Content uploaded by Julie R. Irwin
Author content
All content in this area was uploaded by Julie R. Irwin
Content may be subject to copyright.
188
䉷 2009 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 36 ● August 2009
All rights reserved. 0093-5301/2009/3602-0004$10.00. DOI: 10.1086/597049
Happiness for Sale: Do Experiential Purchases
Make Consumers Happier than Material
Purchases?
LEONARDO NICOLAO
JULIE R. IRWIN
JOSEPH K. GOODMAN*
Previous theories have suggested that consumers will be happier if they spend
their money on experiences such as travel as opposed to material possessions
such as automobiles. We test this experience recommendation and show that it
may be misleading in its general form. Valence of the outcome significantly mod-
erates differences in respondents’ reported retrospective happiness with material
versus experiential purchases. For purchases that turned out positively, experiential
purchases lead to more happiness than do material purchases, as the experience
recommendation suggests. However, for purchases that turned out negatively,
experiences have no benefit over (and, for some types of consumers, induce
significantly less happiness than) material possessions. We provide evidence that
this purchase type by valence interaction is driven by the fact that consumers adapt
more slowly to experiential purchases than to material purchases, leading to both
greater happiness and greater unhappiness for experiential purchases.
P
sychologists (e.g., Gilbert 2006; Kahneman, Diener, and
Schwarz 1999; Van Boven and Gilovich 2003), econ-
omists (e.g., Frank 1985; Veenhoven 1993), and public pol-
icy theorists (e.g., Easterlin 2003) have become increasingly
interested in measuring and understanding human happiness.
For psychologists, research on happiness has proved revo-
lutionary because an overfocus on negative clinical states
had omitted the positive range of human experience by fo-
cusing on what decreases pathology as opposed to what
increases well-being (e.g., Aspinwall and Staudinger 2003;
Seligman 2002). For economists, happiness provides a use-
*Leonardo Nicolao (Leonardo.Nicolao@phd.mccombs.utexas.edu) is a
doctoral student and Julie R. Irwin (jirwin@mccombs.utexas.edu) is as-
sociate professor of marketing at the McCombs School of Business at the
University of Texas at Austin, 1 University Station, B6700, Austin, TX
78712. Joseph K. Goodman (goodman@wustl.edu) is assistant professor
of marketing at the Olin Business School, Washington University in St.
Louis, Campus Box 1133, One Brookings Drive, St. Louis, MO 63130.
Address correspondence to Leonardo Nicolao. The authors acknowledge
the helpful input of the editor, the associate editor, and the reviewers. In
addition, the authors thank the members of the Irwin Lab at the University
of Texas at Austin at the time of this research, Morgan Ward, Michael
Luchs, and Rebecca Naylor, for their valuable input and support. This
article is based on the first author’s dissertation.
John Deighton served as editor and Marsha Richins served as associate
editor for this article.
Electronically published January 14, 2009
ful comprehensive construct with which to analyze human
welfare because “everybody wants to be happy. There is
probably no other goal in life that commands such a high
degree of consensus” (Frey and Stutzer 2002, vii). Consumer
researchers have a stake in both of these general aims as
well. When considering the transformation of consumers
through purchasing (Mick 2006), it makes sense to consider
not only what leads to consumer downfall (debt, drug ad-
diction, etc.) but also what leads to an especially happy life.
As with economics, consumer research can become overly
atomized, measuring the effects of particular consumption
episodes without a sense of the larger picture of where these
episodes lead in the long run.
The specific topics addressed across happiness research
have covered a wide range, from precise neurological map-
ping (LeDoux and Armony 1999) to cross-cultural survey-
based policy recommendations (Veenhoven 1993). Some of
the most compelling recent works on happiness address the
issue we address in this article: the effect of particular human
behaviors on subsequent happiness. This research has fo-
cused on issues such as religious activities and exercise
(Mochon, Norton, and Ariely 2008), marriage and family
experiences (Easterlin 2003), and gratitude (Lyubomirsky,
Sheldon, and Schkade 2005).
However, there has not been much experimental explora-
tion of an issue primary to consumer theory: how particular
purchases affect happiness. As we detail later in this article,
HAPPINESS FOR SALE 189
this gap in the literature may be due to a surprising agreement
among theorists on this issue. Perhaps the theory has proved
so convincing that experimentalists saw no need to test it.
Dating as early as Hume (1737/1975) and through Scitovsky
(1976) and Frank (1985), the sentiment has been that indi-
viduals will be happier if they spend their money on expe-
riences (e.g., theater, concerts, and vacations) as opposed to
material purchases (e.g., fancy cars, bigger houses, and gad-
gets). We term this the experience recommendation.
There is only one published empirical test of this expe-
rience recommendation. Van Boven and Gilovich (2003),
using a number of clever experiments, found that their re-
spondents did derive more happiness from positive expe-
riential purchases when compared to positive material pur-
chases. Our article is a long overdue experimental treatment
of this issue.
In three experiments, we affirm Van Boven and Gilovich’s
(2003) finding, but we show that their results are limited to
positive purchases. For negative purchases, experiences
have no advantage over material goods, and sometimes ma-
terial purchases even induce more happiness than do ex-
periences, the opposite of the experience recommendation.
We provide evidence that the experiences versus material
goods distinction (as opposed to some other correlated var-
iables) is underlying our results. For instance, a well-tested
materialism scale moderates the findings in both our first
and second experiments.
Moreover, in the first experimental test of hedonic ad-
aptation rates across types of purchases, we show that in-
dividuals adapt slower to experiential purchases when com-
pared to material purchases. This differential in adaptation
rates for experiential and material purchases is the under-
lying mechanism for our effects. Because individuals adapt
more slowly to experiential purchases, adaptation leads to
both lower (for negative purchases) and higher (for positive
purchases) levels of retrospective happiness for experiential
versus material purchases.
THEORETICAL BACKGROUND
What Is Happiness, and How Does Purchasing
Affect It?
