Content uploaded by Michael Tsiros
Author content
All content in this area was uploaded by Michael Tsiros
Content may be subject to copyright.
Michael Tsiros & David M. Hardesty
Ending a Price Promotion:
Retracting It in One Step or Phasing
It Out Gradually
Using the literature on both pricing and regret, the authors develop a conceptual model of purchase likelihood and
propose a pricing tactic that appears to have marketplace potential. Sellers currently using a hi–lo pricing tactic
discount a product for a limited time and then raise the price back to its original level in one step. Here, the authors
investigate whether sellers should return prices to their prepromotion levels all at once or in steps. They propose
that sellers should consider an alternative tactic, labeled “steadily decreasing discounting” (SDD). This alternative
tactic requires that the seller offer one or more additional discounts that are less than the prior discount before
returning the product to its original price. Study 1 is a laboratory experiment that tests the proposed underlying
mechanisms (future price expectations and anticipated inaction regret) influencing likelihood to buy. In Study 2, an
additional laboratory experiment is undertaken to provide further empirical support in favor of the SDD tactic, to
address alternative explanations for the findings, and to demonstrate that there are no negative perceptions
associated with using SDD. Study 3 is a field experiment that assesses the effectiveness of SDD, and Study 4
examines scanner panel data to evaluate its generalizability.
Keywords: pricing tactics, regret, price promotions, expectations, hi–lo pricing, everyday low pricing
Michael Tsiros is Associate Professor of Marketing, School of Business
Administration, University of Miami, and Tassos Papastratos Research
Professor of Marketing, ALBA Graduate Business School, Athens,
Greece (e-mail: tsiros@miami.edu). David M. Hardesty is Associate Pro-
fessor of Marketing, Gatton College of Business & Economics, University
of Kentucky (e-mail: david.hardesty@uky.edu). The authors acknowledge
the valuable feedback from the anonymous JMR review team, as well as
Bill Bearden, Allan Chen, Blair Kidwell, Tatiana Levit, Kent Monroe,
Akshay Rao, Terry Shimp, and Danny Weathers. The authors contributed
equally to this research.
Price promotions have been demonstrated to be prof-
itable in the long run, and it has been suggested that
sellers should continue to employ them (Kopalle,
Mela, and Marsh 1999; Pauwels, Hanssens, and Siddarth
2002). In a review of reference pricing research, Mazumdar,
Raj, and Sinha (2005) conclude that how a firm frames a
price offer may influence a consumer’s decision to buy the
brand. Occasionally, after offering an initial price discount,
companies such as New York & Company, Boden Clothing,
and Time Warner Cable offer a series of additional price
promotions before returning the price of a product to its
original level. For example, Boden has advertised a series of
consecutive promotions, such as 15% off for three days, fol-
lowed by 13% off for a day, 11% off for a day, and 10% off
for a day, before returning merchandise to its original price.
These pricing practices raise the question whether prices
should be returned to their original level, after an initial
discount, all at once, as is typically the case for the hi–lo
pricing tactic, or in steps. In the current research, and con-
sistent with these examples, we propose that sellers could
offer one or more additional discounts that are smaller in
size than the prior discount before returning the product to
its original price. We label this practice “steadily decreasing
discounting” (SDD).1
Two particularly popular price promotion tactics are
everyday low pricing (EDLP) and hi–lo pricing. Sellers that
employ an EDLP tactic charge a constant, everyday price
with no (or very infrequent and small) temporary price pro-
motions (Monroe 2003). Alternatively, sellers that employ a
hi–lo pricing tactic set relatively higher prices on an every-
day basis but offer frequent and substantial price promo-
tions. Sellers use a hi–lo pricing tactic in an effort to dis-
criminate between price-sensitive and price-insensitive
consumers.
Given that many sellers employ hi–lo pricing, the cur-
rent research examines the relative effectiveness of the
alternative pricing tactic, SDD, versus the existing and more
often used hi–lo pricing tactic. In addition, Study 1 evalu-
ates the EDLP tactic. Drawing from literature on both
future price expectations and anticipated regret, we argue
that the increasing price trends associated with the SDD
tactic versus the hi–lo pricing tactic may increase con-
1Airlines typically start with low prices for a scheduled flight
and progressively increase the price as certain quotas are met and
the departure date gets closer. Although this pricing resembles
SDD, it does not start with the regular price and eventually reach
that price as the departure date approaches. In addition to the price
not having the same start and end level, prices may fluctuate
because reservations may be cancelled or the company reassigns
quota. Finally, airlines also often offer last-minute sales to reach
capacity. As such, SDD is not a dynamic pricing tactic (Desiraju
and Shugan 1999), nor is it being compared with dynamic pricing
in this research.
© 2010, American Marketing Association
ISSN: 0022-2429 (print), 1547-7185 (electronic)
Journal of Marketing
Vol. 74 (January 2010), 49–64
49
sumers’ purchase likelihood in the current period through
higher future price expectations (Jacobson and Obermiller
1990) and greater anticipated inaction regret (Sevdalis, Har-
vey, and Yip 2006; Tsiros 2009).
We begin our assessment of SDD by conducting a lab
experiment (Study 1) to test a theoretical framework regard-
ing the benefits of SDD’s effectiveness compared with the
hi–lo pricing and EDLP tactics; we show that SDD gener-
ates higher future price expectations and leads to greater
anticipated inaction regret, each of which affects purchase
likelihood. Consequently, we show that SDD generates
more revenue than the existing hi–lo and EDLP pricing tac-
tics.2In Study 2, we offer additional empirical support for
the effectiveness of SDD, examine alternative explanations
for the findings, and demonstrate that there are no negative
perceptions associated with employing the SDD tactic. In
Study 3, we conduct a field study that compares the relative
effectiveness of hi–lo pricing with that of the SDD tactic.
We show that revenues are higher when using the SDD tac-
tic. Finally, in Study 4, we assess scanner panel data to
uncover several instances of the effectiveness of the SDD
tactic. This research contributes to the pricing literature by
demonstrating that the theoretical mechanisms driving the
effectiveness of the SDD tactic relative to the hi–lo tactic
are future price expectations and anticipated inaction regret.
Moreover, this research has implications for managers in
that SDD appears to be a more profitable pricing tactic than
both hi–lo pricing and EDLP and is free of negative percep-
tions associated with using it. In the next section, we outline
the theoretical framework guiding our research hypotheses
(see Figure 1).
Conceptual Framework
Consumer Future Price Expectations
Research by Ariely (1998) and Hsee, Abelson, and Salovey
(1991) shows that evaluation of a stimulus is determined
not only by its position (the actual value of its outcome) but
also by its velocity (the change in the value). Given the pric-
ing pattern associated with SDD, we focus on consumer
expectations of future prices, an understudied, forward-
looking reference price (DelVecchio, Krishnan, and Smith
2007; Sun, Neslin, and Srinivasan 2003). Winer (1985)
finds that consumer expectations of future prices play a sig-
nificant role in purchase decisions. Similarly, and in line
with neoclassical economic theory, Jacobson and Ober-
miller (1990) suggest that consumers compare the sticker
price with expected future price. Consumers expecting
higher future prices are encouraged to purchase sooner;
those expecting lower future prices are more likely to wait.
On the basis of an analysis of 151 weeks of scanner data,
Jacobson and Obermiller find empirical support that con-
2We believe that SDD is most relevant for products with at least
a moderate profit margin and for products that are purchased infre-
quently (once a year or less often). In this article, we provide con-
sistent evidence using different products (a personal digital assis-
tant [PDA] in two lab studies and a wine stopper in a field study).
However, we also show some initial evidence in support of SDD
for grocery store items (e.g., soda, cereal).
Tactica
expectations
Future price
regret
Anticipated
likelihood
Purchase
FIGURE 1
Conceptual Framework
aTactic: hi–lo = 1, SDD = 2.
