How I Decide Depends on What I Spend:

Use of Heuristics Is Greater for Time

than for Money
RITESH SAINI
ASHWANI MONGA*
We demonstrate that decision making is more heuristic in situations that involve
spending time rather than money. Relative to participants in the money condition,
those in the time condition show a higher propensity to choose a compromise option
(experiment 1) and to rely on an arbitrary anchor (experiment 2). We propose that
such heuristics are used more for time because, compared to monetary expenditures,
temporal expenditures are harder to account for. Consistent with this proposition,
when participants in both time and money conditions are primed to account for their
expenditures, they no longer differ in their use of heuristics. The associated response
times offer additional process evidence (experiment 3).

s consumers, we are used to spending both money and

time. We spend money paying for things like furniture,
flights, and books; we spend time assembling do-it-yourself
furniture, searching for the cheapest flight on the Internet,
and waiting for delivery of our book order. Sometimes we
also make a priori decisions regarding such expenditures. For
instance, a parent planning to visit potential universities for
her promising high school child would foresee spending days
as well as money on travel. Because constraints of time and
money will not permit travel to all possible universities, she
makes an a priori decision about the number of universities
to visit. This decision will then help her arrange for time (e.g.,
applying for leave from work) and money (e.g., applying for
another credit card). This decision will also be of consequence
for universities hoping to be in her consideration set. In this
article, we try to understand how such decisions are made
differently for expenditures of time than those of money.
Specifically, we study whether there are time-money differences in the extent to which heuristics are used.

Economists usually treat time in a manner similar to their

treatment of moneyas just another scarce resource that
people spend to achieve certain ends (Becker 1965). However, time and money are often treated differently by consumers. For instance, choices are risk seeking for losses of
money but relatively risk averse for losses of waiting time
(Leclerc, Schmitt, and Dube 1995). Relatedly, while studying time as a medium of exchange, Okada and Hoch (2004)
find that, when individuals pay in time rather than money,
they display greater willingness to pay for riskier options.
In the context of sunk costs, the consideration of past investments has been found to be relatively weaker for time
(Soman 2001). Finally, extending prior work by Soman
(1998), Zauberman and Lynch (2005) show that time costs
are discounted more than money costs because of a difference in how slack (i.e., perceived surplus) changes from the
present to the future; change in slack is greater for time than
for money. To sum up, prior research reveals that, when
information such as magnitude of risk, prior investments,
or time delay is processed by individuals in the context of
time versus money, there are quantitative differences in how
this information is used to arrive at judgments and decisions.
We contribute to this stream of research by demonstrating that a qualitatively different form of decision makingmaking decisions based on quick-and-easy heuristics
rather than an analysis of the available informationgains
prominence when one works with time rather than money.
Recall the aforementioned parent trying to decide how
many universities to visit. Her decision could be based on
a careful analysis (Because it costs $200 to travel to one
university, I should go to just three), or it could be driven

*Ritesh Saini is assistant professor of marketing, School of Management, George Mason University, Enterprise Hall 130, Fairfax, VA
22030 (sritesh@gmu.edu). Ashwani Monga is assistant professor of marketing, College of Business, University of Texas at San Antonio, One UTSA
Circle, San Antonio, TX 78249 (ashwani.monga@utsa.edu). The authors
would like to thank seminar participants at their universities, as well as
the editor, the associate editor, and the three reviewers, for their valuable
input. The sequence of authorship was decided using a coin toss; the authors
contributed equally to this article.
John Deighton served as editor and Ann McGill served as associate editor
for this article.
Electronically published February 28, 2008

914
2008 by JOURNAL OF CONSUMER RESEARCH, Inc. Vol. 34 April 2008
All rights reserved. 0093-5301/2008/3406-0014$10.00

HOW I DECIDE DEPENDS ON WHAT I SPEND

by a qualitatively different process in which she relies on

a rule of thumb, such as choosing the compromise option
(Out of my three friends, one visited two universities, one
visited five, and one visited eight; let me take the middle
option and visit five), or anchoring on a number from
another context (I always evaluate three options before
making a decision; let me visit three universities). These
two heuristicscompromise and anchoringhave been
very well documented in the consumer research literature,
and we will focus on them to test our proposed theory
about time-money differences in the use of heuristics.
The compromise effect is a type of menu dependence
(Huber, Payne, and Puto 1982; Simonson 1989) in which
preferences are irrationally influenced by the menu of options
in the choice set. This effect runs counter to the following
notion of similarity: given two options B and C, the inclusion
of an option A, which is more similar to B than to C, ought
to reduce the preference for B (relative to C) because some
of those who earlier chose B can now choose A. Simonson
(1989) demonstrates that including A can increase the preference for B. Because B is now in the middle of the {A, B,
C} menu, it represents a compromise that is easy to justify.
The other heuristic we examine, anchoring (Tversky and Kahneman 1974), is the tendency to rely heavily, or anchor, on
one piece of information in order to arrive at a decision. For
instance, the willingness to pay for a product can be influenced
by an arbitrary anchor such as the last two digits of ones
Social Security number (Simonson and Drolet 2004). We
believe that such heuristics will be used more in decisions
involving time than in those involving money. We next discuss the reasons for our belief.