We use the terms happiness and subjective well-being
(Diener 1984) interchangeably. Other researchers (e.g.,
Frank 1999; Seligman 2002) have established that these terms
are strongly interrelated and that ratings of happiness correlate
highly with other measures of both psychological and phys-
iological well-being (e.g., Sutton and Davidson 1997). Al-
though measures of subjective well-being sometimes include
other affective and cognitive components (e.g., satisfaction
with life; Pavot and Diener 1993), happiness explains most
of the variance in the subjective well-being construct (Comp-
ton et al. 1996).
Happiness is measurable, predictable, and comparable
across contexts (Diener 1984; Diener et al. 1999; Gilbert
2006; Layard 2005). We follow previous researchers in char-
acterizing happiness as an overall sense that life is good (My-
ers 1992), that life contains many positive situations and emo-
tions (Ahuvia 2008; Lyubomirsky, King, and Diener 2005).
Researchers typically measure happiness by asking people
how happy they are (see Kahneman et al. 1999 for many
examples) or how happy they are with a particular situation
(Raghunathan and Irwin 2001) using multi-item scales.
Scitovsky (1976) suggested some years ago that purchasing
may have a negative impact on happiness because consumers
often buy “joyless” material possessions such as houses and
cars, resulting in comfort but not pleasure. Likewise, Easterlin
(2003) has proposed that investment in “pecuniary” market
objects has no effect on happiness. In his essay “How Not
to Buy Happiness” (2004), Robert Frank echoes these sen-
timents by recommending against the accumulation of “con-
spicuous goods” in the pursuit of happiness. These recom-
mendations lead us to wonder, by encouraging people to buy
more things, are marketers encouraging them toward a less
happy life?
Most of the writers who discuss the negatives of material
goods nevertheless do suggest that some types of purchases
may increase happiness. Scitovsky (1976) argues for cultural
experiences such as vacations and concerts to provide plea-
sure. Thought exercises based on adaptation studies argue
that consumers might be happier trading their material
wealth (e.g., a percentage of the square footage of their
house) for a more pleasant day-to-day lifestyle (e.g., a more
pleasant commute) and more enjoyable experiences through-
out the year (e.g., more vacation time; Frank 1999). In his
book, The Happiness Hypothesis (2006, 101), Jonathan
Haidt suggests that people should, “accumulate less, and
‘consume’ more . . . vacations, and other enjoyable activ-
ities.” In other words, there is a suggestion that purchases
of material goods (e.g., cars, houses, furniture) should, over-
all, lead to less happiness compared to purchases of expe-
riential goods (e.g., vacations, concerts, sporting events).
We call this suggestion “the experience recommendation.”
Besides appearing in the writings outlined above, this rec-
ommendation has been established empirically (Van Boven
and Gilovich 2003) and can be traced back to Hume (1737/
1975, 283), who extols experiences (e.g., theater) as superior
to “the acquisition of worthless toys and gewgaws.”
What distinguishes material from experiential purchases?
Material purchases are tangible; they may be taken from
place to place, they last beyond a couple of days, and they
take up physical space. Stereos, cars, and houses are ex-
amples of material goods. Experiential purchases are not
tangible. Rather, the purchase entitles the consumer to an
event that is finite in time. Movies, amusement parks, and
restaurant dinners are examples of experiential purchases.
The material/experiential distinction is a continuum. For
some purchases, for some consumers, locating particular
purchases on the continuum may be difficult. However, the
findings of Van Boven and Gilovich (2003) and the intui-
tions of many others (e.g., Easterlin 1995; Pine and Gilmore
1999; Scitovsky 1976) suggest that this distinction exists
and that consumers can easily discriminate between the two
types of purchases. Future research will likely uncover ad-
190 JOURNAL OF CONSUMER RESEARCH
ditional subtleties relevant to these two categories of pur-
chases (e.g., performance characteristics [Deighton 1992],
vividness, degree of social interaction, memory distortions),
but for now we concentrate on this basic distinction and
terminology. Doing so allows us to speak to the experience
recommendation and to link our work with previous work
in the area.
Happiness, Consumption, and the Hedonic
Treadmill
In addition to documenting and testing the experience
recommendation, we propose and test an underlying mech-
anism for our findings. We suspect that purchase type affects
the “hedonic treadmill” (Brickman and Campbell 1971; Ra-
ghunathan and Irwin 2001), the adaptation mechanism that
integrates positive purchases into the decision makers’ ref-
erence point, shifting the purchase into the status quo instead
of a gain (Kahneman, Knetsch, and Thaler 1991; Samuelson
and Zeckhauser 1988). Experiences might be less suscep-
tible to this treadmill; people continue to enjoy past pleasant
experiences via memories (Frederick and Loewenstein 1999;
Van Boven and Gilovich 2003), and experiences may have
more of a lasting impact on one’s life (Frank 2004; Scitov-
sky 1976).
The hedonic treadmill is driven by hedonic adaptation
(Frederick and Loewenstein 1999), which refers to the less-
ening of a hedonic response over time. Better things become
less good over time, and worse things become better. Note
that there are two general classes of work on hedonic ad-
aptation (see Frederick and Loewenstein [1999] for many
examples of each): measuring adaptation to a continually
repeating stimulus (e.g., traffic noise [Weinstein 1982] or
incarceration [Wormith 1984]) and measuring adaptation to
a stimulus that occurs once (e.g., buying a luxury good
[Frank 1999]). Our work focuses on the latter, which is more
relevant to most marketing contexts, matches the discussions
of adaptation in most of the happiness literature, and fits
more closely with the experience recommendation.
