
had
a
choice
between
the
low
effort
but
boring
job
of
attendant
and
the
more
effortful
and
more
engaging
job
of
usher.
In
classical
economic
theory,
a
worker
will
accept
lower
wages
to
work
at
whichever
job
is
anticipated
to
grant
the
greatest
utility.
This
prediction
holds
even
if
that
job
requires
the
worker
to
contribute
more
effort
than
does
the
alternative
effort
is
just
another
factor
in
the
utility
function
and
it
should
influence
preference
only
to
the
extent
that
it
matters
to
utility.
In
contrast
to
classical
theory,
our
theory
of
Effort
Aversion
posits
that
the
expenditure
of
effort
is
valued
exceptionally.
Following
previous
research
(e.g.
Akerlof
and
Yellen,
1990;
Amir
et
al.,
2008),
we
propose
that
when
choosing
a
job,
workers
seek
monetary
compensation
for
effort,
even
if
the
expenditure
of
effort
increases
the
worker’s
own
utility.
The
theory
predicts
that
workers
may
choose
a
less
effortful
job
even
if
they
anticipate
experiencing
greater
utility
from
a
more
effortful
alternative.
We
thank
Andrea
Angott,
Dan
Ariely,
Liam
Delaney,
George
Loewenstein
and
two
anonymous
reviewers
for
helpful
comments
and
Anna
Weinberg
for
excellent
research
assistance.
Comerford
gratefully
acknowledges
funding
for
this
research
by
the
Irish
Research
Council
for
the
Humanities
and
Social
Sciences
and
the
Fulbright
Commission.
Corresponding
author
at:
School
of
Management,
University
of
Stirling,
FK9
4LA,
UK.
Tel.:
+44
1786467317.
E-mail
address:
comerfod@gmail.com
(D.A.
Comerford).
0167-2681/$
see
front
matter ©
2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jebo.2013.05.016
D.A.
Comerford,
P.A.
Ubel
/
Journal
of
Economic
Behavior
&
Organization
92 (2013) 152–
162 153
To
test
this
theory,
we
conducted
a
consequential
lab
experiment
that
offered
participants
a
choice
between
an
effortless
boring
job
and
an
effortful
engaging
job.
We
found
that
over
a
fifth
of
participants
both
set
wages
at
a
level
that
implied
choice
of
the
effortless
job
and
also
predicted
that
they
would
enjoy
it
less
than
the
effortful
job.
Participants
were
assigned
to
a
job
on
the
basis
of
the
wage
they
set
those
who
set
wages
at
a
level
that
implied
choice
of
the
effortless
boring
job
were
assigned
to
work
at
it.
In
line
with
the
prediction
of
Effort
Aversion,
workers
at
the
effortless
boring
job
reported
experiencing
less
enjoyment
than
did
those
performing
the
effortful
engaging
job.
The
term
we
propose
for
this
tendency
to
reject
a
job
because
it
is
effortful
is
Effort
Aversion.
The
paper
opens
by
presenting
a
model
that
predicts
the
circumstances
under
which
workers
are
likely
to
overweight
compensation
for
effort
at
the
expense
of
non-monetary
benefits.
We
build
on
Fehr
and
Schmidt’s
model
of
the
demand
for
fairness
in
wage
markets,
which
decomposes
the
utility
engendered
by
a
job
into
two
parts
(1999):
1.
The
utility
derived
from
all
the
substantive
features
of
the
job:
This
component
encompasses
all
the
factors
recognized
in
conventional
utility
functions
the
satisfaction
engendered
by
the
work;
the
purchasing
power
of
the
wages,
etc.
2.
The
utility
derived
from
the
relative
wage
i.e.
how
the
offered
wage
compares
to
a
‘normal’
wage
for
the
job:
A
worker’s
knowledge
of
market
norms
and
how
much
similar
jobs
pay
informs
her
sense
of
how
much
she
should
get
paid.
Independent
of
the
purchasing
power
of
the
wage,
it
hurts
to
feel
that
the
offered
wage
is
lower
than
normal.
The
utility
lost
from
perceiving
one’s
effort
to
be
inadequately
compensated
falls
into
the
relative
wage
component.
In
Fehr
and
Schmidt’s
model
the
degree
to
which
relative
wage
matters
to
utility
is
dictated
by
a
weighting
function,
˛.
The
key
theoretical
contribution
of
our
research
is
to
show
that
˛
varies
systematically
such
that
relative
wage
receives
greater
weight
in
choice
than
in
evaluations
made
while
working.
In
the
words
of
Daniel
Kahneman,
‘nothing
in
life
is
as
important
as
you
think
it
is
when
you
are
thinking
about
it’
(Kahneman,
2011,
p.
402).
Our
model
predicts
a
finding
suggested
by
the
previous
literature:
that
people
care
about
how
the
wage
conforms
to
market
norms
more
when
they
are
choosing
a
job
than
they
do
when
working
at
it.
We
now
review
some
findings
from
the
literature
on
preference
reversals
that
lead
us
to
predict
that
relative
wages
matter
more
in
choice
than
in
experience:
1.