3Kalwani and Yim (1992) gather expected future price data and
assess purchases for a future occasion, as we do in Study 2,
instead of assessing purchase intentions for the current sale, as we
do in Study 1.
sumers conceptualize a reference price as an expectation of
future price. Higher future price expectations resulted in
increased quantity sold in the current period. More recently,
DelVecchio, Krishnan, and Smith (2007) examined the
effect of price promotion format, demonstrating that people
have higher future price expectations for price promotions
presented in percentage terms versus dollar terms. These
higher future price expectations result in greater choice of
the percentage-framed price promotions. Similar to DelVec-
chio, Krishnan, and Smith’s research, we investigate the
effect of expected future price on purchase likelihood dur-
ing both the current and the future promotion periods.3
The hi–lo and the SDD pricing tactics differ in terms of
both the individual discounts offered and the pattern of
prices. Prior research (Alba et al. 1999; DelVecchio, Krish-
nan, and Smith 2007) has demonstrated that deeper dis-
counts produce lower future price expectations. In the cur-
rent research, the average promotion size is held constant
across the two pricing tactics. As such, the impact of the
depth of the discounts offered is controlled for and should
not have a differential impact across the tactics. However,
the pattern of prices is different between the two pricing
tactics and may affect future price expectations. Impor-
tantly, prior research has established that price judgments
are affected by the ordering of past prices (Buyukkurt 1986;
Krishna 1991, 1994; Meyer and Assuncao 1990; Slonim
and Garbarino 1999).
Here, we expect that consumers will have higher future
price expectations for SDD versus hi–lo pricing due to the
greater prevalence of upward price trends when using the
SDD tactic. Adaptation-level theory (Helson 1964) suggests
that consumers judge current prices relative to their internal
norms (adaptation levels), representing the combined
effects of past, present, and future prices. For example, a
marketer employing the SDD tactic might sell a product
regularly at $499 and offer an initial large sale at $349 and
then two smaller sales at $399 and $449 before returning
the price to its original level. These additional sales result in
50 / Journal of Marketing, January 2010
4We assume that consumers expect prices to eventually reach
the regular price level and not exceed it. This is why SDD outper-
forms hi–lo. Otherwise, going from $349 to $499 in hi–lo may
indicate further, more drastic price increases than SDD’s smaller
increments.
cantly from the control group (those never missing the large
sale) in their likelihood to take advantage of the smaller sale
(Tykocinski and Pittman 2001). As such, inaction inertia
should favor SDD because the difference between the con-
secutive sales is smaller (e.g., always 10%, such as 40% off
versus 30% off or 30% off versus 20% off) than the one
experienced under hi–lo pricing (e.g., 40% off versus no
sale). More recently, Sevdalis, Harvey, and Yip (2006) dis-
tinguish between two types of anticipated regret: anticipated
inaction regret (regret anticipated to be experienced after
forgoing the second sale) and anticipated action regret
(regret anticipated to be experienced after buying the item
during the second sale). Across two studies, support for the
role of anticipated inaction regret in predicting likelihood to
buy was evidenced, while anticipated action regret did not
significantly impact purchase intentions.
Here, we expect that SDD will lead consumers to
experience higher levels of anticipated inaction regret at the
current sale price than hi–lo pricing as a result of their
expectations that the price will slowly return to its original
level. For example, if a product is regularly priced at $499
but recently sold for $349 and is now on sale for $399, we
expect consumers to anticipate regretting not buying the
product at $399 (anticipated inaction regret) because they
expect the price to eventually go back to the regular level of
$499. Given that SDD has additional weeks with an upward
trend in price, we expect that anticipated regret will be
greater for SDD than for hi–lo pricing. We also expect that
anticipated inaction regret will mediate the effect of future
price expectation on likelihood to buy.
In summary, we expect that higher future price expecta-
tions and more anticipated inaction regret associated with
the SDD tactic will lead to greater likelihood to buy. In
addition, we expect anticipated inaction regret to play a
dual role because it is also expected to mediate the effect of
future price expectation on likelihood to buy. This greater
likelihood to buy is expected to result in increased purchase
likelihood at higher prices for the SDD tactic than for the
hi–lo tactic as a result of anticipation that the price will
come back to the regular price in stages. As such, and in
addition to greater purchase likelihood, revenues are
expected to be greater when using the SDD tactic than the
hi–lo tactic. Formally, we propose the following:
H1: Compared with hi–lo pricing, SDD generates higher
revenue.
H2: Compared with hi–lo pricing, SDD leads to (a) higher
future price expectations and (b) more anticipated regret
from not buying the product (anticipated inaction regret).
Thus, SDD results in greater likelihood to buy.
H3: Anticipated inaction regret mediates the effect of future
price expectations on likelihood to buy.
Study 1 is a laboratory experiment that assesses the
relative effectiveness of SDD versus hi–lo pricing (i.e., H1)
and EDLP and tests the theoretical rationale (i.e., H2and
H3) pertaining to the expectation that SDD will outperform
hi–lo pricing. Study 2 attempts to resolve several limitations
associated with Study 1. In particular, Study 2 further eval-
uates the effectiveness of the SDD tactic, showing that there
are no negative effects associated with employing SDD over
Ending a Price Promotion / 51
more weeks containing an upward trend in price compared
with using hi–lo pricing, and consumers are expected to
have higher future price expectations as a result. Each of the
additional sales associated with the SDD tactic acts as a
price anchor, leading to an upward shift in price expecta-
tions and a new, higher adaptation level. This increased
adaptation level associated with the SDD tactic makes the
current price appear more attractive and results in greater
likelihood to buy for SDD than for hi–lo. For example,
when a product regularly priced at $499 is discounted to
$349 and then raised to $399, we expect that the upward
trend from $349 to $399 (and because it has not reached its
regular price of $499) will result in higher adaptation levels
and, thus, higher future price expectations.4
In addition to the advantage that SDD receives from its
upward price trend by shifting consumers’ adaptation levels
upward, price promotion research (Mace and Neslin 2004)
has identified the phenomenon of a postpromotion dip (a
significant drop in sales after a large discount is retracted).
Under SDD, we expect this phenomenon to be significantly
reduced compared with the hi–lo tactic because the price
returns to the regular price in stages.
Anticipated Regret
Another explanation for the success of SDD in generating
greater likelihood to buy after missing a previous sale is
consumers’ anticipating feelings of regret. Social psychol-
ogy and behavioral decision theory have given much atten-
tion to regret, but regret is only beginning to be fully exam-
ined in the marketing literature on purchase decisions
(Simonson 1992) and customer satisfaction and repurchase
intentions (Inman, Dyer, and Jia 1997; Inman and Zeelen-
berg 2002; Taylor 1997; Tsiros and Mittal 2000; Zeelenberg
and Pieters 1999). Simonson (1992) finds evidence that
anticipated regret influences brand preference and timing of
a purchase. Simonson asked consumers to anticipate how
they would feel if they passed on a current sale and learned
later that the price was higher. Compared with those who
were not asked to anticipate, consumers who anticipated
such a scenario were more likely to make an immediate
purchase than to wait for a better price. Anticipated regret
provides an important explanation for the success of SDD
in generating greater likelihood to buy after missing a previ-
ous sale.
The current research is also consistent with research on
inaction inertia, or the tendency of consumers to defer
choice following a missed sale (Tsiros 2009; Tykocinski
and Pittman 1998; Tykocinski, Pittman, and Tuttle 1995).
This stream of research has found that after consumers real-
ize that they missed a large sale (e.g., 40% off), they are less
likely to purchase a product at a significantly smaller sale in
the future (e.g., 10% off). When the difference between the
two sales (the one missed and the current one) is small (e.g.,
40% off versus 30% off), consumers do not vary signifi-
TABLE 1
Study 1 Experimental Conditions
Hi–Lo Pricing SDD Pricing
Regular Past Current Regular Past Current
Period Price Price Price Price Price Price
Week 1 $499 $499 $349 $499 $499 $349
Week 2 $499 $349 $349 $499 $349 $379
Week 3 $499 $349 $349 $499 $379 $409
Week 4 $499 $349 $499 $499 $409 $439
Week 5 $499 $499 $499 $499 $439 $469
Week 6 $499 $499 $499 $499 $469 $499
Week 7 $499 $499 $499 $499 $499 $499
Average $499 $435 $435 $499 $435 $435
5The discount sizes used are consistent with prior reference
pricing research (Grewal, Marmorstein, and Sharma 1996).
6The “weeks” terminology is arbitrary; we use it for simplicity
of explication. The periods could be time frames other than weeks
(e.g., days). In addition, each participant saw and responded to
prices for only one of the weeks displayed in Table 1. As such, we
make an assumption that consumers remember the most-recent
promotion. This appears to be a reasonable assumption given the
product category examined here (consumer electronics) and find-
ings from prior research indicating that consumers have some
knowledge about past deals (Dickson and Sawyer 1990; Krishna
1994; Le Boutiller, Le Boutiller, and Neslin 1994; Vanhuele and
Drèze 2002). However, we relax this constraint in Study 2 because
participants see 20 weeks of price data.
7We assumed a 100% conversion rate of the participants who
indicated their willingness to purchase the product at a given price.