Others have suggested that hedonic adaptation rates may
vary by type of stimuli. For instance, Van Boven (2005)
surmises that people may tend to adapt faster to material
purchases because (positive) experiential purchases remain
open to positive reinterpretation (Mitchell et al. 1997). As
a consequence, memory keeps the experiences from declin-
ing in happiness over time. Likewise, in their review of
adaptation rates, Frederick and Loewenstein (1999) find that
people show very little adaptation both to certain positive
experiences such as plastic surgery and certain negative ex-
periences such as loss of a loved one, whereas adaptation
to material gains such as a particular increase in income is
quite rapid. In a chapter of his book Luxury Fever entitled
“Gains That Endure,” Frank (1999, 88) considers whether
some purchases might provide slower adaptation than luxury
goods, which, he argues, place consumers on an endless
hedonic treadmill. He focuses on experiences such as va-
cations, saying: “Provided they are of sufficient duration,
vacations have been shown to have restorative effects that
persist long after people return to work.” The one-time pur-
chase of a vacation may lead to slower adaptation over time,
and thus more happiness, than spending a similar amount
on a luxury object.
We agree with this supposition, and we expand it to sug-
gest that, in general, people adapt to experiences, on average,
more slowly than to material purchases. Note that we apply
this proposition both to negative and positive purchases:
hedonic adaptation would result in a positive experience
inducing more happiness but a negative experience inducing
less happiness than the comparable material purchase with
the same initial happiness level. In our third experiment, we
directly test hedonic adaptation rates over time, by purchase
type and valence, and we show that hedonic adaptation un-
derlies our primary finding. In other words, we propose (and
show) that hedonic adaptation, by reducing the unhappiness
with negative purchases as well as the happiness with positive
purchases at different rates across purchase type, results in
concomitant differences in retrospective happiness. Further-
more, we show that this adaptation happens quite quickly,
often in a matter of minutes. Our work is the first (that we
know of) to test for happiness with both positive and negative
experiential/material purchases and also the first to explicitly
test purchase-type adaptation differences in time.
Thus, we experimentally address three components of the
experience recommendation: (1) whether/when the recom-
mendation is appropriate, (2) whether the appropriateness of
the recommendation depends on consumer characteristics
such as materialism, and (3) the mechanism underlying the
appropriateness of the experience recommendation.
EXPERIMENT 1
This experiment tested whether outcome valence mod-
erates the effect of purchase type on happiness. Respondents
first recalled a purchase (positive or negative, and material
or experiential) and then indicated their retrospective hap-
piness with the purchase. We expected the experience rec-
ommendation to hold for positive purchases but not for neg-
ative purchases.
In addition, we measured respondents’ level of materi-
alism, which is defined as attachment to material posses-
sions, including an enmeshed relationship between the own-
ership of objects and one’s sense of self (Richins and
Dawson 1992, 308). Although researchers have addressed
the societal and personal impact of materialism (many of
them arguing that it has a negative influence [e.g., Burroughs
and Rindfleisch 2002; Kasser 2002]), for our purposes the
measure of materialism allows us to affirm our material/
experiential continuum by isolating consumers for whom
material purchases are especially important versus especially
unimportant. As materialism increases, the adaptation rates
for material purchases (both positive and negative) should
decrease; material purchases should resonate longer for ma-
terialistic consumers, affecting happiness more. Thus, we
expected a three-way interaction between materialism, va-
HAPPINESS FOR SALE 191
FIGURE 1
MEAN HAPPINESS RATING BY OUTCOME
lence, and purchase type: the valence by purchase type in-
teraction should grow stronger as materialism decreases.
Method
A total of 211 undergraduates from various colleges at the
University of Texas at Austin participated in the experiment
in exchange for extra credit. We asked respondents to recall
a personal purchase. There were four between-subjects con-
ditions: the purchases were either experiential or material, and
they were either positive or negative. For the material pur-
chases, we asked respondents the following:
Please describe a time when you spent about $300 on an
object. You kept the object for some time and may even still
have it. It was an object you could touch with your hand.
You bought the object to increase your happiness.
For the experiential purchases, we asked the following:
Please describe a time when you spent about $300 on an
experience. In other words you did not end up with anything
tangible (anything you could hold in your hand) at the end
of the experience except for your memories. You bought the
experience to increase your happiness.
We adapted these instructions from Van Boven and Gi-
lovich (2003), with more neutral wording to accommodate
our positive and negative conditions (Van Boven and Gi-
lovich solicited memories of purchases around $50, and in
an unreported test of our two-way interaction we replicated
our results with this amount as well). The outcome valence
was manipulated in the last sentence of the instructions:
“And, it turned out well and you did enjoy the purchase”
versus “Unfortunately, it did not turn out well and you did
not enjoy the purchase.” For the material purchases, we
instructed participants that the purchases could not be sold
or given away, so that any confound with potential resale
was eliminated.
Next, respondents rated their purchase on three 7-point
happiness scales adapted from Van Boven and Gilovich
(2003): “When you think about this purchase, how happy
does it make you?” (Not Happy–Moderately Happy–Very
Happy), “How much does this purchase contribute to your
happiness in life?” (Not at All–Moderately–Very Much), “To
what extent do you think the money spent on this purchase
would have been better spent on something else—some other
type of purchase that would have made you happier?” (Not
at All–Moderately–Very Much). These measures formed one
scale . (In all three studies the first two questions(a p .86)
were highly correlated and using only the first mea-[r
1 .9],
sure resulted in the same results as using all three.) In addition,
we asked respondents how much they spent and how many
months ago they made the purchase. After the happiness task,
participants answered the nine-item version of “The Material
Values Scale” (MVS; Richins 2004), which we combined into
one measure (a p .84).
Results and Discussion
Twenty-one participants reported the wrong purchase va-
lence and were dropped from the analyses. These partici-
pants reported positive purchases when asked for negative
purchases and vice versa. The likelihood of not following
the instructions was not influenced by the conditions of this
experiment NS).