WHY WOULD HEURISTICS BE USED

MORE FOR TIME THAN FOR MONEY?
Decisions, such as arriving at the number of universities
to visit, could be based on a careful analysis in which one
considers factors such as availability of money, how much
it costs to visit one university, and knowledge about how
useful university visits are. Such thinking reflects an analytic
decision making in which costs and benefits are carefully
considered before arriving at a decision. This kind of decision making is consistent with the economists view of a
rational choice theory and is an integral part of many economic models relating search costs to behavior (Stigler
1961). However, the information processing approach (Bettman 1979) suggests that decision making may be qualitatively different if decision makers are limited in their capacity to process information methodically (Simon 1955).
Our review of the literature suggests that people differ in
their capacity to process time as opposed to money.
Studying time-money differences in how people consider
sunk costs, Soman (2001) shows that, while individuals account for past investments of money, they do not do so for
time. He argues that people face difficulties in accounting for
time because they do not routinely transact in time as they
do in money. Although people in some professions (e.g.,

915

lawyers) do keenly monitor their time expenditures, most

other people are not trained to do so. Consistent with this
logic, Soman (2001) finds that accounting for time can be
improved by various means, such as making a wage rate
available, increasing the salience of opportunity costs, or
teaching economic approaches to time. However, in the absence of such active steps, people find it difficult to account
for time. Beyond lack of transaction experience and training,
what seems to contribute to this difficulty with time is its
ambiguity. Okada and Hoch (2004) suggest that, unlike
money, which is unambiguousa dollar is a dollar in all
circumstancesthe value of time is malleable and cannot be
precisely assessed because its use may vary from one situation
to another. That is, the value of time is not concrete but instead
is open to interpretation. Okada and Hoch (2004) demonstrate
that, when people spend time (vs. money), they are more
prone to rationalizing their expenditures and displaying riskseeking behavior because they are not able to accurately assess
the opportunity costs of their time. Overall, these findings
suggest that temporal (vs. monetary) information is harder to
process. Research on decision making suggests that this could
lead to qualitative differences in how decisions are formed.
It is well known that, if people are limited in their capacity
to scrutinize information, they are less likely to be able to
identify and process the relevant and important pieces (Alba
and Hutchinson 1987) and less likely to analyze the information in order to arrive at a judgment or decision (Ratneshwar and Chaiken 1991). In such situations, instead of
continuing efforts to use this relevant information, people
make a qualitative shift to a different form of decision making
in which they rely on heuristics (Payne, Bettman, and Johnson
1993). This idea of relying on cognitive shortcuts instead of
a careful analysis of relevant information is also integral to
many dual-process models (see Chaiken and Trope [1999]
for a review). For instance, in the heuristic-systematic model
(Chen and Chaiken 1999), people make judgments based on
simple cues or rules of thumb if they are limited in their
cognitive ability or capacity to scrutinize judgment-relevant
information. Coupling these findings about heuristics being
used when information is hard to process with findings about
temporal (vs. monetary) information being hard to process,
we predict that people faced with temporal information will
rely relatively more on easy-to-use heuristics. For instance,
they will choose the middle of three available options because
it is a quick and easily justifiable compromise solution (Simonson 1989) or they will simply apply an anchor that they
have already formed (Tversky and Kahneman 1974).
Our prediction about time-money differences in the use
of heuristics is predicated on the idea that accounting is
harder to do for time than for money. However, Soman
(2001) has demonstrated that people can be encouraged to
account for time (e.g., by increasing the salience of opportunity costs), which, in turn, makes them consider time the
way they consider money. In line with this, we predict that,
when accounting is encouraged (vs. not encouraged) for both
time and money, the effect of heuristic use being higher for
time than for money will weaken. This will happen because