Although this conversion rate is for explication purposes only,
note that we obtain similar results for any other conversion rate,
and we assume that conversion rate does not vary systematically
with price.
8After contacting store managers of two major electronics stores
and the headquarters of major manufacturers of PDAs, we deter-
mined that estimated profit margin for the retailer was 30%.
To empirically test the two hypotheses, we collected
data from 463 undergraduate business students who were
then entered into a cash prize raffle for their participation.
Because of several missing values, we removed three
respondents. There was a similar number of participants per
condition. To begin the experiment, participants were given
ascenario regarding the sale of an iPAQ 4155 PDA. The
Appendix shows an example of the scenario for the $499,
$379, $409 condition. After reading the scenario and look-
ing at two price advertisements, participants indicated
whether they would buy the PDA. In addition, they listed
their thoughts regarding their buying decision. Then, they
estimated what the price of the PDA would be one week
from now (Janiszewski and Lichtenstein 1999). Next, par-
ticipants completed two-item measures for store and brand
image. Each pair of items constituting the image measures
was significantly correlated (ps < .01), and we averaged the
items to form composite variables for each construct.
Finally, we obtained a single-item measure for anticipated
inaction regret. These scales appear in the Appendix.
Results
H1states that sellers will generate higher revenue when
employing an SDD pricing tactic rather than a hi–lo pricing
tactic. To assess this prediction, we first compared the reve-
nue generated across the weeks.7We calculated revenue as
the current price times the percentage willing to purchase at
the given price. As Table 2 shows, and consistent with H1,
the results reveal that across the weeks, SDD generated
$1,076.73 of revenue, while hi–lo generated only $990.88
(t406 = 3.55, p< .01; d = .35). Thus, the SDD tactic resulted
in 8.7% higher revenue. These findings support H1and pro-
vide initial evidence that SDD may be a viable pricing tac-
tic for sellers to employ. We next examined profits for the
case in which the product cost was assumed to be $349 (the
estimated cost to the retailer for the iPAQ 4155 PDA at the
time of the study), a 30% profit margin that is representative
of the PDA marketplace.8The SDD tactic generated
52 / Journal of Marketing, January 2010
a substantial period, and it assesses consumers’ likelihood
to visit a store that employs SDD versus hi–lo pricing.
Importantly, in both Studies 1 and 2, store and brand image
are compared across tactics because price promotion activ-
ity has been shown to negatively affect consumer percep-
tions (Grewal et al. 1998). Specifically, SDD offers an addi-
tional (though shallower) price promotion and, because of
the higher frequency of promotions, may result in more
negative perceptions. Finally, in Studies 3 and 4, a field
experiment is conducted and an evaluation of scanner panel
data is undertaken to demonstrate further the effectiveness
and generalizability of the SDD pricing tactic.
Study 1: Theoretical Assessment
Method
In an initial effort to investigate the effectiveness of SDD
versus hi–lo pricing and to evaluate the theoretical rationale
that might explain the relative effectiveness of SDD, we
manipulated prices between participants at several levels for
an iPAQ 4155 personal digital assistant (PDA) (see Table
1).5Each participant was randomly placed in one of the
week conditions displayed in Table 1, and each saw three
price points.6For example, the $499, $379, $409 condition
(see Week 3 of SDD in Table 1) represents a regular price of
$499, a most-recent past price of $379, and a current price
of $409. Importantly, we manipulated these prices so that
the average regular price ($499), the average most-recent
past price ($435), and the average current price ($435) were
the same for both tactics. Thus, this design allowed for a
fair comparison between the two tactics.
TABLE 2
Study 1 Results
Expected Percentage
Pricing Tactic Future Anticipated Likely to Potential Revenue Total
Condition Price Regret Buy per Person per Week Revenue
Hi–Lo $415.89 2.52
$499, $499, $349 $387.33 3.50 76% $349 ×.76 = $265.24
$499, $349, $349 $365.56 3.01 67% $349 ×.67 = $233.83
$499, $349, $349 $370.55 3.10 68% $349 ×.68 = $237.32
$499, $349, $499 $407.62 1.78 10% $499 ×.10 = $49.90 $ 990.88
$499, $499, $499 $462.56 2.12 14% $499 ×.14 = $69.86
$499, $499, $499 $462.09 2.10 14% $499 ×.14 = $69.86
$499, $499, $499 $455.55 2.06 13% $499 ×.13 = $64.87
SDD $430.14 3.32
$499, $499, $349 $382.54 3.52 74% $349 ×.74 = $258.98
$499, $349, $379 $402.18 4.37 62% $379 ×.62 = $234.98
$499, $379, $409 $411.83 3.78 50% $409 ×.50 = $204.50
$499, $409, $439 $442.40 3.54 30% $439 ×.30 = $131.70 $1,076.73
$499, $439, $469 $451.05 3.21 24% $469 ×.24 = $112.56
$499, $469, $499 $461.67 2.70 13% $499 ×.13 = $64.87
$499, $499, $499 $459.33 2.09 14% $499 ×.14 = $69.86
Random Discounting $419.51 2.79
$499, $499, $349 $383.61 3.47 75% $349 ×.75 = $261.75
$499, $349, $439 $449.88 3.60 30% $439 ×.30 = $131.70
$499, $439, $409 $390.44 2.91 38% $409 ×.38 = $155.42
$499, $409, $469 $475.08 2.68 21% $469 ×.21 = $98.49 $ 983.15
$499, $469, $379 $371.11 3.09 57% $379 ×.57 = $216.03
$499, $379, $499 $412.68 1.79 11% $499 ×.11 = $54.89
$499, $499, $499 $453.76 2.02 13% $499 ×.13 = $64.87
resulted in higher anticipated regret levels for not buying
now than hi–lo pricing (3.32 versus 2.52; t406 = 2.46, p<
.01; d = .24). These results suggest that the relatively higher
future price expectations and greater anticipated inaction
regret associated with SDD pricing should enhance its
effectiveness compared with the hi–lo tactic. Each of these
findings supports H2.
In addition to the preceding analyses, we content-
analyzed responses to the open-ended question that asked
participants to describe their thoughts leading to their deci-
sion. We expected hi–lo to generate a greater number of
lower future price expectations and SDD to generate greater
anticipated regret from not buying the product. We asked
two graduate students to code the responses on the basis of
the following categories: (1) lower future price expecta-
tions, (2) anticipated regret from not buying the product,
and (3) other thoughts that could not be classified in the
previous two categories. Interjudge agreement was 88%,
and all disagreements were resolved by a third judge. As
expected, we observed moderately more lower future price
expectations for the hi–lo pricing tactic than for the SDD
pricing tactic (46% versus 32%, χ2= 2.83, p< .10; w = .15)
and significantly more thoughts dealing with anticipated
regret from not buying the product for the SDD pricing tac-
tic than for the hi–lo pricing tactic (40% versus 12%; χ2=
13.81, p< .01; w = .34). A sample of lower-future-price-
expectation thoughts from the hi–lo $499, $349, $499 con-
dition includes “Why buy the PDA for $499 if I can buy it
for $150 less at a later time?” and “The fact that the PDA
has sold for $349 in the past led me to conclude there could
Ending a Price Promotion / 53
$144.90 in cumulative profit, while the hi–lo tactic only
resulted in $76.50 in cumulative profit per person. We also
observed similar results for profit margins within 10% of
the estimated profit margin for retailers.
Although SDD generated more profit, reducing prices
may negatively affect the image of both the brand and the
store; these variables have been shown previously to be
affected by price promotion activity (Grewal et al. 1998).
Because SDD involves two additional sales compared with
hi–lo, we examine the effect of both tactics on brand and
store image. For store image, we find no significant effects
across pricing tactics (4.76 versus 4.65; t406 = .49, p> .10).
Similarly, brand image was not significantly affected by the
SDD tactic (5.19 versus 5.22; t406 = –.21, p> .10). These
results are consistent with Monroe and Krishnan’s (1985)
finding that discounts on branded products may not affect
brand image. Overall, SDD benefits the seller through
higher revenue without any negative impact on store or
brand image.
H2posits that SDD will generate higher future price
expectations and result in higher levels of anticipated regret
for not buying the product so that SDD will lead consumers
to be more likely to buy than hi–lo pricing. To test this
hypothesis, we asked participants to estimate what the prod-
uct would cost in one week. In addition, we asked partici-
pants to determine how much regret they would feel if they
did not buy the product now (see the Appendix). Partici-
pants estimated next week’s average price to be $415.89 for
hi–lo pricing compared with $430.14 for SDD (t406 = 2.41,
p< .01; d = .24). Similarly, and as Table 2 shows, SDD
Tactica
expectations
Future price
regret
Anticipated
likelihood
Purchase
.24**
.30**
.43**
.19**
FIGURE 2
Study 1 Path Analyses
*p< .05.