2
(x ! 1.5,
Overall, we did not find support for the experience rec-
ommendation: there was not a reliable relationship between
the material versus experiential variable and reported hap-
piness with the purchases As we(F(1, 185) p 1.95, NS).
expected, the effect of purchase type was significantly
moderated by the outcome valence of the purchase
( , ; see fig. 1). When
2
F(1, 185) p 3.85 p ! 0.05, h p .02
we fit the “Happiness p Purchase Type” model at each
level of purchase outcome valence (Irwin and McClelland
2001), we replicated the experience recommendation for
positive purchases: experiential purchases induced more
reported happiness than did material pur-(M p 5.75)
chases ( If the pur-M p 5.27; F(1, 185) p 5.53, p
! 0.05).
chase did not turn out positively, however, the effect did
not hold. Fitting the model for the negative purchases,
there was not a significant difference in happiness be-
tween experiential and material purchases(M p 2.52)
; NS). Note that the difference(M p 2.60 F(1, 185) p .16,
in effect across valence is unlikely to be due to statistical
power; if purchase type had significantly affected happiness
with negative purchases, the effect would be in the opposite
direction from the effect for positive purchases. Thus, pos-
itive outcomes result in a different pattern of results from
negative outcomes. Controlling for the purchase amount
and the time since the purchase negligibly affected the
interaction results when controlling for(F(1, 171) p 3.39)
amount and ( ) when controlling for time,F(1, 184) p 3.69
192 JOURNAL OF CONSUMER RESEARCH
FIGURE 2
PURCHASE TYPE BY OUTCOME VALENCE BY
MATERIALISM INTERACTION
and a Sobel test showed that amount paid and time since
purchase did not reliably mediate the results (t’s
! 1.65,
Thus, the interaction cannot be explained byp’s
1 .10).
differences in market value of the purchase types or by
differences in recall of the happiness afforded by the two
types of purchases over time.
Although there was not a main effect of materialism on
happiness with the purchase this con-(F(1, 181) ! 1, NS),
struct moderated the valence by purchase type interaction
To interpret this three-way in-(F(1, 181) p 6.06, p
! .05).
teraction, we fit the valence by purchase type model at two
levels of materialism, low (one standard deviation below
the mean) and high (one standard deviation above the mean).
As the top graph in figure 2 shows, the consumers lower
on materialism showed a particularly strong version of the
purchase type by happiness interaction (F(1, 181) p 9.81,
Fitting the model both at low materialism andp
! .001).
at each level of valence shows that experiential purchases
led to more happiness if there was a positive outcome
but that material purchases(F(1, 181) p 6.81, p
! .001)
led to marginally more happiness if there was a negative
outcome For less materialistic(F(1, 181) p 3.03, p
! .08).
consumers, switching valence resulted in a switch in the
effect of purchase type on happiness.
For more materialistic consumers (see bottom graph of
fig. 2), purchase type also did not influence happiness overall
NS). There was a main effect of outcome(F(1, 181) ! 1,
valence for these consumers, but they showed neither a prod-
uct type main effect nor the interaction we found in previous
studies NS). For negative outcomes, the(F(1, 181)
! 1,
more materialistic respondents were just as unhappy with
material purchases as with experiential purchases (presum-
ably their high hopes for material purchases were dashed).
For positive outcomes, material purchases resonate just as
much as experiences for more materialistic respondents be-
cause they are just as happy with the positive material pos-
sessions. The results provide support that it is indeed the
material-experiential dimension that is responsible for our
purchase type effects across the experiments.
EXPERIMENT 2
In experiment 1, we gave participants specific instructions
to recall either a material or an experiential purchase. This
design assumes that material and experiential purchases are
equally accessible to memory and that they are equally as-
sociated with negative and positive outcomes. Experiment 2
instead measures associations between purchase type and out-
come valence that occur naturally in consumers’ memories,
without any prompting for one kind of purchase or another.
We asked individuals to freely recall three different purchases
and then to rate each one on a material-experiential contin-
uum. Thus, this method is more reflective of the likely con-
tinuous nature of the experiential/material construct.
This experiment also rules out the possibility that chronic
happiness underlies our results by measuring Satisfaction
with Life (Pavot and Diener 1993) before the primary task,
with a 10-minute filler task in between. These Satisfaction
with Life scores did not have any main effects or interactions
with the happiness measure, and therefore they will not be
discussed further.
Method
We randomly assigned 198 undergraduate and MBA
students from the University of Texas at Austin to either
the positive or negative purchase conditions. Participants
recalled three purchases that turned out either well (pos-
itive condition) or poorly (negative condition) and then
briefly described the purchases. Afterward, they were
given the same definitions of purchase type used in our
first experiment and were asked to rate each of these
purchases on a 7-point scale, anchored by “completely
material” and “completely experiential.” Following the
ratings questions, participants received the same three
happiness questions and the MVS (Richins 2004) used
in the previous experiment.
HAPPINESS FOR SALE 193
FIGURE 3
PURCHASE TYPE ON HAPPINESS SLOPES BY
VALENCE AND MATERIALISM
N
OTE
.—Higher values in the y-axis denote greater happiness with experi-
ential purchases.
Results and Discussion
All participants reported three purchases, rated them on
the experiential or material purchases continuum, and then
rated the purchases on the happiness scales. Thus, we em-
ployed a two-step hierarchical analysis. In the first step, we
regressed each of the three happiness scores onto the three
purchase classification ratings for each participant. This
model provided us with slopes (one per participant) de-
scribing the influence of purchase type (material vs. expe-
riential) on individual levels of happiness with the purchase.
In the second step, we regressed these happiness by purchase
type slopes onto the valence of the outcome condition, as
well as the materialism scale and their interaction. This final
model captures the effect of materialism, valence, and their
interaction on the relationship between purchase type and
happiness. In other words, does the influence of purchase
type on happiness depend on valence? And does this de-
pendency differ by materialism? Figure 3 plots these results.
The y-axis represents the influence of purchase type on hap-
piness, where the more positive the number, the greater the
influence of experiential purchases on happiness. Con-
versely, the more negative the number on the y-axis, the
greater the influence of material purchases on happiness.