Choice
is
made
in
a
Joint
Evaluation
context
but
experience
is
based
on
Single
Evaluation:
By
its
nature,
job
choice
is
made
in
a
context
with
a
salient
alternative
we
can
either
accept
a
job
that
is
offered
or
continue
to
spend
our
time
on
the
status
quo
occupation.
Thus
job
choice
occurs
in
a
context
where
the
attributes
of
each
alternative
can
be
readily
compared
relative
to
one
another
what
Hsee
et
al.
(2012)
refer
to
as
a
joint
evaluation
context
(JE).
By
contrast,
a
job
is
experienced
in
a
single
evaluation
context
(SE)
we
can
only
work
one
job
at
a
time.
Since
it
is
easier
to
compare
wages
in
JE
than
in
SE,
appraisals
made
at
the
moment
of
choice
will
be
more
sensitive
to
relative
wage
than
they
will
at
the
moment
of
experience.
2.
Construal
level
is
different
in
choice
and
experience:
Construal
level
theory
posits
that
evaluations
made
at
a
temporal
distance
place
greater
weight
on
high-level
attributes,
whereas
those
made
in
the
moment
of
experience
place
greater
weight
on
low-level
attributes
(Trope
and
Liberman,
2010).
High-level
attributes
are
abstract
and
conceptual
(e.g.
being
treated
justly).
Low-level
attributes
are
concrete
(e.g.
being
physically
comfortable).
Previous
research
shows
that
the
concern
accorded
distributive
justice
in
the
workplace
(e.g.
receiving
fair
compensation)
increases
with
temporal
distance
(Cojuharenco
et
al.,
2011).
To
the
extent
that
job
choice
occurs
at
a
temporal
remove
from
actually
working,
the
high-level
concept
of
fair
compensation
will
be
weighted
more
heavily
in
choice
than
in
experiential
evaluations.
3.
Rules
apply
to
choice:
Making
a
choice
can
require
time
and
cognitive
effort.
An
adaptive
means
to
cut
the
costs
of
decision
making
is
to
adopt
a
rule.
By
their
nature,
rules
are
inflexible
and
so,
given
specific
circumstances,
they
can
cause
people
to
choose
an
inferior
option
over
a
superior
alternative.
Inconsistencies
across
predicted
utility
and
choice
have
been
attributed
to
the
application
of
rules
such
as
“waste
not”
(Arkes
and
Blumer,
1985),
“seek
variety”
(Simonson,
1990),
and
“don’t
pay
for
a
delay”
(Amir
and
Ariely,
2007).
Following
Akerlof
and
Yellen’s
fair-wage
hypothesis
(1990),
we
propose
that
there
is
a
rule
of
thumb
to
demand
higher
wages
for
additional
expenditure
of
effort.
That
rule
could
lead
choice
astray
in
the
case
where
a
worker
has
the
option
to
work
at
an
enjoyable
task
that
requires
additional
expenditure
of
effort.
In
that
case,
the
worker
would
experience
greater
utility
by
accepting
the
more
effortful
job,
but
the
rule
would
prompt
them
to
reject
it.
Any
one
of
the
three
mechanisms
summarized
above
would
lead
compensation
to
receive
greater
weight
in
the
moment
of
job
choice
than
it
does
in
the
moment
of
experience.
It
is
important
to
note
that
there
is
more
than
one
procedure
by
which
workers
can
choose
to
work
at
a
job,
and
this
too
will
impact
the
weight
accorded
compensation.
Some
workers
look
through
a
menu
of
jobs
(on
a
website
or
in
a
newspaper)
and
selectively
apply
for
the
ones
they
prefer.
Other
workers
choose
their
job
by
negotiating
wages
e.g.
plumbers
receive
a
phone
call
asking
for
a
quote
and
decide
on
a
figure
that
will
both
compensate
their
costs
and
appeal
to
the
caller.
A
further
prediction
of
the
current
research
is
that
the
weighting
attached
to
the
relative
wage
will
differ
across
these
two
choice
procedures.
This
prediction
is
informed
by
a
series
of
preference
reversals
reported
in
Amir
et
al.
(2008)
and
in
Boothe
et
al.
(2007).
Boothe
et
al.
attribute
these
reversals
to
the
Market
Value
Heuristic.
They
presented
participants
with
a
good
154 D.A.
Comerford,
P.A.
Ubel
/
Journal
of
Economic
Behavior
&
Organization
92 (2013) 152–
162
that
was
high
in
hedonic
value,
but
low
in
market
value
(e.g.
a
can
of
soda)
and
a
good
that
was
low
on
hedonic
value,
but
high
in
market
value
(e.g.
an
eggplant
roulette).
Some
participants
chose
between
these
two,
whereas
others
assigned
minimum
selling
prices
to
them.
Preference
differed
systematically
across
procedures
such
that
respondents
were
more
likely
to
choose
the
high
hedonic
value
good
than
they
were
to
assign
it
a
higher
minimum
selling
price.
The
market
value
heuristic
suggests
that
people
will
place
greater
weight
on
compensation
for
effort
when
setting
wages
than
they
do
when
choosing
from
a
menu
of
jobs.