JOURNAL OF CONSUMER RESEARCH

916

encouraging accounting for time will facilitate the consideration of temporal expenditures and, therefore, reduce the
reliance on heuristics, whereas encouraging accounting for
money will have no influence because monetary expenditures are anyway considered carefully.
Finally, if encouraging accounting for time leads to a
deliberate analysis of temporal expenditure rather than the
use of a heuristic, people will take longer to arrive at a
decision. This increase in response time will occur because
analytic processes consume more time than heuristic processes (De Neys 2006). In contrast, encouraging accounting
for money will not change response times because people
anyway engage in an analysis of monetary expenditure.
We test these predictions in three experiments. In experiment 1, we create a forced-choice situation and measure
whether the heuristic of choosing a compromise option is
used more for time than for money. In experiment 2, we
test whether this effect also holds for the influence of anchors
on decisions. In experiment 3, we reexamine the anchoring
heuristic and test the moderating role of a prime that encourages accounting. We also examine whether data on response times provide convergent validity for the proposed
heuristic processes. Overall, these experiments support our
theorizing across different heuristics (compromise and anchoring) and different indicators of heuristic use (forced
choice, decisions, and response times).
In our experiments, we use the context of consumers
searching for information (Moorthy, Ratchford, and Talukdar
1997; Punj and Staelin 1983). Specifically, we focus on nonsequential search (Burdett and Judd 1983; Stigler 1961) in
which people make a search decision (e.g., number of stores
to visit) based on their knowledge about the expected search
costs (e.g., time to visit stores) and the potential benefits (e.g.,
finding the lowest-priced product). This context captures the
essential elements of our area of inquiry: how a priori decisions are made when time rather than money is to be spent.
Another reason to use this set-up is that, unlike purchase
situations in which money is the dominant currency, search
situations involve time as well. For instance, an individual
trying to sell a house could spend time searching for a buyer
or he could spend money on a real estate agent who finds
the highest bidder. Therefore, the context of search presents
an ecologically valid forum in which to study how consumers
make decisions for time and money.

EXPERIMENT 1
Overview
In experiment 1, participants considered searching for a
moving company. If one chose to search more (i.e., invite
more moving companies for estimates), the likelihood of
a desirable payoff (i.e., finding the cheapest moving company) increased but so did the search costs (i.e., cost of
inviting the moving companies). Participants had to choose
one option (i.e., number of companies to invite) from a
menu of options. Our prediction was that a change in the
menu (two vs. three options) would lead to a stronger

change in the relative share of the focal option if the currency used for search was time (vs. money). The focal
option was the lower search option in the binary set but
the compromise (between the lowest and the highest search
options) in the trinary set.

Design
A between-subjects design was used in which currency
(time vs. money) and menu of options ({3, 5} or {1, 3, 5})
were manipulated. The trinary set included one option (invite
one company), which was more similar to the option of three
companies than to that of five companies. Therefore, in the
absence of a compromise effect, the one-company option
would more adversely affect the share of the three-companies
option (vs. five-companies option). However, if people follow
the compromise effect, the share of the three-companies option will actually increase relative to the five-companies option (Simonson 1989) when the one company option is introduced. This change in the relative share of the threecompanies option (vs. the five-companies option) was the
dependent variable.

Procedure
In exchange for partial course credit, 211 undergraduate
students at the University of Texas at San Antonio participated
in this experiment. The participants read the following statement. The words in brackets denote the conditions to which
the participants were randomly assigned. We considered $5
to be roughly equivalent to 30 minutes. This wage rate of
$10 per hour is based on our pretests and is close to a wage
rate of $12.50 per hour used in recent work (Okada and Hoch
2004).
Imagine that you are moving to a different location and have
decided to hire a moving company. Because you have never
tried movers before, you ask a knowledgeable friend for help.
He picks up the Yellow Pages and tells you the names of 50
moving companies that provide a good level of service. He
adds that the prices could vary a lotanywhere from $500
to $1,000 for the amount of stuff you have. He therefore
suggests that you randomly choose some of these 50 moving
companies, get them to your apartment so that they can give
you their price estimates, and then pick the one that is the
cheapest.
The problem that you now face is to decide how many you
should get home for an estimate. To get the cheapest rate, the
best thing to do would be to let each of 50 companies visit
your apartment and give you an estimate so that you may pick
the cheapest one. However, there is a cost involved in terms
of the amount of [money/time] you spend on this activity. Each
moving company will [charge $5/take 30 minutes] to inspect
your stuff and provide an estimate.

Participants in the binary set condition then went on to

read the following:

HOW I DECIDE DEPENDS ON WHAT I SPEND

As you think about the number of moving companies you
should get over for estimates, you consider getting either 3 or
5 moving companies. So you need to make a decision now.
How many of the 50 moving companies will you get over for
estimates? (Please mark only one of the following two options.)