Memory for material versus experiential purchases did
not differ by valence: respondents did not recall experiences
or material purchases more in the positive versus negative
conditions ( NS). Replicating the results in
F(1, 192) p 1,
our previous experiments, the “overall” line shows that the
relationship between happiness with the purchase and pur-
chase type was significantly predicted by purchase valence
For positive pur-
2
(F(1, 192) p 6.22, p ! .05, h p .03).
chases, happiness was positively related to how experiential
the purchase was
(M p .104, F(1, 192) p 12.69, p
!
slope
For negative purchases, there was no relationship between
001).
happiness and purchase type ;
(M p .0001 F(1, 192)
! 1,
slope
NS).
As in experiment 1, there was a marginally significant
interaction between outcome valence and materialism
At a low level of materialism
(F(1, 192) p 3.57, p p .06).
(one standard deviation below the mean) experiences led
to more happiness than did purchases of material posses-
sions but only if the outcome turned out positively
However, for individuals
(F(1, 192) p 9.53, p
! 0.05).
with high levels of materialism (one standard deviation
above the mean), experiential and material purchases did
not differentially affect happiness, regardless of the va-
lence of the outcome of the purchase NS).(F(1, 192)
! 1,
Note that the results from this experiment are especially
conservative. We allowed respondents to freely recall any
purchases that came to mind. This design allows us to make
stronger claims about the experience recommendation be-
cause it mimics the consumer’s process when remembering
a positive (or negative) purchase without any constraint as
to the type of purchase.
EXPERIMENT 3
Experiments 1 and 2 showed that the effect of purchase
type on happiness depends on the purchase outcome. How-
ever, we have not directly tested our proposed mechanism
for this effect. Previously we suggested that adaptation dif-
ferences drive this interaction. People adapt more slowly to
experiences than to material purchases and, as a conse-
quence, happiness is mitigated less over time for experiences
than for material purchases. In experiment 3, participants
were given the chance to choose an actual product (material
purchase) or experience (experiential purchase). We then
measured their happiness with the purchase, using the same
happiness measures that have been used throughout our stud-
ies. We collected the ratings (1) immediately following con-
sumption of the chosen option, (2) 7 minutes later, (3) 1
day later, (4) 1 week later, and (5) 2 weeks later.
Method
Three hundred fifty-five students from the University of
Texas at Austin participated in this experiment in exchange
for extra credit in an introductory marketing course. Each
participant was assigned (without their knowledge) to either
a “material purchase” or an “experiential purchase” condi-
tion. In the experiential condition, the participants were told
that they could use three “lab dollars” to purchase one ex-
perience from a set of three possible options (a video clip,
a song, or a video game) to consume (i.e., watch the video,
listen to the song, or play the video game) in the lab session.
The three items presented to the respondents were chosen
randomly from a pool of seven possible experiences (two
video clips, two songs, and three video games) and were
each priced at $3. After the choice task, participants watched
194 JOURNAL OF CONSUMER RESEARCH
the video, listened to the song, or played the video game in
the experiential purchase condition. Once the experience
was over, participants answered the same three-item hap-
piness scale used in our previous studies. Then, participants
took a 7-minute break. More specifically, the experimenter
made the following announcement, “We are going to have
a 7-minute break in this session. During this break, you can
do anything you want. The lab door will remain open, so
you can come and go at your will.” This break provided us
with the first adaptation period.
After the 7-minute break, participants answered the same
set of dependent measures used in the first part of the study
(before the break). They saw the following instructions:
We are going to ask you some final questions about your
choice of a [their choice]. We know we asked these questions
earlier. However, we are interested in your answers to them
now (your answers may be different or they may be the same).
The participants were then debriefed with instructions to
answer the follow-up questionnaires. The follow-up ques-
tionnaires were collected through the Internet 1 day, 1 week,
and 2 weeks after the in-lab experimental session. Each
follow-up questionnaire consisted of a brief introduction
reminding our participants of the in-lab study, without any
reference to the options from which they had chosen. Fol-
lowing this introduction, participants responded, via e-mail,
with (1) a description of what they had chosen (to check
that they remembered what they had chosen, which all of
them did) and (2) their answer to the same happiness scale
question used in our lab sessions.
Individuals in the material purchase condition went
through the same procedure used in the experiential purchase
condition except that they chose to “purchase” one item from
a group of three products (instead of three experiences),
randomly sampled from a set of seven options (a set of
pencils, a can holder, a keychain, a ruler, a deck of cards,
a screwdriver, and a small picture frame). The items all had
retail prices close to $3.00 (the amount the participants were
“charged” in lab dollars for the purchase). After the choice,
participants in the material purchase condition received their
product and were told they could take this product home
with them.
Results and Discussion
We had 100% response rates for dependent measures col-
lected immediately after choice and 7 minutes after choice
(i.e., the adaptation rates collected in the lab). Response rates
were 83.3% for dependent measures collected the day after,
69.8% for the week after, and 61.1% for 2 weeks after
choice. A hazard regression indicated that response rates
were not influenced by our between-subject type of purchase
manipulation NS). In other words, purchase
2
(x (1) !p 1,
type did not affect response rates. In addition, there was
variance in the choice of experiences and material goods;
the choice probabilities for the seven experiences were .26,
.22, .19, .13, .11, .09, and .05, and for the seven material
goods they were .30, .21, .20, .14, .08, .07, and .03.
As in our previous studies, these data had two between-
subjects variables: purchase type (which was manipulated)
and purchase valence (which was measured). In addition,
we had a within-subject longitudinal variable, time since
purchase. Consistent with many (if not most) longitudinal
data of psychological responses (e.g., Drew and Abbott
2006), our data followed a power law function:
k
y p ax . (1)
In figure 4, parts A and B depict happiness scores for material
and experiential purchases, respectively, over time (in
minutes) by the initial happiness scores (which we term the
“set points”). The set points are the continuous measure of
the purchase valence (positive to negative). As the figure
shows, we obtained a classic power function shape, with
most of the change happening in the first time periods. The
average AIC (Akaike Information Criterion, for which
lower values indicate better model fit) across set point was
⫺21.08 for the power law model, ⫺14.29 for a quadratic
model, and ⫺11.08 for a linear model.