**p< .01.
aTactic: hi–lo = 1, SDD = 2.
A: Initial Path Model
B: Full Path Model
Tactica
expectations
Future price
regret
Anticipated
likelihood
Purchase
.21**
.31**
.42**
.11*
.33**
because EDLP has no price volatility. Forty-four undergrad-
uate student participants from the same population were
assigned to the EDLP condition. For the EDLP condition,
the normal price, the most-recent past price, and the current
price were all $435. Thus, both the average most-recent past
price and the average current price were the same for the
hi–lo and SDD pricing conditions. The results revealed that
27% of participants were likely to buy, yielding revenues of
$829.71 across the seven weeks for EDLP compared with
$990.88 in revenues generated from hi–lo pricing and
$1,076.73 from SDD. Thus, EDLP resulted in 16.3% lower
revenues than hi–lo pricing and 22.9% lower revenues than
SDD. These results rule out the price volatility alternative
explanation because EDLP would be superior if price
volatility explained the effects. Moreover, these results sug-
gest that the SDD tactic results in higher revenues than
EDLP as well.
To further test the proposed mechanism advanced here,
we extended our study by varying the order of the interme-
diate steps for SDD. In this design, we modified the order
of the steps for SDD presented in Study 1 (see Table 2). The
only difference in the design of the study was in the four
intermediate discounts, which in this case did not follow a
strict “steadily decreasing” trend, though on a couple of
occasions the discounts were larger than the one offered in
54 / Journal of Marketing, January 2010
be another sale and I would rather wait and save $150.” An
example of an anticipated-inaction-regret thought in the
SDD $499, $349, $379 condition was “$379 is not that
much more than the sale price of $349, and you never know
when it will be that price again.” Another respondent in the
SDD $499, $439, $469 condition wrote, “The price would
go up now if I don’t take it now and I would miss an
opportunity.”
On the basis of our findings thus far, SDD appears to
perform better than hi–lo pricing because of the higher
future price expectations and increased anticipated inaction
regret associated with the SDD price combinations. In addi-
tion to the preceding analyses, we performed a path analysis
(see Figure 2) to test the mediating effect proposed in H3.
First, and in support of H2, “tactic” (hi–lo = 1, SDD = 2; see
Figure 2, Panel A) had a positive influence on both expected
future price expectations (.30, p< .01) and anticipated
regret (.24, p< .01). As we expected, both future price
expectations and anticipated regret were positively related
to purchase likelihood (.19, p< .01, and .43, p< .01,
respectively). In addition, when we included a path from
future price expectations to anticipated regret (see Figure 2,
Panel B), we observed a positive relationship between
future price expectations and anticipated regret (.33, p<
.01). Consistent with H3, it appears that anticipated inaction
regret plays a dual role in influencing purchase likelihood.
First, and consistent with prior studies (Sevdalis, Harvey,
and Yip 2006), anticipated inaction regret has a direct effect
on purchase likelihood. Second, anticipated inaction regret
partially mediates the effect of future price expectations on
purchase likelihood. In the model in Figure 2, Panel A,
when the path from future price expectations to anticipated
regret is set to zero, the direct effect from future price
expectations on purchase likelihood is more significant (.19,
p< .01) than when the path from future price expectations
to anticipated regret is estimated (.11, p< .05). Thus, antici-
pated inaction regret partially mediates the relationship
between future price expectations and purchase likelihood
because the coefficient for this path is reduced (from .19 to
.11) but is still significant.
Several researchers (Baron and Kenny 1986; Sobel
1982) have proposed a more formal test of mediation. Con-
ducting a Sobel (1982) test (and the Aroian version of the
Sobel test popularized by Baron and Kenny [1986]), we
find a significant mediation effect of anticipated inaction
regret on the relationship between future price expectations
and purchase likelihood (Sobel: z = 2.33, p< .05; d = 28;
Aroian: z = 2.31, p< .05; d = 28). Thus, H3is supported.
Although our results provide support for our conceptual
framework, suggesting that future price expectations and
anticipated inaction regret are the underlying mechanisms
for SDD’s success over hi–lo pricing, it is possible that
SDD outperforms hi–lo pricing because there is less price
volatility associated with the SDD tactic. Price volatility in
Study 1 (as measured by the standard deviation) was 80.2
for hi–lo and 58.6 for SDD. To rule out price volatility as an
alternative explanation for the superiority of SDD over
hi–lo pricing, we also empirically evaluated EDLP in this
study. If price volatility drives the results related to SDD
and hi–lo pricing, EDLP should be preferred over both
an opportunity to evaluate whether one tactic generates
greater store traffic than the other. Finally, and similar to
Krishna’s (1994) assessment of the certainty consumers
associate with a deal occurring, Study 2 examines whether
SDD generates higher levels of price certainty than hi–lo.
That is, if SDD generates greater price certainty, this could
be an alternative explanation for its relative effectiveness.
Study 2: Price History Design
Method
Having established the underlying theoretical mechanisms
(future price expectations and anticipated inaction regret)
associated with the effectiveness of SDD in Study 1, we
conducted Study 2 to further investigate the relative effec-
tiveness of SDD versus hi–lo pricing and to address the lim-
itations associated with Study 1. Specifically, we manipu-
lated the price histories that participants were exposed to
using similar discount sizes to those used in Study 1. Par-
ticipants were randomly assigned to one of five conditions
(see Table 3). Three of the conditions involved a single-
store pricing tactic, and two conditions involved two-store
pricing tactics. For example, in the single-store conditions,
each participant saw only one pricing tactic (A, A′, or B).
Condition A involved a hi–lo pattern (four sales of $349).
Condition A′was a variation of hi–lo involving the same
number of promotions as in SDD (six sales of $399).
Finally, Condition B involved the SDD pattern of prices
(two sets of the following pattern of sales: $349, $399,
$449). In addition, we included two more conditions in
which participants observed two stores (one following the
hi–lo and the other the SDD pricing pattern). In one condi-
tion (AB), the store that used hi–lo pricing included four
sales (A), and the store that used the SDD tactic (B)
included six sales (see Table 3). In the other condition
(A′B), both stores had six sales, with Store A′using hi–lo
pricing and Store B using SDD. Importantly, and for all
conditions, each store had the same average price ($469)
across each 20-week period.
We collected data from 247 undergraduate business stu-
dents who were entered into a cash prize raffle for their par-
ticipation. Participants began by reading a scenario and
looking at price information available from two stores (or
one store in the three single-store conditions). Under the
scenario, participants were told to imagine that they had
consulted pricetrack.com (a fictitious Web site) to gather
past prices for a PDA described in the scenario and that the
Web site revealed the 20 most-recent weekly prices for the
PDA at two stores (A and B, or A′and B) or at one store (A,
B, or A′). After evaluating the price information, partici-
pants were asked to assume that they had gone home to visit
family during a school break. Returning to school a few
weeks later, they had gone to pricetrack.com to check prices
again because they wanted to purchase the PDA, but the
Web site was no longer available and had not been available
while they were away. After reading this scenario, partici-
pants indicated the highest price they were willing to pay
for the PDA, their likelihood of visiting the store, their best
estimate of the price of the PDA after they went away, and
Ending a Price Promotion / 55
the previous promotion period. We collected data from 196
undergraduate students from the same population as previ-
ously; they saw the same stimuli. Each of the seven condi-
tions had the same number of participants. The results show
that the new pricing tactic, without the steadily decreasing
trend in the discounts, performs similar to hi–lo and worse
than SDD ($983.15 versus $1,076.73; t397 = 3.88, p< .01;
d = .39). Similarly, participants estimated next week’s aver-
age price to be $419.51 compared with $430.14 for SDD
(t397 = 2.09, p< .05; d = .21), and anticipated regret levels
from not buying now were 2.79 compared with 3.32 for
SDD (t397 = 2.25, p< .05; d = .23). If we take as an exam-
ple the condition in which the regular price is $499, the last
price was $439, and the current price is $409, even with a
higher last price than the SDD condition with the same cur-
rent price ($499, $379, $409), future price expectations are
lower ($390.44 versus 411.83), anticipated inaction regret is
lower (2.91 versus 3.78), and likelihood to buy is lower
(38% versus 50%). In this case, consumers might expect the
price to keep going down. Therefore, we argue that there is
something fundamental about the order and magnitude of
the price increments that offer a signal to consumers that
the price is indeed in a trajectory that will reach the regular
price, and this signal drives both higher future price expec-
tations and greater anticipated inaction regret, which lead to
greater likelihood to buy and higher revenue.