917
FIGURE 1
EXPERIMENT 1: EFFECT OF CURRENCY AND MENU
OF OPTIONS ON RELATIVE SHARE OF ONE
OPTION OF INVITING MOVING COMPANIES

__ I will ask 3 companies to come over for price estimates.

__ I will ask 5 companies to come over for price estimates.

Participants in the trinary set condition read a paragraph

similar to the one above except that they considered three
options: one, three, or five moving companies.

Results and Discussion

We employed a binary logistic regression model in which
we analyzed the proportion of participants choosing the threecompanies option over the five-companies option. The dependent variable was the relative share. The currency and the
menu of options were the between-subjects independent variables. Figure 1 depicts the percentage relative share ((number
choosing three companies number choosing either three or
five companies) # 100).
There was no main effect of either currency (b p .22;
Wald p .52; p 1 .47) or menu of options (b p .47;
Wald p 2.43; p 1 .11), but the predicted interaction was
significant. The menu of options # currency interaction
(b p 1.29; Wald p 4.62; p ! .05) confirmed that the
compromise effect emerged differently in the money condition versus the time condition. Specifically, when the
currency was time, the focal three-companies option was
chosen by 53.9% (28/52) participants in the binary condition and 78.0% (39/50) in the trinary condition. In line
with the compromise effect, this difference was significant
(z p 2.25; p p .01); the share of the three-companies
option relative to the five-companies option increased
when the one-company option was introduced. In contrast,
there was no statistical change in relative share when the
currency was money (z p .43; p 1 .67); the three-companies option was chosen by 64.2% (34/53) participants
in the binary condition and by 60.0% (30/50) in the trinary
condition.
When we offered an easy-to-pick compromise, people relied on that option more for time than for money, presumably
because, as we theorized earlier, temporal costs are generally
harder to process. However, could these results have been
driven, either partly or fully, by the nature of the payoff
(getting a low price)? Because the payoff was less congruent
with time than with money, it might have made it harder to
analyze temporal costs and thus have increased the reliance
on heuristics. To examine this issue, we conducted a followup experiment with a separate group of 162 undergraduate
students at the University of Texas at San Antonio in which
we used a payoff that is more congruent with time than with
moneydelivery time of moving company. The scenario presented to the participants was as follows. The words in brackets denote the conditions to which the participants were randomly assigned.

Imagine that you are moving to a different location and have

decided to hire a moving company that can deliver your stuff
within a convenient time frame. Because you have never tried
movers before, you ask a knowledgeable friend for help. He
picks up the Yellow Pages and tells you the names of 50
moving companies that charge very similar prices, but differ
in terms of the time frame in which they guarantee delivery.
Your friend also adds that moving companies guarantee delivery time only after they have had a look at the stuff to be
moved. He therefore suggests that you randomly choose some
of these 50 moving companies, get them to your apartment
so that they can give you their delivery-time estimates, and
then pick the one with the best delivery time.
The problem that you now face is to decide how many
you should get home for an estimate. You could ask each of
the 50 companies to visit your apartment to give you an
estimate. However, there is a cost involved in terms of the
amount of [money/time] you will spend on this activity. Each
moving company will [charge $5/take 30 minutes] to inspect
your stuff and provide an estimate.

Participants chose one option from either a binary or a

trinary set, just as in the earlier study. We analyzed the
proportion of participants choosing the three-companies option over the five-companies option. The results were encouraging. There was no main effect of currency (b p
.25; Wald p .33; p 1 .56). There was a main effect of the
menu of options (b p 1.03; Wald p 5.46; p ! .05),
which was qualified by a menu of options # currency interaction (b p 1.76; Wald p 3.98; p ! .05), confirming
that the compromise effect emerged differently in money
versus time. Specifically, when the currency was time, the
focal three-companies option was chosen by 64.1% (25/39)

JOURNAL OF CONSUMER RESEARCH

918

participants in the binary condition and 92.3% (36/39) in

the trinary condition. This increase was significant (z p
3.02; p ! .01). In contrast, there was no statistical change
in relative share when the currency was money (z p .27;
p 1 .39); the three-companies option was chosen by 76.9%
(30/39) participants in the binary condition and 79.5% (31/
39) in the trinary condition. Having observed robust results
for the compromise effect across two scenarios with different
payoffs, we now examine time-money differences in the anchoring heuristic.