In
Section
2
we
present
the
amended
model
of
job
utility
in
which
the
relative
wage
parameter,
˛,
is
weighted
differently
across
procedures.
We
propose
a
pathway
by
which
˛
is
reweighted
and
how
this
in
turn
leads
to
preference
reversals.
The
proposed
pathway
is
designed
to
encompass
the
various
sources
of
preference
reversal
that
were
discussed
in
the
preceding
paragraphs.
It
is
intended
to
make
meaningful
predictions
about
when
Effort
Aversion
will
occur
and
under
what
conditions
it
will
be
exacerbated.
We
test
the
predictions
of
the
model
in
Sections
3–5.
Section
3
reports
Study
1,
which
shows
the
preference
reversal
across
choice
procedures,
and
shows
that
this
is
mediated
by
a
focus
on
salient
attributes.
In
this
stated
preference
study,
participants
were
told
that
they
had
been
offered
an
effortful,
engaging
job
and
an
effortless,
boring
one.
Some
respondents
were
asked
to
set
the
minimum
wage
they
would
accept
for
the
effortful
job.
Others
were
asked
which
job
they
would
prefer
if
both
jobs
paid
the
same
amount.
Immediately
afterwards,
respondents
reported
what
factors
they
considered
when
responding.
Respondents
who
answered
the
preference
question
were
more
likely
to
imply
preference
for
the
effortful,
engaging
job
than
were
those
who
set
wages.
They
were
also
more
likely
to
explain
their
choice
with
reference
to
the
hedonic
quality
of
the
working
experience.
Those
who
set
wages
were
more
likely
to
refer
to
the
effort
the
jobs
require,
and
to
subsequently
imply
preference
for
the
effortless
job.
Mediation
analysis
supports
the
proposed
pathway,
showing
that
job
choice
is
explained
by
consideration
of
effort.
Section
4
reports
Study
2,
a
consequential
lab
experiment
that
documents
the
preference
reversal
across
choice
and
wage-setting
in
an
incentive-compatible
setting.
In
the
study,
we
asked
MBA
students
to
act
in
a
film
for
5
min.
They
had
the
option
of
playing
the
role
of
a
Worker
engaged
in
a
mildly
challenging
puzzle
or
an
Onlooker,
who
would
sit
passively
and
watch
the
Workers.
In
a
within-sample
design,
participants
set
a
reservation
wage
and
made
an
explicit
choice
between
the
two
jobs.
Again,
participants
were
more
likely
to
set
wages
at
a
level
that
implied
choice
of
the
effortless
job
than
they
were
to
explicitly
choose
it.
Consistent
with
Effort
Aversion,
participants
who
set
wages
at
a
level
that
assigned
them
to
work
at
the
effortless
job
reported
having
experienced
lower
levels
of
enjoyment
while
working
than
did
those
who
worked
at
the
effortful
job.
Moreover,
many
of
them
predicted
that
the
effortful
engaging
job
would
be
more
enjoyable.
In
Section
5
we
test
a
low
cost
intervention
to
overcome
Effort
Aversion.
Study
3
reports
the
results
of
a
manipulation
whereby
we
asked
some
participants
to
predict
the
enjoyment
of
the
jobs
before
choosing
between
them.
This
enjoyment
question
should
draw
attention
to
the
experiential
content
of
the
jobs.
We
show
in
a
stated
preference
study
that
the
manipulation
successfully
attenuates
the
preference
reversal,
but
find
that
it
has
no
effect
in
the
consequential
lab
study.
We
conclude
that
Effort
Aversion
cannot
be
overcome
simply
by
drawing
attention
to
the
experiential
qualities
of
a
job.
The
final
sections
of
the
paper
discuss
the
ecological
validity
of
the
experiments
presented,
and
consider
the
implications
of
Effort
Aversion
for
labor
markets.
2.
Model
Relative
wage
is
the
difference
between
the
actual
wage
a
job
offers
and
the
wage
that
the
worker
perceives
as
fair
or
reasonable
for
that
job.
Relative
wage
matters
for
job
preference,
independent
of
absolute
wages
(Kahneman
et
al.,
1986;
Bewley,
1999;
Fehr
and
Schmidt,
1999).
The
utility
a
worker
i
perceives
a
job
to
deliver
can
be
decomposed
into
two
discrete
sources
the
utility
engendered
by
its
relative
wage
and
the
utility
engendered
by
all
its
other
features.
The
dominant
theories
of
fairness
consider
the
weight
placed
on
relative
wage
to
be
a
stable
individual
preference
(e.g.
Fehr
and
Schmidt,
1999).
Our
contention,
based
on
the
literature
reviewed
in
the
previous
section,
is
that
the
weight
placed
on
the
relative
wage
varies
systematically
across
preference
elicitation
procedures.
The
model
we
propose
looks
like
this:
uip(job)
=
(1
˛ip)(utility
from
substantive
features
of
job)
+
˛ip(relative
wage)
The
˛
term
in
our
model
is
systematically
related
to
the
preference
elicitation
procedure,
hence
the
subscript
p.