In summary, Study 1 provides initial empirical support
in favor of SDD and the conceptual model. However, this
study has a few limitations that must be addressed. First, the
study limited participants in terms of the number of prices
received. Study 2 addresses this limitation by employing 20
weeks of price information to allow for a more complete
assessment of image perceptions. These new stimuli pro-
vide a stronger test of the potentially deleterious impact on
image perceptions because it has been shown that such
assessments are long-term (Blattberg, Briesch, and Fox
1995; Mela, Gupta, and Lehmann 1997).
A second limitation is that we considered only one store
and one pricing tactic at a time. Study 2 addresses this lim-
itation by giving some participants historical pricing infor-
mation from two stores, one employing a hi–lo pricing tac-
tic and the other employing SDD. A third limitation is that
the hi–lo pricing condition included three large discounts of
$150, while SDD employed five discounts (one at $150,
one at $120, one at $90, one at $60, and one at $30). In
Study 2, we include a condition in which both tactics
employ the same number of discounts to assess the poten-
tial alternative explanation of discount frequency (Alba et
al. 1999). Moreover, by including an equal number of dis-
counts, we alleviate concerns about potential administrative
cost differences associated with the SDD tactic. In particu-
lar, and as we tested in Study 1, SDD would require store
managers to make more price changes and potentially incur
higher advertising expenses because more sales would need
to be promoted. In addition, Study 2 assesses participants’
willingness to pay and likelihood to visit the store in the
future after exposure to 20 past prices for the SDD and
hi–lo pricing tactics. The willingness-to-pay assessment
allows for another comparison of the effectiveness of SDD,
and our measure of the likelihood to visit the store provides
TABLE 3
Study 2 Experimental Conditions
Conditions
AB(Hi–Lo and SDD) A′B(Hi–Lo′and SDD)
Period A(Hi–Lo) A′(Hi–Lo′) B(SDD) A(Hi–Lo) B(SDD) A′(Hi–Lo′) B(SDD)
Week 1 VaVVVVVV
Week 2 VVVVVVV
Week 3 VXVVVXV
Week 4 VVVVVVV
Week 5 VVVVVVV
Week 6 WX WWWX W
Week 7 VVXVXVX
Week 8 VVYVYVY
Week 9 WX V WV X V
Week 10 VVVVVVV
Week 11 VVVVVVV
Week 12 VXVVVXV
Week 13 VVVVVVV
Week 14 WV V WV V V
Week 15 VXWVWXW
Week 16 VVXVXVX
Week 17 WV Y WY V Y
Week 18 VVVVVVV
Week 19 VXVVVXV
Week 20 VVVVVVV
Average ZZZZZZZ
aV: $499, W: $349, X: $399, Y: $499, Z: $469.
TABLE 4
Study 2 Results
Conditions
AB(Hi–Lo and SDD) A′B(Hi–Lo′and SDD)
Measures A(Hi–Lo) A′(Hi–Lo′) B(SDD) A(Hi–Lo) B(SDD) A′(Hi–Lo′) B(SDD)
Willingness to pay $385.25 $392.76 $435.63a,b $393.95 $393.95 $371.29 $371.29
Likelihood to visit store 3.86 3.88 4.61a,b 4.13 5.07a4.05 5.10b
Store image 4.27 4.47 4.62 4.94 4.92 4.99 4.92
Brand image 5.39 5.60 5.53 —c—c—c—c
aIndicates significant differences (p< .05) between conditions B and A within the same row.
bIndicates significant differences (p< .05) between conditions B and A′within the same row.
cBrand image measures were not included in the AB and A′B conditions because participants were exposed to both pricing tactics and their relative impact on
brand image is confounded.
In the two-store conditions, we estimated only one willing-
ness to pay for the PDA (as opposed to one for each store),
and both versions of hi–lo (AB and A′B) generated similar
results ($393.95 versus $371.29; t86 = 1.22, p> .10,
respectively).
Likelihood to visit the store. Next, we wanted to assess
whether the SDD tactic resulted in greater likelihood to
visit the store than the hi–lo pricing tactic. We evaluated
this by assessing participants’ likelihood to visit each store
after experiencing (1) each tactic for a period of 20 weeks
or (2) both tactics for a period of 20 weeks. Respondents
were asked how likely they were to visit the store (7 = “very
likely,” and 1 = “very unlikely”). Compared with the SDD
(B) condition’s average store traffic (4.61), the hi–lo (A)
condition averaged 3.86 (t92 = 2.71, p< .01; d = .57) and
the hi–lo (A′) condition averaged 3.88 (t90 = 2.01, p< .05;
d = .42). As with willingness to pay, store traffic was similar
for the two hi–lo conditions (A and A′). In the AB condi-
56 / Journal of Marketing, January 2010
their level of certainty regarding their price estimate. Then,
respondents indicated their image of the stores along with
their image of the brand using the two-item measures from
Study 1 (see the Appendix).
Results
Willingness to pay. Table 4 includes all the results for
Study 2. We first evaluated willingness to pay across the
two pricing tactics. We assessed willingness to pay here
instead of likelihood to buy (as in Study 1) because respon-
dents were not presented with a current price and, therefore,
a decision of whether to buy. The results revealed that SDD
generated a higher willingness to pay than both versions of
hi–lo (B versus A: $435.63 versus $385.25; t92 = 3.42, p<
.01; d = .71; and B versus A′: $435.63 versus $392.76; t90 =
3.43, p< .01; d = .72). These findings provide further sup-
port for the relative effectiveness of the SDD tactic. More-
over, willingness to pay was not significantly different
between Conditions A ($385.25) and A′($392.76, p> .10).
9Consistent with our predictions and the results of Study 1, the
expected price for hi–lo (A) was $462.12; under hi–lo (A′), it was
$458.11; and under SDD, it was $489.31. In other words, SDD led
to significantly higher future price expectations than both versions
of hi–lo (F2, 119 = 4.03, p< .05, χ2= .12). In addition, SDD led to
a higher future price expectation in the A′B condition (469.88 ver-
sus 444.59; t66 = 2.01, p< .05; d = .52) but not in the AB condition
(471.03 versus 452.85; t59 = 1.16, p> .10; d = .30).
10We did not include brand image measures in the AB and A′B
conditions because participants were exposed to both pricing tac-
tics, and their relative impact on brand image is confounded.
In the two-store condition (AB), in which the hi–lo tac-
tic has four promotions and SDD has six promotions (this
replicates Study 1), store image averaged 4.94 for the hi–lo
store and 4.92 for the SDD store (t44 = .20, p> .10). Thus,
we observed no significant differences in store image across
tactics. Importantly, these results suggest that the SDD tac-
tic yields no additional negative long-term impact on store
image. However, it is possible that SDD benefits in the AB
condition by having more sales (i.e., six versus four) than
the hi–lo tactic. For the A′B condition, store image averaged
4.99 in the hi–lo store and 4.92 in the SDD store (t41= .54,
p> .10). Again, we observed no significant differences in
store image across tactics. Store image appears to be similar
for the two tactics when we used the longer 20-week time
frame as well as when an equal number of promotions were
employed. Moreover, the store image values were similar
for the AB and A′B hi–lo conditions.
In summary, across Studies 1 and 2, we provided par-
ticipants with past price information and asked them to
indicate their future price expectations and anticipated inac-
tion regret to assess the relative likelihood to buy and will-
ingness to pay for a product when using the SDD versus
hi–lo pricing tactic. Collectively, these two studies revealed
that SDD generates greater revenue and profit than hi–lo
pricing and EDLP and has no deleterious effects associated
with negative assessments of brand or store image. More-
over, we established that SDD appears to be effective for
both short and long patterns of price promotion and that
future price expectations and anticipated inaction regret are
the underlying mechanisms driving these effects. Note that
Studies 1 and 2 used different dependent variables, and we
did not measure anticipated inaction regret in Study 2, pre-
cluding a further assessment of the underlying mechanism
proposed. In addition, although Study 1 provided some evi-
dence against price volatility as a possible explanation for
the improved performance of SDD, note that though SDD
had less price volatility than hi–lo (52.3 versus 61.6 stan-
dard deviations, respectively), this was not the case com-
pared with hi–lo (47.0). This finding offers further evidence
that price volatility may not be a feasible alternative expla-
nation for the results.