EXPERIMENT 2
Overview
For experiment 2, participants considered the situation of
searching for quality information before purchasing a used
car. Although it would be desirable to get information about
each car under consideration, searching for such information
usually involves costs, as is evident by the presence of companies such as Carfax that sell vehicle history reports for
a price. If a participant chose to search more (i.e., view more
historical records), the likelihood of a desirable outcome
(i.e., finding a used car of acceptable quality) increased but
so did the search costs (i.e., cost of obtaining the records).
The classic two-step procedure of anchoring was used (Tversky and Kahneman 1974). First, participants were anchored
on a specific number by asking them whether the number
of records they would view would be less than or equal to
that number or more than that number. Then they were asked
to provide the exact number of records they would view.
Our prediction was that, when the currency is time rather
than money, the change in the anchor value (2 vs. 40) would
more strongly influence the decision regarding the number
of records to view.

Design
A between-subjects design was used in which the currency
(time vs. money) and the anchor value (low vs. high) were
manipulated. The dependent variable was the number of historical records that the participants decided to view.

(price range, brand name, etc.), and find that there are 80
cars that meet these criteria. You are now thinking about
viewing their historical records. However, the website informs you that it takes [$1/5 minutes] to view the details of
one historical record. Therefore, the more records you view,
the more [money/time] you spend on this website. You now
have to decide how many of these 80 historical records you
are going to view.
Will you view up to [2/40] records or will you view more
than [2/40] records? (Mark one option):
(a) I will view [2/40] records or less.
(b) I will view more than [2/40] records.
Exactly how many records will you view? (Write one number, not a range.)
I will view ____ records.

Results and Discussion

Two participants provided inconsistent responses to the
two questions, indicating that they had not read the questions
well (e.g., one participant marked I will view 2 records or
less in response to the first question and then wrote I will
view 5 records in response to the second question). Therefore, these two participants were excluded. The following
results are based on a sample size of 139 rather than 141.
We used a between-subjects analysis in which number-ofrecords was the dependent variable and currency and anchor
value were the between-subjects independent variables. Figure 2 depicts the results of the analysis of variance.
The currency # anchor interaction (F(1, 135) p 3.5;
p p .06) indicated that the number-of-records difference beFIGURE 2
EXPERIMENT 2: EFFECT OF CURRENCY AND ANCHOR
VALUE ON DECISION REGARDING THE NUMBER
OF CAR RECORDS TO VIEW

Procedure
One hundred and forty-one undergraduate students at
George Mason University participated in this experiment in
exchange for partial course credit. They read the following
statement. The words and numbers in brackets denote the four
conditions to which the participants were randomly assigned.
Youre planning to buy a used car within the next few days.
Your friends recommend that you consult a reputed used-car
website which offers extensive information that is hard to
find anywhere else. On this website, you can not only view
the listings of the used cars but can also view detailed historical records (accidents, water damage, number of previous
owners, etc.).
You go to the used-car website, put in your search criteria

NOTE.Numbers are means (standard deviations). Error bars show 95%

confidence intervals.

HOW I DECIDE DEPENDS ON WHAT I SPEND

tween the low and high anchor conditions was greater for
time (M p 14.6) than for money (M p 3.4). That is, participants in the time (vs. money) condition were more
sensitive to changes in the anchor. In fact, the effect of
anchor was significant only for time and not for money.
The number of records in the time condition was higher
(F(1, 135) p 13.5; p ! .001) when the anchor value was
high (M p 23.7) rather than low (M p 9.1) , but the number of records in the money condition was statistically
the same (F(1, 135) p .57; p 1 .44) irrespective of
whether the anchor value was high (M p 19.2) or low
(M p 15.8). Consistent with experiment 1, we demonstrate greater susceptibility to heuristics when the currency is time rather than money.

EXPERIMENT 3
Overview
The scenarios of experiment 3 were similar to that of experiment 2 except that the costs were higher ($3/15 minutes
instead of $1/5 minutes). The main aim was to provide evidence for the underlying process. Given our theorizing about
heuristic use differences occurring because people do not account well for time (vs. money), we predict that time and
money participants will no longer differ in their use of heuristics if they are encouraged to account for their expenditures.
In prior research, time accounting has been encouraged by
making a wage rate salient, increasing the salience of opportunity costs, or teaching economic approaches to time (Soman 2001). Because these manipulations are focused on encouraging accounting for time (not money), we could have
used them on only time participants to examine whether the
use of heuristics reduces. However, we could not have used
these manipulations on both time and money participants to
examine whether the heuristic use difference between time
and money goes down when accounting is encouraged. So
we used a more general procedure in which both time and
money participants wrote essays aimed at priming accounting
for their respective currencies. MediaLab was used for this
study because we also wanted to measure response times for
the number of records questiontime taken from the moment
this question was posed to the moment it was answered.