Neces-
sarily,
if
the
worker
places
greater
weight
on
the
relative
wage
component
of
the
utility
function,
they
place
correspondingly
less
weight
on
the
other
features;
hence
the
term
1
˛ip is
the
weighting
function
on
the
utility
derived
from
the
substantive
features
of
the
job.
On
the
basis
of
the
preference
reversals
discussed
in
the
previous
section,
we
propose
the
following:
˛p=wage
setting >
˛p=choice >
˛p=experienced
utility
rating
The
hypothesis
that
compensation
for
effort
will
be
weighted
more
heavily
in
wage
setting
than
in
choice
stems
from
the
Market
Value
Heuristic,
which
posits
that
market
norms
are
especially
salient
in
monetary
evaluations
(e.g.
Boothe
et
al.,
2007;
Amir
et
al.,
2008).
The
prediction
that
relative
wage
is
likely
to
be
weighted
more
heavily
in
choice
than
in
experience
stems
from
findings
on
rules
(e.g.
Amir
and
Ariely,
2007),
on
construal
level
(e.g.
Cojuharenco
et
al.,
2011),
and
on
differences
in
attribute
weighting
across
joint
and
single
evaluation
modes
(e.g.
Hsee
et
al.,
2012).
D.A.
Comerford,
P.A.
Ubel
/
Journal
of
Economic
Behavior
&
Organization
92 (2013) 152–
162 155
Enjoyment
rang
Choice from
menu
Reservao
n
wage
seng
To enjoy o
neself
Various
as det
ermine
d b
y
chronic in
dividual
diff
eren
ces
& contextual
cues
To be
compe
nsated
accordi
ng to
market norms
Hed
onic, e
xperie
nal
aributes
Effort
Sali
ent
aribute
Revealed
prefe
rence
Efforu
l, engag
ing work
Effortl
ess,
boring work
Fig.
1.
Potential
pathways
from
elicitation
procedure
to
revealed
preference.
Fig.
1
illustrates
a
pathway
through
which
the
preference
elicitation
procedure
could
influence
choice.
The
pathway
encompasses
the
various
mechanisms
that
were
discussed
in
Section
1.
It
allows
that
the
elicitation
procedure
might
activate
a
rule
or
goal,
following
a
stream
of
the
literature
that
suggests
this
process
(e.g.
Fischer
et
al.,
1999;
Suk
and
Yoon,
2012).
It
also
allows
that
elicitation
procedures
might
have
a
direct
effect
on
the
allocation
of
attention
in
choice,
following
a
literature
that
predicts
preference
reversals
resulting
from
the
efficient
allocation
of
attention
(e.g.
Kozsegi
and
Szeidl,
2013;
Sims,
2003;
Woodford,
2012).
Fig.
1
presents
examples
of
some
processes
by
which
respondents
might
respond
to
different
preference
elicitation
procedures.
When
asked
to
rate
the
enjoyment
of
working
at
a
job,
respondents
might
ask
themselves
“how
does
this
option
advance
me
toward
the
goal
of
experiencing
enjoyment?”
With
that
criterion
in
mind,
they
will
more
heavily
weight
attributes
that
are
diagnostic
of
the
hedonic
experience.
Given
the
alternatives
of
an
engaging
job
that
requires
effort
and
a
boring
job
that
requires
no
expenditure
of
effort,
a
respondent
is
likely
to
rate
the
engaging
job
as
more
enjoyable.
In
contrast
with
the
procedure
of
enjoyment
rating,
setting
a
reservation
wage
is
unlikely
to
prompt
the
goal
of
expe-
riencing
enjoyment.
Amir
et
al.
propose
that
“monetary
assessment
tends
to
focus
on
the
transaction
(e.g.
how
fair
the
transaction
seems)”
(2008,
p.
1055).
The
goal
of
the
wage
setter
is
more
likely
then
to
be
something
like
“to
receive
adequate
compensation”.
Attributes
that
will
be
weighted
heavily
are
those
that
relate
to
market
norms.
Market
norms
of
fairness
and
reciprocity
suggest
that
greater
effort
should
be
compensated
with
higher
wages
(Akerlof
and
Yellen,
1990).
When
told
that
a
job
that
requires
no
expenditure
of
effort
pays
$x,
a
worker
will
demand
a
wage
of
at
least
$x
in
order
to
work
at
a
job
that
requires
effort.
Even
if
the
effortful
job
is
engaging
and
the
effortless
job
is
boring,
we
predict
that
the
wage
setter
will
be
reluctant
to
accept
a
lower
wage
for
the
more
effortful
work.
In
summary,
˛
is
likely
to
be
granted
substantial
weight
by
the
procedure
of
wage
setting.
When
asked:
If
the
two
jobs
paid
the
same
amount,
which
would
you
choose?
there
are
a
number
of
different
criteria
that
might
come
to
mind.
Some
people
might
spontaneously
consider
the
effort
the
two
jobs
require
and
choose
the
easier
of
them,
but
others
might
consider
how
enjoyable
each
of
the
jobs
is,
or
how
proud
they
would
feel
telling
their
brother-in-law
about
the
job.