Next, in Study 3, to further test the generalizability of
the SDD tactic, we assess the relative effectiveness of SDD
versus hi–lo pricing by adapting existing industry examples
of the SDD tactic. We provide consumers with the product’s
regular price and a series of upcoming prices to assess SDD
in a field setting. Study 3 improves on Studies 1 and 2 by
testing the effectiveness of SDD in an actual consumption
setting with greater external validity and by using a differ-
ent presentation of the price promotions in an attempt to
demonstrate another condition in which SDD may be more
effective than the hi–lo pricing tactic.
Study 3: Field Study
The site for this study was an upscale kitchen appliance
store that is located in a small and wealthy suburb of a large
metropolitan area in the United States (population:
>30,000; median age: 34 years; median household income:
~$66,000; education level: >80% high school, >40% with a
Ending a Price Promotion / 57
tion, participants averaged 5.07 for the SDD store and 4.13
for the hi–lo store (t44 = 3.34, p< .01; d = 1.03). Thus, par-
ticipants were significantly more likely to visit a store using
the SDD tactic than one using the hi–lo pricing tactic. For
the A′B condition, participants averaged 5.10 for the SDD
store and 4.05 for the hi–lo store (t43 = 3.47, p< .01;
d = 1.06), again showing significant differences in likeli-
hood to visit in favor of the SDD store. Thus, the SDD
tactic appears to increase the potential traffic a store experi-
ences relative to the hi–lo pricing tactic. Moreover, store
traffic for the hi–lo tactic was similar for the AB and A′B
conditions. Importantly, these results occur when con-
sumers are made aware of prices. Recall that in Study 1, we
assumed that store traffic was constant across SDD and
hi–lo. As such, the Study 2 results suggest that the Study 1
results are conservative given that SDD may lead to an
increase in store traffic if promotions are also advertised.
Price uncertainty. To examine the level of price uncer-
tainty generated by the two pricing tactics, we first asked
participants to estimate the price of the PDA a week after
they went away for the break. Then, we asked participants
to state their certainty with their price estimates (see the
Appendix). A potential alternative to SDD generating
higher future price expectations is that SDD may generate
less uncertainty than hi–lo in consumer price estimates.
However, the study results indicate that SDD (B) generated
the same level of certainty as both hi–lo versions (A and A′)
(84% versus 80% versus 77%, F2, 119 = 1.49, p> .10). As
such, price uncertainty does not appear to be driving differ-
ences between SDD and hi–lo pricing.9
Store and brand image. We included store image and
brand image measures in the single-store conditions (A, B,
and A′), but we included only store image measures in the
two-store conditions (AB and A′B).10 Compared with the
SDD (B) condition’s average store image (4.62), the hi–lo
(A) condition averaged 4.27, and the hi–lo (A′) condition
averaged 4.47 (t94 = 1.93, p< .10; d = .40, and t88 = .69, p>
.10). Thus, there were no significant differences in store
image across stores A′and B. However, the SDD store (B)
had a slightly higher store image level than the hi–lo store
(A). Store image did not vary between the two hi–lo condi-
tions (A and A′). Similarly, compared with the SDD (B)
condition’s average brand image (5.53), the hi–lo (A) con-
dition averaged 5.39, and the hi–lo (A′) condition averaged
5.60 (t90 = .77, p> .10, and t90 = –.36, p> .10, respec-
tively). Thus, we observed no significant differences in
brand image across tactics. In addition, brand image was
similar for the two hi–lo conditions (A and A′).
11Note that the same-depth hi–lo was offered for two-thirds of
the time; the price was $17.45 for two days, and on the third day,
it was set to the regular price ($24.95) to maintain the same aver-
age price across all three conditions ($19.95).
12However, note that it is not clear from observing the data that
the store policy is to use SDD. We are not able to deduce whether
this pricing pattern is due to a conscious choice by the store
manager or to other factors (e.g., trade deals).
stoppers (10 at $17.45 and 3 at $24.95); and during the
SDD promotion period, the store sold 24 wine stoppers (14
at $17.45, 6 at $19.95, and 4 at $22.45). The product costs
the store $12.475 and has a 100% profit margin. Compared
with when no promotion is offered, same-frequency hi–lo
increased sales by 75%, same-depth hi–lo increased sales
by 63%, and SDD increased sales by 200%. Importantly,
the increase in sales associated with SDD relative to same-
frequency and same-depth hi–lo is statistically significant
(t58= 2.18, p< .05; d = .57, and t58= 2.41, p< .05; d = .63,
respectively). Thus, H1is supported. In addition to the sales
results, compared with when no promotion tactic was
offered, same-frequency hi–lo led to a 5% increase in profit,
same depth hi–lo led to a 12% decrease in profit, and SDD
led to a 55% increase in profit. Thus, SDD performs better
than the more established hi–lo pricing tactic. In the next
study, using available scanner panel data, we provide anec-
dotal evidence that SDD can be an effective pricing tactic
even in grocery store settings.
Study 4: Dominick’s Finer Foods
Data
We performed a final test for SDD by examining the
Dominick’s Finer Foods data sets, which have been widely
used in marketing (Mace and Neslin 2004). These data sets
include weekly sales volume, price, and profit data at the
stockkeeping unit (SKU) level from several product cate-
gories (e.g., soft drinks, cereal, analgesics, beer) across 399
weeks (1989–1997) from Dominick’s stores in the Chicago
metropolitan area. We begin by examining colas, which rep-
resent 34 SKUs from the soft drink category. To avoid
aggregation of the data, which could lead to erroneous con-
clusions about the pricing tactic of each store, we focus our
analysis in the most popular store.
An examination of the data set shows that on some
(albeit rare) occasions, the store manager was already using
an SDD pricing tactic.12 For example, examining the prices
for two-liter bottles of Pepsi, we observe 14 occasions when
the price was brought back to the regular level in two or
more steps. For the six-pack of Pepsi 12-ounce cans, we
observe 12 occasions of SDD. These represent approxi-
mately 14% of the weeks. In each of these occasions, no
major holiday was included in any of the weeks for SDD or
hi–lo pricing.
The next step was to find patterns that would enable us
to compare the revenues of the store when using SDD
instead of hi–lo pricing patterns. For example, for two-liter
bottles of Pepsi, there was only one such case in which the
patterns were comparable: Both patterns begin with the
same regular price of $1.59, lower the price to the same
level of $1.09, and eventually return it to the regular level.
Note that we were unable to control for other important
variables (e.g., depth and frequency of sales before the
examined period). Thus, the results we present here are not
58 / Journal of Marketing, January 2010
bachelor’s degree or better, and >16% with a master’s
degree or better, according to the 2000 census). The product
category selected was wine bottle stoppers, which are sold
in the store in different styles for a regular price of $24.95.
We selected this product because of several characteristics,
such as the store was allowed to offer promotions by the
manufacturer, the product was fairly popular, and the price
was not too high, which made the study financially manage-
able. There had been no other promotions in the focal cate-
gory all year. In addition, during the promotion periods, all
other activity in the store (e.g., number of salespeople)
remained constant.
Design and Procedure
The store used two separate pricing tactics (hi–lo and SDD)
during the test period. These pricing tactics were alternated
every week for a period of 30 weeks. The average price of
the product was kept constant between the promotion tac-
tics. In addition, after discussing the procedure with the
store owner, we determined that customers do not visit the
store weekly, and thus we decided to run the promotions on
a weekly basis and to alternate the two tactics.
To be consistent with the similar instances of the com-
panies mentioned in the introduction (New York & Com-
pany, Boden Clothing, and Time Warner) and to use dis-
count sizes consistent with prior research (Grewal et al.
1996), SDD was run at 30% off the first day, 20% off the
second day, and 10% off the third day. Hi–lo was run in two
versions (at the same frequency as SDD): three days at 20%
off and at lower frequency but similar depth as SDD and
two days at 30% off. More specifically, the wine stopper,
which was regularly priced at $24.95, was discounted under
the SDD tactic at $17.45, $19.95, and $22.45 before return-
ing to the regular price. Under same-frequency hi–lo, the
product was discounted at $19.95 for all three days, and
under same-depth hi–lo, the product was discounted at
$17.45 for two days and returned to the original price of
$24.95 on the third day.11 The SDD and hi–lo pricing tactic
stimuli used in Study 3 appear in the Appendix.