Design
A between-subjects design was used in which currency
(time vs. money), anchor value (low vs. high), and accounting prime (absent vs. present) were manipulated. The dependent variables were the number of records decision and
the associated response time.

Procedure
One hundred and forty-five undergraduate students at the
University of Texas at San Antonio participated in exchange
for partial course credit. If participants were assigned to the
time (or money) condition, they either saw the scenario as

919

in experiment 2 or first wrote an essay on the value of time

(or money) according to the following instructions. The
words in brackets denote the conditions to which the participants were randomly assigned:
Please take a couple of minutes to write a short essay on
the value of [time/money]. In this essay, please explain
how [time/money] is a resource that one should spend
carefully.

When accounting prime was manipulated to be present,

participants saw the above instructions, typed in their essays,
and pressed Continue. The computer then displayed the
used car scenario. When accounting prime was absent, participants were directly presented with the used car scenario.
At the bottom of the scenario, the anchor was manipulated
by asking participants to choose one of two options: they
marked whether they would view less than or equal to a
certain number of records (2 vs. 40) or more than that number. Then, on the next screen, participants were asked to
type in the exact number of records they would view.

Results and Discussion

As in experiment 2, some participants were excluded because their inconsistent responses (for the anchoring question vs. the number of records question) revealed that they
had not read the questions well. The following results are
based on a sample of 143 rather than 145.

Number of Records. We used a between-subjects analysis in which number of records was the dependent variable
and currency, anchor value, and accounting prime were the
between-subjects independent variables. Figure 3 depicts the
results of the analysis of variance.
The currency # anchor # accounting prime interaction (F(1, 135) p 2.8; p ! .1) indicated that the accounting prime influenced the pattern of results. When the
accounting prime was absent, the currency # anchor interaction (F(1, 135) p 3.4; p p .06) was similar to that
of experiment 2; the number of records difference between the two anchor conditions was greater for time
(M p 23.6) than for money (M p 8.4). More specifically,
the anchoring effect was significant only for time and not for
money; the number of records in the time condition was
higher (F(1, 135) p 16.3; p ! .001) when the anchor value
was high (M p 29.4) rather than low (M p 5.9), but the
number of records in the money condition was statistically
the same (F(1, 135) p 2.1; p 1 .15) irrespective of whether
the anchor value was high (M p 11.8) or low (M p 3.4).
When the accounting prime was present, however, these
time-money differences vanished. The currency # anchor
interaction was no longer significant (F(1, 135) p .3; p 1
.58); the number-of-records difference between the two anchor conditions was similar for time (M p 5.9) and money
(M p 10.3).
The heuristic use difference between time and money
vanished because, as predicted, priming accounting re-

JOURNAL OF CONSUMER RESEARCH

920
FIGURE 3
EXPERIMENT 3: EFFECT OF CURRENCY, ANCHOR VALUE,
AND ACCOUNTING PRIME ON DECISION REGARDING THE
NUMBER OF CAR RECORDS TO VIEW

full information, we included it in our analysis.) The dependent variable used was log-transformed response times. This
transformation was done to adjust for extreme response times
(Fazio 1990). However, the results reported below are substantively identical to the results we obtained without doing
any transformation. To ease interpretation, the means reported
below are the actual response times (in milliseconds).
As expected, the currency # anchor # accounting
prime interaction was not significant (F(1, 135) p .7;
p 1 .39). In fact, the only term that was statistically significant was the critical currency # accounting prime interaction (F(1, 135) p 8.2; p ! .01). Consistent with our
prediction, the accounting prime made time participants
deliberate more but did not influence money participants.
Specifically, for time participants, the response time was
greater (F(1, 135) p 8.9; p ! .01) when the accounting
prime was present (M p 14,912) than when it was absent
(M p 9,144). For money participants, the response time
was statistically the same (F(1, 135) p 1.1; p 1 .29) irrespective of whether the accounting prime was present
(M p 11,204) or absent (M p 13,065).
Viewing the two-way interaction from another perspective,
the data revealed an interesting cross-over that we did not
predict. In line with our theorizing about heuristic use being
relatively higher for time, the response time was marginally
lower (F(1, 135) p 2.9; p ! .1) for time (M p 9,144) than
for money (M p 13,065) when the accounting prime was
absent. However, when the accounting prime was present,
there was an unexpected reversal such that the response time
was now higher (F(1, 135) p 5.5; p ! .05) for time (M p
14,912) than for money (M p 11,204). With the benefit of
hindsight, these results make sense. If time is indeed harder
to process, accounting for time (when one is encouraged to
do so) ought to take longer than accounting for money.