The proper analysis for the data is a hierarchical mixed
design model that first captures the longitudinal effects and
then tests for differences in these effects by the between-
subjects variables. The first step captured the power function
by using the standard model (the powerlog (y) p log (x)
function becomes linear when transformed in this way). Thus,
for our model we regressed log-transformed happiness scores
onto the log-transformed time since choice. Time since choice
was measured in minutes—1, 7, 1,440 (1 day), 10,080 (1
week), and 20,160 (2 weeks) minutes after the choice.
This step calculates a slope for each respondent reflecting
the effect of elapsed time on happiness with the choice (note
that the slope from this model corresponds to the exponent
in the power function).
log (happiness) p b + b x + error. (2)
01i log(time)
The resulting individual slopes were then regressed onto
type of purchase (experiential vs. material purchases), each
individual’s set point, and the interaction between both var-
iables:
b p b + b x + b x
1i 0 a type of purchase b set point
+ b (x # x ) + error. (3)
c type of purchase set point
Thus, in this model we test for different adaptation rates for
experiential and material purchases as well as our purchase
type by valence interaction.
This regression confirmed that the relationship between
time since choice and happiness is moderated by(b )
1i
type of purchase and initial happiness with the choice
Figure 5 helps explicate this(F(1, 245) p 14.50, p
! .001).
interaction (using the methods described in Irwin and
McClelland [2001]). Part A of the figure plots (log) happiness
HAPPINESS FOR SALE 195
FIGURE 4
HAPPINESS WITH MATERIAL PURCHASES OVER TIME IN MINUTES (A) AND
HAPPINESS WITH EXPERIENTIAL PURCHASES OVER TIME IN MINUTES (B)
scores for choices made in the material purchase condition
by (log) time since consumption. Each line corresponds to a
model fit at different set points (1, 2, 6, and 7, where 1 is
the lowest happiness score and 7 is the highest). Part B pres-
ents the analogous data for experiential purchases.
As the figure shows, the slopes are steeper in the material
than in the experiential conditions (F(1, 245) p 4.83, p
!
showing that material purchases indeed show faster.05),
adaptation than experiential purchases. In fact, in this data
we do not see much adaptation at all for particularly negative
purchases. As Frederick and Loewenstein (1999) explain,
some situations do not induce adaptation, and sometimes
they induce sensitization (increased sensitivity over time).
Perhaps some negative experiences are of this nature, al-
though of course we hesitate to make too much of this
possible sensitization apart from noting that it is evocative
(and consistent with our expectation of lack of adaptation
to experiences).
To underscore the effect these adaptation differences have
on subsequent happiness, we tested the effects of purchase
type and set point (valence of the outcome, i.e., the inter-
action originally described in our previous studies) across
the different points in time. When the model is fit (Irwin
and McClelland 2001) at time 2 (7 minutes after the choice),
there is not an interaction between type of purchase and
valence of the outcome NS). However, as(F(1, 245) ! 1,
time passes, this interaction appears (F(1, 245) p 8.14,
and ; all p’s
! .05) forF(1, 245) p 9.58, F(1, 245) p 9.96
measures taken 1 day, 1 week, and 2 weeks after the lab
session, respectively. This set of results not only replicates
our retrospective happiness findings from previous studies
but also supports our contention that hedonic adaptation is
driving the effects.
Note that the stimuli in this study were nested under the
material and experiential categories. There are pros and cons
to nested designs in this context. Although participants chose
196 JOURNAL OF CONSUMER RESEARCH
FIGURE 5
LOG(HAPPINESS) BY LOG(TIME) IN MATERIAL PURCHASE CONDITION (A) AND
LOG(HAPPINESS) BY LOG(TIME) IN EXPERIENTIAL PURCHASE CONDITION (B)
from among a wide variety of potential purchases and no
particular purchase dominated choice within the two pur-
chase types, as in all nested designs the material/experiential
classification is confounded with the particular choices
within the two categories. The benefit of a nested design,
which probably explains its popularity in marketing re-
search, is that the items naturally embody the categories
(i.e., games actually are experiences; rulers actually are ma-
terial goods).
GENERAL DISCUSSION
In order to test the experience recommendation, we mea-
sured retrospective happiness with material and experiential
purchases by valence, and we found that experiences tend
to produce both more (for positive purchases) and less (for
negative purchases) happiness than do material purchases.
In addition, we mapped actual hedonic adaptation across
time for material and experiential purchases, and we found
that adaptation happens more quickly for material purchases
than for experiential purchases. Thus, it is not surprising
that experiences end up inducing a wider variance of hap-
piness. We even showed that, after only a day of adaptation,
purchases that started at the same level of rated happiness
had diverged enough to induce a purchase type by valence
interaction in retrospective happiness ratings.
Our results replicate those of Van Boven and Gilovich
(2003), showing that, for positive purchases, experiences
lead to greater happiness than material purchases. Also, there
is a sense in which our results support the experience rec-
ommendation because we show that on average the most
happiness obtained through purchasing is likely to be ob-
tained through experiential purchases that turn out well.
However, the experience recommendation in its pure form
HAPPINESS FOR SALE 197
is incomplete. Our findings suggest that a lifetime of neg-
ative experiential purchases might lead to quite an unhappy
life and furthermore that negative material purchases may
not leave as much of a negative mark.
Why are there different adaptation rates across purchase
types? The answer may lie in the human condition. The
General Social Survey has consistently suggested that mar-
riage and family experiences increase happiness and the
negatives of those experiences (divorce, death of a loved
one) have the opposite effect (Easterlin 2003). In his review
of the correlates of happiness, Argyle (1999) finds that pos-
itive social interaction is a major source of happiness; many
experiential purchases involve activities with other people,
including family. In addition to social interaction, the ac-
complishment of goals and the ability to be lost in a task
(Csikszentmihalyi and Csikszentmihalyi 1988) seem to be
correlated with happiness. Even solitary experiential pur-
chases, such as the purchase of an on-line game, allow for
the possibility of this kind of flow. In addition, Argyle (1999)
shows in his meta-analysis that exercise increases happiness.