Similar to the previous study, by allowing hi–lo to have
two versions, we can test both the effect of frequency of
promotion and the depth of promotion. All three conditions
were run for the same number of weeks (10) and had the
same average price ($19.95) across the three days. As such,
each tactic was run for 30 days, and sale signs were
removed during the nonpromotion periods. The total store
sales volume during the 30 promotional periods was similar
across conditions.
Results
On average, under no promotion, the store sells 8 wine stop-
pers during a 30-day period (the length of time that each
tactic was run). During the same-frequency hi–lo promotion
period, the store sold 14 wine stoppers at $19.95; during the
same-depth hi–lo promotion period, the store sold 13 wine
13Note that the regular price and the low price of soft drinks
changed several times during the span of the seven years included
in the data set. Although the hi–lo tactic was used several times,
the changes in the band of price points limited the number of
testable occasions.
average temperature was higher during the average hi–lo
promotion than during the SDD promotion (57°F versus
53°F; p< .05), making these tests conservative.
To assess whether this evidence of SDD was only a
single-store phenomenon or whether there were other stores
employing SDD pricing, we assessed the same four SKUs
for the next three largest stores in the soft drink category.
We found that each of these stores was also practicing SDD
at some limited level (approximately 10% of the time) with
similar results. We also investigated data from other cate-
gories (i.e., analgesics, beer, canned soup, cereal, and crack-
ers) for the largest store (in terms of sales volume) in each
category. For categories with a few instances (i.e., canned
soup, cereal, and crackers), SDD yielded a significant
increase in profit over hi–lo (SunBelt Berry Basic cereal
showed an 80% increase). These results indicate that the
SDD pricing tactic may not be category specific. Instead, it
appears that there is some use of this tactic within grocery
store chains, across stores, and across product categories,
and it appears to be effective.
Discussion
Summary of Results
In this research, we set out to determine whether SDD is an
effective tactic because of its impact on consumers’ pur-
chase likelihood based on higher future price expectations
and increased anticipated inaction regret. To assess the
effectiveness of SDD empirically, we conducted four stud-
ies. The results from these studies suggest that SDD is an
effective alternative to both hi–lo pricing and EDLP. Study
1 found SDD to yield higher revenue than both hi–lo pric-
ing and EDLP. It also provided support for the proposed
framework in which SDD leads to higher future price
expectations and anticipated inaction regret, which in turn
lead to greater likelihood to buy. It appears that the
“steadily decreasing” part of the discount is fundamental in
providing consumers with a signal for higher future prices,
which encourages them to buy now. Study 2 showed that
SDD leads to higher willingness to pay than hi–lo pricing,
even when we control the number of promotions. Thus, we
ruled out a frequency-of-promotions explanation (Alba et
al. 1999). Study 2 also provided participants with multiple
past price points and allowed for a simultaneous compari-
son between the hi–lo and the SDD tactics. In addition, we
observed no deleterious effects on store or brand image
across both lab experiments. Next, the field study showed
that SDD yields higher revenue than hi–lo pricing. Finally,
an examination of grocery store scanner panel data revealed
that SDD may be a more profitable tactic than hi–lo pricing.
The overall results confirm our prediction and indicate that
SDD generates greater revenue than hi–lo pricing.
In addition, Study 2 tested participants’ likelihood to
visit a store on the basis of its pricing tactic. The findings
show that the SDD tactic generated a greater likelihood of
visiting the store than hi–lo pricing. Thus, SDD may benefit
the retailer by yielding greater revenues and increased store
traffic. Because store traffic was constant in the field experi-
ment (Study 3) and was assumed to be constant in Study 1,
Ending a Price Promotion / 59
meant to provide a definitive test of the two tactics. Instead,
these results demonstrate an occasion when SDD was evi-
denced in the marketplace, thus providing a rough assess-
ment of its effectiveness.
On this occasion, the price of Diet Pepsi increased from
its low point of $1.09 in two intermediate steps ($1.29,
$1.49). We compared this trend of price increases with
prices from a couple of weeks later, when the price of Diet
Pepsi increased from its low point of $1.09 directly to
$1.59. The average purchase price across the hi–lo promo-
tion period was $1.42, and the average purchase price
across the SDD promotion period was $1.41. Consistent
with prior results, consumption of soft drinks is positively
related to atmospheric temperature (Bello and Al-Hammad
2006; Hays 1999). To make the comparisons as accurate
and fair as possible, we contacted the National Oceanic and
Atmospheric Association to obtain daily temperature data
for the period and location involved in these tests. Because
SDD actually occurred in late April and hi–lo occurred in
the mid-May, the average temperature was higher during
the hi–lo promotion than during the SDD promotion (69°F
versus 53°F, p< .01). Thus, the actual test is a conservative
estimate of the effectiveness of SDD. The revenue gener-
ated from SDD during the four-week period was
$14,846.80, and from hi–lo, it was $12,471.00. This repre-
sents a 19.1% increase in revenue and a 25% increase in
profit from using SDD versus hi–lo. Another example was
for a six-pack of Pepsi 12-ounce cans. The price started at
$2.79 and dropped to $1.99. The SDD series brought the
price back up to $2.79 with one intermediate step ($2.33),
while the hi–lo practice brought the price back to $2.79
directly. The revenue generated from SDD was $303.08,
and from hi–lo, it was $149.76. This represents a 102%
increase in revenue from using SDD versus hi–lo during the
three-week period. In addition, SDD generated 43% more
profit than hi–lo. There were no significant differences in
temperature between the two tactics. Thus, we were able to
find examples of SDD pricing, and it appeared to be effec-
tive. Keeping in mind the limitations we mentioned previ-
ously, this is consistent with the results from the experimen-
tal studies and the field study we presented.
To get a more reliable measure of the effectiveness of
the different tactics, we examined all occurrences of the
same pricing tactic throughout the available data set. We
observed two more occasions of hi–lo with the same previ-
ously described constraints (same regular price, same low
price, return to the same regular price, and no prices above
or below those extremes for a few weeks before the focal
period of the tactic).13 When we compare the SDD promo-
tion period with the average hi–lo promotion period (aver-
age of four promotion periods of the same hi–lo pattern),
the revenue generated from SDD was 29% higher than the
revenue generated from average hi–lo. In addition, SDD
generated 44% higher profit than average hi–lo. Finally, the
Prior research by Alba and colleagues (1999) demon-
strates that for dichotomous price distributions, promotions
with greater depth result in lower price estimates than pro-
motions offered more frequently but at lower depths. More-
over, their results indicate that for nondichotomous price
distributions, greater frequency of promotions results in
lower price estimates than deeper depth. However, in the
research presented here, we do not directly compare differ-
ent depths and frequency of promotions for dichotomous or
nondichotomous distributions. Instead, we compare hi–lo
pricing, which has a dichotomous price distribution, with
SDD, which has a nondichotomous price distribution.
Importantly, because of the nature of the SDD tactic, its
price distribution cannot be dichotomous. Future
researchers are encouraged to examine the condition in
which hi–lo pricing is nondichotomous to compare it with
SDD. From the work of Alba and colleagues, SDD may be
preferred in this instance because it has more frequent pro-
motions, which lead to lower price estimates.
Lalwani and Monroe (2005) replicate and extend Alba
and colleagues’ (1999) results and suggest that it is not only
the dichotomous versus nondichotomous nature of the price
distribution that affects the depth and frequency effects but
also the salience of depth and frequency. Future researchers
are encouraged to examine the relative effectiveness of
hi–lo pricing versus SDD when the depth of the discount is
larger (e.g., reduced from $499 to $299 instead of to $349,
as in our first two studies). Lalwani and Monroe’s results
suggest that the magnitude of promotions should be more
salient for a higher-priced product, such as the PDA used in
Studies 1 and 2, and should result in a depth effect favoring
hi–lo pricing. However, with this larger depth of discount,
the marketer has the ability to offer additional discounts
before returning the product to its regular price. These addi-
tional discounts should enhance the frequency effect and
favor the SDD tactic. Further research needs to disentangle
these competing effects. An evaluation of the depth of the
discounts and the number of steps may uncover additional
boundary conditions associated with the effectiveness of
SDD versus hi–lo pricing.
Moreover, future researchers need to consider the role
of consumer stockpiling behavior on the relative effective-
ness of hi–lo pricing versus SDD. Mela, Jedidi, and Bow-
man (1998) show that consumers wait for deep discounts,
and Ailawadi and colleagues (2007) suggest that consumer
stockpiling does not necessarily hurt sales. We note that in
Study 3, no consumer purchased more than one wine stop-
per. Future researchers are encouraged to explore these
stockpiling-related issues regarding the relative effective-
ness of the hi–lo and SDD pricing tactics.