Crucially,
norms
of
compensation
are
more
likely
to
be
considered
when
setting
wages
than
they
are
by
other
procedures,
and
hence
we
predict
that
(a)
workers
will
be
most
likely
to
choose
an
effortless
and
boring
job
by
the
procedure
of
wage
setting;
(b)
they
will
be
less
likely
to
choose
it
from
a
menu
of
options;
and
(c)
they
will
be
least
likely
to
rate
it
as
the
most
enjoyable
job.
3.
A
stated
preference
study
to
test
for
Effort
Aversion
Our
first
study
tests
whether
preference
elicitation
procedures
affect
the
likelihood
that
a
job
is
preferred
and
assesses
the
role
of
Effort
Aversion
in
this
process.
We
presented
respondents
with
a
hypothetical
choice
between
an
effortless,
boring
job
and
an
effortful,
engaging
job.
We
asked
half
the
sample
to
state
their
reservation
wage
and
to
explain
their
response
and
the
other
half
was
asked
‘which
job
do
you
prefer?’
and
to
explain
their
response.
Our
first
hypothesis
is
that
a
higher
proportion
of
workers
would
state
a
preference
for
the
effortful,
engaging
job
than
would
imply
preference
for
it
by
wage
setting.
The
scenario
was
designed
such
that
the
only
thing
that
discriminates
between
the
two
jobs
is
the
activities
they
entail.
When
explaining
their
response,
respondents
in
the
preference
condition
are
predicted
to
refer
to
the
pleasures
and
pains
156 D.A.
Comerford,
P.A.
Ubel
/
Journal
of
Economic
Behavior
&
Organization
92 (2013) 152–
162
of
those
activities,
which
would
manifest
itself
through
explanations
that
describe
the
hedonic
quality
of
working
e.g.
‘enjoyable’,
‘boring’,
‘interesting’,
etc.
Our
second
hypothesis
is
that
wage
setters
will
be
more
likely
than
those
in
the
preference
condition
to
explain
their
response
with
reference
to
effort
than
will
those
stating
preference.
Our
theory
predicts
that
preference
varies
with
the
weight
placed
on
the
relative
wage.
A
useful
feature
of
the
open-ended
explanations
is
that
they
reveal
what
criteria
are
considered
and
which
are
decisive
in
shifting
preference.
The
influence
that
a
criterion
has
in
job
selection
can
be
measured
by
coding
a
variable
to
indicate
reference
to
it
in
the
open-ended
explanation
and
then
regressing
job
choice
on
that
indicator.
Thus,
we
can
test
whether
a
preference
reversal
is
explained
by
consideration
of
effort.
Our
third
hypothesis
is
that
any
difference
in
preference
across
conditions
will
be
explained
by
wage
setters’
higher
propensity
to
consider
effort.
3.1.
Methods
We
asked
the
respondents
to
imagine
they
had
been
offered
a
choice
of
two
short-term
jobs:
working
as
an
invigilator
in
an
art
gallery
at
a
cultural
festival;
or
working
as
an
usher
at
the
same
festival.
The
usher
will
‘publicize
events,
escort
performers
to
their
venues
and
clean
venues
after
performances’.
As
an
invigilator
the
‘only
task
you
are
expected
to
perform
is
to
alert
a
security
guard
in
the
highly
unlikely
event
that
one
is
required.
.
.
Activities
such
as
reading,
listening
to
music
or
using
a
mobile
phone
are
strictly
prohibited
while
on
duty’.
Respondents
were
explicitly
told
that
all
other
features
(hours,
contract
duration,
etc.)
were
identical
across
the
two
jobs.
One
version
of
the
questionnaire
asked:
“The
job
of
invigilator
pays
D
1000
for
2
weeks.
What
is
the
minimum
amount
you
would
accept
to
work
as
an
usher
for
2
weeks?”
The
other
asked:
“Assuming
the
two
jobs
have
the
same
conditions
(e.g.
pay,
breaks,
etc.),
which
would
you
prefer?”
3.2.
Hypotheses
H1.
Respondents
in
the
wage
setting
condition
will
be
more
likely
to
imply
preference
for
the
effortful
job
of
usher
than
will
respondents
in
the
preference
condition.
H2.
Explanations
of
preference
will
be
more
likely
to
refer
to
the
hedonic
quality
of
the
job.
Explanation
of
wages
will
be
more
likely
to
refer
to
effort.
H3.
Reference
to
the
effort
the
job
requires
will
predict
job
selection
and
mediate
the
difference
in
preference
elicited
across
procedures.
3.3.
Procedures
149
respondents
were
recruited
in
public
spaces
on
the
campus
of
University
College
Dublin
and
asked
to
complete
a
brief
pen
and
paper
questionnaire
about
summer
jobs.
No
compensation
was
offered.
Questionnaires
were
alternated
and
placed
in
an
opaque
bag
for
distribution.
5
responses
were
dropped
as
incomplete.
We
coded
response
to
the
open-ended
question
with
dummy
variables:
‘con-
sidered
effort’
if
effort
or
hard
work
is
mentioned
by
the
respondent;
‘considered
hedonic
quality’
if
the
explanation
referred
to
how
interesting,
boring,
enjoyable,
etc.
the
work
would
be;
and
‘considered
career
prospects’
if
the
respondent
referred
to
on-the-job
learning;
value
for
resumé,
etc.