Although it is a less common component of experiential
purchases than social interaction and active engagement,
exercise nevertheless may vary by the experiential/material
distinction (e.g., walking around an amusement park or zoo
vs. sitting in a car). Thus, positive experiences may be cor-
related with the basics of human happiness, and negative
experiences may represent the thwarting of these basics.
Adaptation may be slower for many experiences for this
reason; future research can show whether experiential pur-
chases are a proxy for more fundamental human needs.
In addition, there are several psychological findings that
corroborate (and may underlie) our adaptation results. For
instance, experiences may be more self-involving than ma-
terial goods on average (Van Boven 2005). Research also
has suggested that positive experiences not only live on in
memories but also lend themselves to even more positive
reinterpretations over time as the negative aspects of them
fade (Mitchell et al. 1997; Van Boven 2005; Van Boven and
Gilovich 2003). Thus, positive experiences resist adaptation
and remain more positive over time. It is not clear how this
mechanism would affect negative experiences, but Mitchell
et al. (1997) acknowledge that perhaps negative experiences
are framed even more negatively as time passes.
Our results evoke many such possibilities, and future re-
search can address all of these subtleties. In the process,
future research is likely to uncover subcategories of the
material/experiential distinction and to find that demograph-
ics such as age and gender influence the effect of product
type on happiness. We are particularly interested in the ways
in which memory differences might underlie the material/
experiential distinction, and we believe that memory may
be an important subconstruct of this continuum. Also, in
our work we did not make subtle distinctions between con-
structs such as life satisfaction, happiness, quality of life,
and so forth. However, we should acknowledge that some
researchers distinguish among these terms, for instance, us-
ing satisfaction with life as a more cognitive judgment of
how well life is going (Diener et al. 1999). Future explo-
rations of this topic might profitably identify how different
conceptualizations of happiness are related to purchase type.
Our research does not address the general question of
whether accumulated purchase patterns affect overall hap-
piness. There also remain unanswered questions about the
relationship between materialism and the experience rec-
ommendation. For example, if the material/experiential dis-
tinction applies to consumption as well as to products (Holt
1995), then it may be possible to consume experiences ma-
terialisticaly and vice versa.
In addition, future explorations could directly contrast the
more atomic judgments of purchases (e.g., ratings of effec-
tiveness, satisfaction with the purchase, intent to repurchase
within the brand, and quality) that are common in marketing
research with the global happiness judgments we measure.
One interesting component of this comparison is the potential
inconsistency between what people believe they want and
what actually makes them happy. As Daniel Gilbert (2006)
writes in his book, Stumbling on Happiness, humans spend
much of their time trying to behave in ways that will make
their future selves happy. Why do we guess wrong so often?
Applied to the marketplace, perhaps initial judgments and
affective responses to products are not particularly predictive
of how these products might contribute to happiness in the
long run.
REFERENCES
Ahuvia, Aaron C. (2008), “Wealth, Consumption, and Happiness,”
in The Cambridge Handbook of Psychology and Economic
Behaviour, ed. Alan Lewis, New York: Cambridge University
Press, 199–226.
Argyle, Michael (1999), “Causes and Correlates of Happiness,” in
Well-Being: The Foundations of Hedonic Psychology,ed.
Daniel Kahneman, Ed Diener, and Norbert Schwarz, New
York: Russell Sage Foundation, 353–73.
Aspinwall, Lisa G. and Ursula M. Staudinger (2003), A Psychology
of Human Strengths: Fundamental Questions and Future Di-
rections for a Positive Psychology, Washington, DC: Amer-
ican Psychological Association.
Brickman, Philip and Donald Campbell (1971), “Hedonic Rela-
tivism and Planning the Good Society,” in Adaptation-Level
Theory: A Symposium, ed. Mortimer H. Appley, New York:
Academic Press, 287–304.
Burroughs, James E. and Aric Rindfleisch (2002), “Materialism
and Well-Being: A Conflicting Values Perspective,” Journal
of Consumer Research, 29 (December), 348–70.
Compton, William C., Maggie L. Smith, Kim A. Cornish, and
Donald L. Qualls (1996), “Factor Structure of Mental Health
Measures,” Journal of Personality and Social Psychology,71
(2), 406–13.
Csikszentmihalyi, Mihaly and Isabella S. Csikszentmihalyi (1988),
Optimal Experience: Psychological Studies of Flow in Con-
sciousness, New York: Cambridge University Press.
Deighton, John (1992), “The Consumption of Performance,” Jour-
nal of Consumer Research, 19 (December), 362–72.
Diener, Ed (1984), “Subjective Well-Being,” Psychological Bul-
letin, 95 (3), 542–75.
198 JOURNAL OF CONSUMER RESEARCH
Diener, Ed, Eunkook M. Suh, Richard E. Lucas, and Heidi L. Smith
(1999), “Subjective Well-Being: Three Decades of Progress,”
Psychological Bulletin, 125 (2), 276–302.
Drew, Patrick and L. F. Abbott (2006), “Models and Properties of
Power-Law Adaptation in Neural Systems,” Journal of Neuro-
physiology, 96 (2), 826–33.
Easterlin, Richard A. (1995), “Will Raising the Incomes of All
Increase the Happiness of All?” Journal of Economic Be-
havior and Organization,27(1),35–48.
——— (2003), “Explaining Happiness,” Proceedings of the Na-
tional Academy of Sciences, 100 (19), 11176–83.
Frank, Robert H. (1985), “The Demand for Unobservable and
Other Nonpositional Goods,” American Economic Review,75
(1), 101–16.
——— (1999), Luxury Fever: Why Money Fails to Satisfy in an
Era of Excess, New York: Free.
——— (2004), “How Not to Buy Happiness,” Daedalus, 133 (2),
69–76.