GENERAL DISCUSSION

NOTE.Numbers are means (standard deviations). Error bars show 95%

confidence intervals.

duced the use of heuristics in the time condition but did

not influence the money condition. Specifically, for time,
the number-of-records difference between the two anchor
conditions was significantly lower (F(1, 135) p 4.5; p !
.04) when the accounting prime was present (M p 5.9)
than when it was absent (M p 23.6); for money, the number-of-records difference between the two anchor conditions was similar (F(1, 135) p .06; p 1 .80) irrespective
of whether the accounting prime was present (M p
10.3) or absent (M p 8.4).

Response Time. We used a between-subjects analysis in

which currency, anchor value, and accounting prime were the
between-subjects independent variables. (We did not expect
the anchor value to influence response times, but, to utilize

We demonstrated that decision making is more heuristic

in situations that involve spending time rather than money.
In experiment 1, a compromise heuristic was applied more
in time than in money. In experiment 2, similar results
emerged for the anchoring heuristic. In experiment 3, when
participants in both time and money conditions were primed
to account for their expenditures, they no longer differed in
their use of heuristics; priming reduced heuristic use in the
time condition but had no influence in the money condition.
Consistent with this, priming increased response times in
the time condition but had no influence in the money condition; time participants presumably shifted from heuristic
use to a more deliberate analysis that money participants
did anyway. Viewing the data from another angle, response
time in the prime-absent condition was lower for time (vs.
money), indicating more heuristic thinking in time. However, response time in the prime-present condition was
higher for time. It seems that, because time is harder to
process than money, doing accounting for time (when one
is encouraged to do so) takes longer than doing accounting
for money.

HOW I DECIDE DEPENDS ON WHAT I SPEND

It is important to revisit the issue of payoffs, which might

have been perceived to be more congruent with money than
with time. Our experiments demonstrated robust timemoney differences across different payoffs: price and delivery time of moving company in experiment 1 and quality
of used car in experiments 2 and 3. Moreover, experiment
3 provided direct evidence for the proposed process rather
than for congruency. By demonstrating that encouraging
accounting for time reduces the use of heuristics (whereas
encouraging accounting for money has no effect), we established that accounting differences are indeed at the heart
of our results. Could even this effect be due to congruency
such that encouraging accounting for time actually made
participants perceive time as money, which then led to a
greater perception of congruency with payoff? Responses
of 35 participants in the prime-present condition of time
indicate that is clearly not the case. Only three participants
interpreted time in relation to money, and they did so quite
vaguely. The other 32 participants penned philosophical discourses on the value of time and how it should be spent
carefully, and they did so without trying to monetize the
issue. We believe that congruency could be a valid theory
in itself and that it should be examined further in future
research. However, it does not speak to the current set of
results, which offer robust support for heuristic use differences between time and money, as well as the underlying
process of accounting differences. These results are significant from multiple perspectives.
Prior research on time and money has shown quantitative
differences in how information such as magnitude of risk
(Leclerc et al. 1995; Okada and Hoch 2004), prior investments (Soman 2001), or time delay (Soman 1998; Zauberman and Lynch 2005) is used to arrive at decisions. We
contribute to this stream by demonstrating that a qualitatively different form of decision makingrelying on quickand-easy heuristics rather than on an analysis of the available
informationgains prominence when one works with time
rather than with money.
We also add to the literature on the use of heuristics in
decision making. The reliance on heuristics has been found
to vary with such factors as involvement (Chaiken 1980),
constraints on processing capacity (Ratneshwar and Chaiken
1991), and type of emotion (Tiedens and Linton 2001). We
believe that the factor we explorecurrencyis significant
because a considerable portion of peoples spending decisions relate to time and therefore may be driven by heuristics. In this article, we have examined spending decisions
only in the context of search, but it would be interesting to
examine how our results extend to other domains.
Our results are also relevant for models of search in which
decisions are dictated by costs and payoffs (Stigler 1961).
Given that time seems to evoke different decision processes
than money, search theorists could also consider modeling
the currency in which a search is conducted.
Implications also arise for practice. People sometimes
have the choice to expend either money or time. When trying
to sell a house, one could either pay a real estate agent or

921

spend ones own time looking for the highest bidder. If

sellers spend time instead of money and therefore make
decisions differently, consequences arise for both consumers
(i.e., home buyers) and businesses (i.e., real estate firms).
To sum up, people often make spending decisions. Our
results suggest that the nature of the spent resource is important. Decisions related to time rather than money foster
an enhanced use of heuristics. The two seem economically
equivalent but are psychologically different.