3.4.
Results
In
line
with
Hypothesis
1,
there
is
a
significant
difference
in
preference
across
procedures.
82%
of
respondents
stated
that
they
preferred
the
effortful
job
of
ushering.
62.5%
of
wage
setters
set
the
minimum
wage
they
would
accept
for
ushering
as
equal
to
or
less
than
the
wage
offered
to
invigilators.
A
Probit
estimation
shows
this
difference
in
preference
across
procedures
to
be
significant
(z(144)
=
2.60,
p
<
.01).
39%
of
respondents
in
the
wage
setting
condition
referred
to
‘effort’
and
‘hard
work’
in
their
explanation,
whereas
only
18%
of
those
who
answered
the
preference
question
did,
z(144)
=
2.76,
p
<
.01.
88%
of
explanations
of
preference
referred
to
enjoyment,
whereas
just
41%
of
explanations
of
wage
did,
z(144)
=
5.51,
p
<
.01.
As
intended
by
the
design
of
the
scenario,
very
few
respondents
reported
that
they
considered
career
prospects
in
selecting
amongst
these
jobs.
Only
one
respondent
a
student
of
art
history
referred
to
it.
Consistent
with
Hypothesis
3,
the
effect
of
elicitation
procedure
on
choice
is
mediated
by
consideration
of
effort.
Table
1
shows
that
elicitation
procedure
is
a
significant
predictor
of
job
selection
(model
1),
until
consideration
of
the
effort
that
a
job
requires
is
included
in
the
model
(model
2).
D.A.
Comerford,
P.A.
Ubel
/
Journal
of
Economic
Behavior
&
Organization
92 (2013) 152–
162 157
Table
1
Marginal
effects
on
the
probability
of
selecting
the
effortless
job.
Model
1
Model
2
Selection
procedure:
wage
setting
(base
category
=
stated
preference)
.194*** (.073)
.055
(.086)
Considered
effort
.532*** (.091)
Considered
hedonic
quality
.147
(.099)
Constant
.269
.226
*** p
<
.01.
3.5.
Discussion
The
mediation
analysis
in
model
2
of
Table
1
shows
that
the
attention
paid
to
effort
is
a
decisive
factor
in
explaining
differences
in
preference
across
procedures,
demonstrating
a
link
in
the
causal
chain
from
procedure
to
elicited
preference.
In
the
next
section
we
demonstrate
that
this
preference
reversal
has
meaningful
consequences.
4.
Consequential
lab
study
This
consequential
lab
study
was
designed
to
test
whether
compensation
for
effort
is
overweighted
in
choice
at
the
expense
of
non-monetary
benefits.
The
subject
pool
for
this
study
was
102
MBA
students
who
were
invited
to
participate
as
actors
in
the
shooting
of
a
video
for
an
experiment.
Those
who
agreed
to
take
part
had
to
choose
between
playing
the
role
of
an
Onlooker
or
a
Worker
for
5
min.
The
choice
scenario
was
designed
so
that
the
only
discriminating
feature
between
the
two
roles
was
the
activity
required
of
the
participant
Onlookers
were
required
to
do
nothing;
workers
were
required
to
do
an
engaging
puzzle.
We
hypothesized
that
(1)
participants
would
be
more
likely
to
imply
choice
of
the
effortless
job
by
the
procedure
of
wage
setting
than
they
would
be
to
favor
it
by
the
other
elicitation
procedures;
and
that
(2)
those
whose
wage
demands
assigned
them
to
work
at
the
effortless
job
would
enjoy
the
time
spent
filming
less
than
those
who
worked
at
the
effortful
job.
4.1.
Methods
Before
a
break
in
a
two-hour
class,
all
students
were
told
that
they
had
the
opportunity
to
participate
in
the
shooting
of
a
5
min
film
and
given
an
informed
consent
document
which
they
could
sign
to
indicate
their
willingness
to
take
part.
The
filming
would
take
place
over
5
min
at
the
close
of
the
break
and
those
who
did
not
sign
the
informed
consent
were
not
permitted
to
be
in
the
room
at
the
time
of
filming.
The
informed
consent
document
described
the
roles.
Onlookers
were
required
to
do
‘nothing
other
than
sit
silently
and
passively
for
5
min’
and
Workers
would
spend
the
5
min
completing
a
set
of
pen
and
paper
puzzles.
On
the
informed
consent
document,
participants
could
see
an
example
of
the
sort
of
task
that
Workers
would
receive1and
were
told
that
Onlookers
would
not
be
permitted
to
talk,
eat,
read
or
use
phones
or
laptops
during
the
5
min
of
filming.
Having
read
these
descriptions,
respondents
were
asked
to
‘answer
the
following
questions
so
we
can
efficiently
allocate
people
to
roles’:
A.
Onlookers
will
receive
a
$2.50
Amazon
gift
card.
What
is
the
minimum
value
gift
card
you
would
accept
to
be
a
Worker
instead?
B.
If
the
two
jobs
paid
the
same
amount,
which
would
you
choose?