Frederick, Shane and George Loewenstein (1999), “Hedonic Ad-
aptation,” in Well-Being: The Foundations of Hedonic Psy-
chology, ed. Daniel Kahneman, Ed Diener, and Norbert
Schwarz, New York: Russell Sage Foundation, 302–29.
Frey, Bruno S. and Alois Stutzer (2002), Happiness and Econom-
ics: How the Economy and Institutions Affect Human Well-
Being, Princeton, NJ: Princeton University Press.
Gilbert, Daniel (2006), Stumbling on Happiness, New York: Knopf.
Haidt, Jonathan (2006), The Happiness Hypothesis: Finding Mod-
ern Truth in Ancient Wisdom. Cambridge: Basic.
Holt, Douglas B. (1995), “How Consumers Consume: A Typology
of Consumption Practices,” Journal of Consumer Research,
22 (June), 1–16.
Hume, David (1737/1975), A Treatise of Human Nature,ed.L.
A. Selby-Bigge and P. H. Nidditch, Oxford: Clarendon.
Irwin, Julie R. and Gary H. McClelland (2001), “Misleading
Heuristics and Moderated Multiple Regression Models,”
Journal of Marketing Research, 38 (February), 100–109.
Kahneman, Daniel, Ed Diener, and Norbert Schwarz (1999), Well-
Being: The Foundations of Hedonic Psychology, New York:
Russell Sage Foundation.
Kahneman, Daniel, Jack L. Knetsch, and Richard H. Thaler (1991),
“The Endowment Effect, Loss Aversion, and Status Quo
Bias,” Journal of Economic Perspectives, 5 (1), 193–206.
Kasser, Tim (2002), The High Price of Materialism, Cambridge,
MA: MIT Press.
Layard, Richard (2005), Happiness: Lessons from a New Science,
New York: Penguin.
LeDoux, Joseph and Jorge Armony (1999), “Can Neurobiology
Tell Us Anything about Human Feelings?” in Kahneman et
al. 1999, 489–99.
Lyubomirsky, Sonja, Laura King, and Ed Diener (2005), “The
Benefits of Frequent Positive Affect: Does Happiness Lead
to Success?” Psychological Bulletin, 131 (6), 803–55.
Lyubomirsky, Sonja, Kennon M. Sheldon, and David Schkade
(2005), “Pursuing Happiness: The Architecture of Sustainable
Change,” Review of General Psychology, 9 (2), 111–31.
Mick, David Glen (2006), “Meaning and Mattering through Trans-
formative Consumer Research,” presidential address before
the Association for Consumer Research, Advances in Con-
sumer Research,33(1),1–4.
Mitchell, Terence R., Leigh Thompson, Erika Peterson, and Randy
Cronk (1997), “Temporal Adjustments in the Evaluation of
Events: The ‘Rosy View,’” Journal of Experimental Social
Psychology, 33 (4), 421–48.
Mochon, Daniel, Michael I. Norton, and Dan Ariely (2008), “Get-
ting Off the Hedonic Treadmill One Step at a Time: The
Impact of Regular Religious Practice and Exercise on Well-
Being,” Journal of Economic Psychology, 29 (5), 632–42.
Myers, David G. (1992), The Pursuit of Happiness: Who Is
Happy—and Why? New York: Morrow.
Pavot, William and Ed Diener (1993), “Review of the Satisfaction
with Life Scale,” Psychological Assessment, 5 (2), 164–72.
Pine, B. Joseph and James H. Gilmore (1999), The Experience
Economy: Work Is Theatre and Every Business a Stage, Bos-
ton: Harvard Business School Press.
Raghunathan, Rajagopal and Julie R. Irwin (2001), “Walking the
Hedonic Product Treadmill: Default Contrast and Mood-
Based Assimilation in Judgments of Predicted Happiness with
a Target Product,” Journal of Consumer Research, 28 (De-
cember), 355–68.
Richins, Marsha L. (2004), “The Material Values Scale: Measure-
ment Properties and Development of a Short Form,” Journal
of Consumer Research, 31 (June), 209–19.
Richins, Marsha L. and Scott Dawson (1992), “A Consumer Values
Orientation for Materialism and Its Measurement: Scale De-
velopment and Validation,” Journal of Consumer Research,
19 (December), 303–16.
Samuelson, William and Richard Zeckhauser (1988), “Status Quo
Bias in Decision Making,” Journal of Risk and Uncertainty,
1 (1), 7–59.
Scitovsky, Tibor (1976), The Joyless Economy: An Inquiry into
Human Satisfaction and Consumer Dissatisfaction,New
York: Oxford University Press.
Seligman, Martin E. P. (2002), Authentic Happiness: Using the
New Positive Psychology to Realize Your Potential for Lasting
Fulfillment, New York: Free Press.
Sutton, Steven K. and Richard J. Davidson (1997), “Prefrontal
Brain Asymmetry: A Biological Substrate of the Behavioral
Approach and Inhibition Systems,” Psychological Science,8
(3), 204–11.
Van Boven, Leaf (2005), “Experientialism, Materialism, and the
Pursuit of Happiness,” Review of General Psychology,9(2),
132–42.
Van Boven, Leaf and Thomas Gilovich (2003), “To Do or to Have?
That Is the Question,” Journal of Personality and Social Psy-
chology, 85 (6), 1193–1202.
Veenhoven, Ruut (1993), Happiness in Nations: Subjective Ap-
preciation of Life in 56 Nations, 1946–1992, RISBO, Studies
in Social and Cultural Transformation, no. 2, Rotterdam: Eras-
mus University Rotterdam, Netherlands.
Weinstein, Neil (1982), “Community Noise Problems: Evidence
against Adaptation,” Journal of Environmental Psychology,
2 (2), 87–97.
Wormith, J. Stephen (1984), “The Controversy over the Effects of
Long-Term Imprisonment,” Canadian Journal of Criminol-
ogy, 26 (4), 423–37.