REFERENCES
Alba, Joseph A. and J. Wesley Hutchinson (1987), Dimensions
of Consumer Expertise, Journal of Consumer Research,
13 (March), 41154.
Becker, Gary S. (1965), A Theory of the Allocation of Time,
Economic Journal, 75 (September), 493517.
Bettman, James R. (1979), An Information Processing Theory of
Consumer Choice, Reading, MA: Addison-Wesley.
Burdett, Kenneth and Kenneth Judd (1983), Equilibrium Price
Dispersion, Econometrica, 51 (July), 95569.
Chaiken, Shelly (1980), Heuristic versus Systematic Information
Processing and the Use of Source versus Message Cues in
Persuasion, Journal of Personality and Social Psychology,
39 (November), 75266.
Chaiken, Shelly and Yaacov Trope (1999), Dual-Process Theories
in Social and Cognitive Psychology, New York: Guilford.
Chen, Serena and Shelly Chaiken (1999), The Heuristic-Systematic Model in Its Broader Context, in Dual-Process Theories
in Social Psychology, ed. Shelly Chaiken and Yaacov Trope,
New York: Guilford, 7396.
De Neys, Wim (2006), Automatic-Heuristic and ExecutiveAnalytic Processing during Reasoning: Chronometric and
Dual Task Considerations, Quarterly Journal of Experimental Psychology, 59 (June), 10701100.
Fazio, Russell H. (1990), A Practical Guide to the Use of Response
Latency in Social Psychological Research, in Research Methods in Personality and Social Psychology, ed. Clyde Hendrick
and Margaret S. Clark, Newbury Park, CA: Sage, 7497.
Huber, Joel, John W. Payne, and Christopher Puto (1982), Adding
Asymmetrically Dominated Alternatives: Violations of Regularity and the Similarity Hypothesis, Journal of Consumer
Research, 9 (June), 9098.
Leclerc, France, Bernd H. Schmitt, and Laurette Dube (1995),
Waiting Time and Decision Making: Is Time Like Money?
Journal of Consumer Research, 22 (June), 11019.
Moorthy, Sridhar, Brian T. Ratchford, and Debabrata Talukdar
(1997), Consumer Information Search Revisited: Theory
and Empirical Analysis, Journal of Consumer Research,
23 (March), 26377.
Okada, Erica M. and Stephen J. Hoch (2004), Spending Time
versus Spending Money, Journal of Consumer Research, 31
(September), 31323.
Payne, John, James Bettman, and Eric Johnson (1993), The Adaptive Decision Maker, New York: Cambridge University Press.
Punj, Girish N. and Richard Staelin (1983), A Model of Consumer
Information Search Behavior for New Automobiles, Journal
of Consumer Research, 9 (March), 36680.
Ratneshwar, S. and Shelly Chaiken (1991), Comprehensions Role
in Persuasion: The Case of Its Moderating Effect on the Per-

922
suasive Impact of Source Cues, Journal of Consumer Research, 18 (June), 5262.
Simon, Herbert A. (1955), A Behavioral Model of Rational
Choice, Quarterly Journal of Economics, 69 (February),
99118.
Simonson, Itamar (1989), Choice Based on Reasons: The Case
of Attraction and Compromise Effects, Journal of Consumer Research, 16 (September), 15874.
Simonson, Itamar and Aimee Drolet (2004), Anchoring Effects
on Consumers Willingness-to-Pay and Willingness-toAccept, Journal of Consumer Research, 31 (December),
68190.
Soman, Dilip (1998), The Illusion of Delayed Incentives: Evaluating Future Effort-Money Transactions, Journal of Marketing Research, 35 (November), 42737.
(2001), The Mental Accounting of Sunk Time Costs: Why

JOURNAL OF CONSUMER RESEARCH

Time Is Not Like Money, Journal of Behavioral Decision
Making, 14 (July), 16985.
Stigler, George J. (1961), The Economics of Information, Journal of Political Economy, 69 (June), 21325.
Tiedens, Larissa Z. and Susan Linton (2001), Judgment under
Emotional Certainty and Uncertainty: The Effects of Specific
Emotions on Information Processing, Journal of Personality
and Social Psychology, 81 (December), 97388.
Tversky, Amos and Daniel Kahneman (1974), Judgment under
Uncertainty: Heuristics and Biases, Science, 27, 185 (September), 1124131.
Zauberman, Gal and John G. Lynch Jr. (2005), Resource Slack
and Propensity to Discount Delayed Investments of Time versus Money, Journal of Experimental Psychology: General,
134 (February), 2337.