C.i.
After
4m30s
spent
as
an
Onlooker,
how
much
do
you
think
you
would
be
enjoying
yourself,
on
a
scale
from
0
to
5?
C.ii.
After
4m30s
spent
as
a
Worker,
how
much
do
you
think
you
would
be
enjoying
yourself,
on
a
scale
from
0
to
5?
After
filming
was
completed,
participants
rated
how
much
they
had
been
enjoying
themselves
4m30s
into
the
film.
These
questions
yield
the
measures
that
will
be
used
to
test
the
hypotheses.
We
asked
about
enjoyment
after
4m30s
so
that
the
before
and
after
measures
of
enjoyment
would
compare
like
with
like.2A
pretest
allays
any
concerns
that
the
enjoyment
predicted
to
occur
at
4m30s
is
not
representative
of
the
enjoyment
experienced
overall.
It
asked
55
people
at
Chicago
O’Hare
airport
to
predict
the
enjoyment
of
both
jobs
after
4m30s
and
then
to
state
which
of
the
two
jobs
they
thought
would
be
more
enjoyable
to
perform.
For
52
of
them
(94%)
the
more
enjoyable
job
was
also
the
one
to
which
they
gave
a
higher
enjoyment
rating.
1A
scrambled
sentence
task
in
which
a
4-word
sentence
is
formed
from
5
words
e.g.
eagle
apple
majestic
the
soars
=
the
majestic
eagle
soars.
2A
concrete
landmark
should
help
attenuate
the
Peak-End
effect,
a
memory
bias
that
focuses
attention
on
the
peak
and
end
of
an
experience
(Kahneman
et
al.,
1993).
Also
there
is
evidence
that
people
do
not
spontaneously
consider
adaptation
i.e.
hedonic
response
to
an
experience
might
not
be
well
summarized
by
initial
response
to
it
(e.g.
Ubel
et
al.,
2005).
4m30s
seemed
a
good
landmark
to
draw
attention
to
the
possibility
of
adaptation.
158 D.A.
Comerford,
P.A.
Ubel
/
Journal
of
Economic
Behavior
&
Organization
92 (2013) 152–
162
0
10
20
30
40
50
60
70
80
Enjoy Ch
oose
Wage
Onlo
oker
Indifferent
Worker
Fig.
2.
Number
of
participants
indicting
preference
for
each
job,
by
elicitation
procedure.
Note:
The
column
on
the
left
shows
the
number
who
rated
each
job
more
enjoyable.
The
center
column
shows
the
number
who
chose
each
job
when
they
were
assumed
to
pay
the
same
amount.
The
column
on
the
right
shows
preference
for
each
job,
as
indicated
by
whether
the
participant
would
be
willing
to
accept
lower
wages
to
work
at
it.
4.2.
Procedures
Immediately
above
the
preference
elicitation
questions
on
the
informed
consent
form
was
written
in
bold
font:
“please
answer
the
following
questions
so
we
can
efficiently
allocate
people
to
roles”.
Participants
were
assigned
to
be
Workers
unless
they
either
answered
that
they
would
choose
to
be
an
Onlooker
or
sought
a
higher
wage
for
being
a
worker
than
was
offered
for
being
an
Onlooker.
74
of
the
102
MBA
students
consented
to
participate
in
the
film.
7
of
them
were
dropped
from
the
sample
because
they
did
not
answer
all
the
questions
required
for
analysis.
On
the
basis
of
the
complete
responses,
46
participants
were
assigned
to
the
role
of
Onlooker,
and
21
participants
were
assigned
to
the
role
of
Worker.
Non-participants
took
an
extended
break
so
as
not
to
disturb
filming.
4.3.
Results
Categorical
measures
were
constructed
for
each
of
the
preference
elicitation
procedures
to
indicate
preference
for
Onlooker,
preference
for
Worker,
or
indifference.
A
respondent
could
indicate
indifference
by
giving
both
tasks
the
same
rating
on
the
enjoyment
rating
scale;
by
seeking
exactly
$2.50
to
be
a
Worker;
or
by
ticking
the
‘no
preference’
box
when
answering
the
choice
question.
A
related
samples
McNemar
test
was
used
to
test
for
differences
within
sample
in
the
probability
of
preferring
a
job
across
procedures.
36
participants
stated
that
they
would
require
a
gift
card
worth
more
than
$2.50
to
accept
the
role
of
Worker.
12
of
those
contradicted
that
answer
by
stating
that
they
would
choose
to
be
a
Worker
if
both
jobs
paid
$2.50,
Chi
squared
(66)
=
12.07,
p
<
0.01.
This
pattern
of
results
supports
Hypothesis
1
(see
Fig.
2).
The
participants
who
sought
higher
wages
to
be
a
Worker
than
an
Onlooker
were
all
assigned
to
work
as
Onlookers.
After
the
job,
participants
rated
how
much
they
had
been
enjoying
themselves
4m30s
into
the
task,
on
a
scale
from
0
(not
at
all)
to
5
(very
much).
Consistent
with
Hypothesis
2,
the
mean
enjoyment
rating
reported
by
Workers
was
higher
than
that
reported
by
Onlookers
(Mworker =
3.8,
Monlooker =
3,
z
=
2.75,
p
<
01).
4.4.
Discussion
This
lab
study
replicates
the
preference
reversal
across
choice
procedures
in
an
incentive-compatible
setting:
19%
of
participants
both
demanded
a
wage
in
excess
of
$2.50
for
the
job
of
Worker,
and
contradicted
that
by
stating
that
they
would
choose
it
if
both
jobs
paid
$2.50.
In
addition,
Onlookers
experienced
less
enjoyment
than
did
Workers.
If
participants
were
to
maximize
the
utility
derived
from
the
substantive
features
of
the
job,
they
would
choose
whichever
one
they
predicted
to
be
more
enjoyable,
since
the
content
of
the
task
was
the
only
thing
that
differed
across
jobs.3Clearly
a
large
proportion
of
the
participants
did
not
maximize
the
utility
derived
from
the
substantive
features
of
the
work:
22%
of
participants
set
wages
at
a
level
that
implied
3A
follow
up
confirmed
that
participants
perceived
no
feature
that
differed
across
the
two
jobs
aside
from
the
content
of
the
work.
We
had
the
opportunity
to
address
the
participants
of
the
study
during
their
class
three
weeks
after
the
study
took
place.
We
reminded
them
of
the
study
and
the
two
roles
and
D.A.
Comerford,
P.A.
Ubel
/
Journal
of
Economic
Behavior
&
Organization
92 (2013) 152–
162 159
choice
of
the
job
that
they
predicted
to
be
less
enjoyable.
In
every
one
of
those
cases,
participants
predicted
that
the
effortful
job
of
Worker
would
be
more
enjoyable,
but
chose
the
effortless
job
of
Onlooker.
This
pattern
is
consistent
with
Effort
Aversion:
participants
avoided
choosing
effortful
work
even
when
they
predicted
that
it
would
provide
them
with
a
better
experience.
5.
Can
Effort
Aversion
be
easily
overcome?
In
the
previous
study,
people
experienced
worse
working
conditions
because
they
placed
value
on
receiving
compensation
for
effort
expended.
A
low
cost
intervention
to
overcome
Effort
Aversion
could
generate
welfare
gains:
workers
would
derive
greater
benefit
from
the
experience
of
working,
and
productive
resources
could
be
allocated
more
efficiently.
In
Section
1
we
alluded
to
various
potential
cognitive
mechanisms
that
might
explain
the
preference
reversals
observed
in
these
studies.
It
is
beyond
the
scope
of
this
enquiry
to
isolate
any
single
one,
but
there
is
a
class
of
cognitive
mechanism
that
might
admit
a
simple
corrective
to
bias.
An
emergent
strand
of
economic
decision
theories
considers
attention
to
be
a
scarce
resource
and
so
models
human
information
processing
as
efficiently
incomplete
(e.g.
Kozsegi
and
Szeidl,
2013;
Sims,
2003;
Woodford,
2012).
Michael
Woodford,
who
put
forth
a
recent
iteration
of
the
idea,
calls
his
theory
Efficient
Perceptual
Distortion
(EPD).
Woodford
demonstrates
that
EPD
could
lead
an
optimizing
decision
maker
to
completely
ignore
a
consequential
attribute
(2012).
In
other
words,
EPD
could
lead
a
worker
to
choose
a
job
without
considering
how
enjoyable
it
might
be,
which
might
explain
the
results
of
our
lab
study.
According
to
EPD,
different
preference
elicitation
procedures
lead
to
differences
in
the
cognitive
costs
of
attending
to
consequential
attributes,
which
can
result
in
preference
reversals.
Prefacing
preference
elicitation
with
questions
that
draw
attention
to
neglected
attributes
is
a
proven
technique
to
attenuate
bias
induced
by
perceptual
distortion
(Comerford,
2011).
A
relevant
example
of
how
attention
can
be
drawn
comes
from
Study
4
in
Amir
et
al.
(2008).
Respondents
were
presented
with
the
scenario
that
they
have
lost
data
from
their
computer.
In
a
between-subjects
manipulation,
one
group
was
told
that
5
years
of
data
was
lost
and
that
the
technician
anticipates
taking
5
min
to
retrieve
it.
In
the
other
condition,
1
month
of
data
was
lost
and
it
will
take
12
h
to
retrieve
it.
Respondents
were
asked
the
maximum
they
would
be
willing
to
pay
the
technician
to
retrieve
the
data
and
to
rate
the
utility
they
would
derive
from
having
the
data
restored.
Respondents
in
the
“5
min
for
5
years
of
data”
condition
gave
a
higher
score
to
the
utility
of
having
that
data
retrieved
but
respondents
in
the
“12
h
for
1
month
of
data”
condition
were
willing
to
pay
more
to
have
the
data
retrieved.
This
inconsistency
in
preference
was
exhibited
for
respondents
who
first
reported
their
willingness
to
pay,
and
then
rated
utility.
A
second
manipulation
in
the
experiment
was
question
order.
Some
respondents
were
assigned
to
rate
utility
before
reporting
willingness
to
pay.
The
respondents