In addition, this research was restricted to SDD. Addi-
tional research is needed to explore the effectiveness of
using random or uneven decreasing discounts before return-
ing the product to its original price. Our initial attempt to
investigate random discounting patterns in Study 1 seems to
suggest that they are less effective than SDD and not signif-
icantly different from hi–lo. Research also needs to con-
sider the impact of perceptions of deal frequency and per-
ceptions of average deal price when determining the relative
effectiveness of SDD versus hi–lo pricing (Krishna and
60 / Journal of Marketing, January 2010
the results represent a conservative estimate of SDD’s effec-
tiveness over hi–lo pricing. Finally, Study 4 provides some
anecdotal illustration from the field regarding grocery prod-
ucts using a pricing tactic that resembles SDD and shows
that it is profitable.
Managerial Implications
The practices of hi–lo pricing and EDLP are ubiquitous in
today’s retail landscape. Managers often discount a product
for a period and then return the price to its original level all
at once (hi–lo pricing). For example, managers might regu-
larly charge $999 for a television, put it on sale for $799 for
a week, and then raise the price back to $999 after a week.
Alternatively, some retail managers choose to employ an
EDLP tactic and price the television at $919 every week.
Our research supports the use of an SDD tactic, in which
the television described is discounted to $799, and then
instead of returning it to its original price all at once, the
retailer offers at least one additional sale, such as $899.
Higher future price expectations and greater anticipation of
inaction regret appear to be the underlying mechanisms that
lead to the effectiveness of the SDD tactic.
The SDD tactic is especially relevant given the current
economic downturn. Many marketers have reduced prices
in an effort to encourage consumers to buy. How these mar-
keters return the prices to their original level as the eco-
nomic landscape improves can have a great impact on their
bottom line. The research presented here suggests that man-
agers should highly consider bringing the prices of their
products back up to their original levels in steps instead of
all at once to take advantage of higher future price expecta-
tions and greater anticipated inaction regret.
Limitations and Future Research Directions
This research has a some limitations. First, further research
could evaluate the effectiveness of SDD in choice sets
involving multiple brands to examine its effect on brand
switching (Zeelenberg and Van Putten 2005). Second,
although the results from the field study are encouraging,
the study included only one product category and was con-
ducted in one store for a period of 30 weeks. Third, study-
ing the longer-term impact of SDD on store image and
brand image when using less known or store-branded prod-
ucts is also warranted. Fourth, we tested our theoretical
rationale only in Study 1. More evidence regarding the
underlying mechanisms of future price expectations and
anticipated inaction regret is warranted. Fifth, although we
assessed price uncertainty in Study 2, our measure captured
participants’ level of certainty (i.e., confidence) about their
price estimates and not the level of certainty about the
store’s future prices from each of the pricing tactics.
Further research is also warranted to determine whether
SDD can be effective for grocery store products, for prod-
ucts purchased frequently, for products with low price and
profit margin, or for products varying in their level of
necessity. The examination of the Dominick’s scanner panel
data provides some initial evidence that SDD may prove to
be effective for grocery store products.
In conclusion, SDD may be an effective pricing tactic
for sellers to employ. Evidence from the lab and from the
field corroborated this assessment. We presented and empir-
ically validated a conceptual framework that provides the
theoretical underpinnings for SDD’s effectiveness, and we
provided some potential boundary conditions that might
limit its application and success. We encourage future
researchers to investigate the effectiveness of SDD and its
variations in diverse settings.
Appendix
Study 1 Scenario Example: Week 3 of SDD ($499,
$379, $409)
Page 1
Recently you realized that you were having trouble keeping
track of your schedule and decided to purchase a PDA.
You find a store where you like one of the PDAs avail-
able. It has the features that you desire to better keep track
of your schedule. A week ago, the store had advertised a
one-week sale and the PDA that normally sells for $499 was
offered at $379. At that time, you were not looking to buy
one but you remember the sale. On page two of this packet
is the store’s advertisement for the PDA.
Please take a couple of minutes to look at the
PDA in the ad on page two before continuing.
Page 2 Regularly $499
Today $379
HP iPAQ 4155 PDA
• The pacesetting iPAQ showcasing Microsoft® Pocket PC Soft-
ware 2003
• 3.5″TFT transreflective color display for easy viewing, both
indoors and outdoors
• 400MHz Intel® Xscale processor
• iPAQ File Store protects critical data in nonvolatile storage
• Integrated Bluetooth, MP3 Stereo
• 64 MB of RAM for applications, files, music and more
Last Sunday while out shopping you stopped by the
store and saw that the store was selling the product at $409.
On page three of this packet is the store’s advertisement for
the PDA.
Ending a Price Promotion / 61
Johar 1996). In particular, Krishna and Johar (1996) find
that the greater the perception of deal frequency and the
greater the perceived average deal price, the greater is con-
sumers’ willingness to pay. The importance of these two
factors on the relative effectiveness of SDD versus hi–lo
pricing warrants additional research. Research related to the
impacts of the depth of the discount, the duration of the dis-
count, the number of steps in SDD, and the duration of each
step is also warranted. Further research could also assess
the effectiveness of the SDD strategy for categories in
which capacity management (e.g., services) or inventory
management (e.g., perishables) is critical. Because we stud-
ied volume herein, further research might also be conducted
to evaluate the relative impact of hi–lo pricing and SDD on
the speed with which products are sold.
From a theoretical perspective, this research focused on
the roles of future price expectations and anticipated inac-
tion regret in predicting purchase likelihood. Both of these
constructs are forward looking in nature. Future researchers
are encouraged to evaluate the roles of experienced regret
(Inman and Zeelenberg 2002; Tsiros and Mittal 2000) and
past prices serving as historical reference prices (Briesch et
al. 1997) as additional predictors of purchase likelihood. In
particular, it is possible that experienced regret (from miss-
ing a prior larger sale) influences consumer behavior by
reducing purchase likelihood (Tsiros 2009). We expect that
experienced regret will be lower on average under SDD
than under hi–lo, especially as the number of steps in
returning the price to the original level increases as the dif-
ference between the two consecutive promotions decreases.
Future researchers might also consider the role of past
prices serving as external reference prices on the relative
effectiveness of SDD versus hi–lo pricing. For example, for
a particular set of prices, such as those displayed in Table 1,
a reference price could be calculated by exponentially
smoothing a brand’s own shelf prices on previous purchase
occasions (Kalyanaram and Little 1994; Lattin and Bucklin
1989). Briesch and colleagues (1997) find this reference-
pricing model to be predictive. This brand-specific reference-
pricing model could be used to determine the break-even
smoothing constant (representing the degree to which past
prices are incorporated into current reference price esti-
mates) in which SDD and hi–lo pricing are equally effec-
tive. This type of research might uncover boundary condi-
tions associated with when hi–lo is preferred versus when
SDD is more effective, assuming that consumers employ
past prices to form their reference price estimates. Finally,
another potential boundary condition might be consumer
expectations for the product price to increase above the
regular price. As we mentioned previously, an assumption
we made in building and testing our conceptual framework
was that there are well-known and advertised regular prices
that are steady, and consumers do not expect future prices to
exceed those levels. If this assumption does not hold, SDD
may not enjoy any benefit over hi–lo, and indeed the rapid
and large price increment may serve as a strong signal for
potential price spikes, which may lead to higher purchase
behavior. Note that in both Studies 1 and 2, participants did
not forecast the price of the PDA to go above its regular
price ($499).
Chef’s Special
THREE BIG DAYS TO SAVE!
All Glass Wine Stoppers
Regular price $24.95
Buy it on March 13th for $17.45
Buy it on March 14th for $19.95
Buy it on March 15th for $22.45
2. The store carries high quality merchandise.
Strongly Agree Strongly Disagree
7654321
Brand image
1. My image of the HP IPAQ is positive
Strongly Agree Strongly Disagree
7654321
2. This HP IPAQ appears to be of quality
Strongly Agree Strongly Disagree
7654321
Study 2 Additional Measures
Willingness to pay
If you were to buy this HP iPAQ, what is the highest
price you would pay? $_____
Likelihood to Visit Store
How likely are you to visit Store A?
Very Likely Very Unlikely
7654321
How likely are you to visit Store B?
Very Likely Very Unlikely
7654321
Future price expectation
What was the most likely price during the week after
you went away? $_____
Uncertainty of price expectation
How certain are you about your price estimate? ___%
(include a number from 0 to 100)
Study 3 Stimuli
SDD
Hi–Lo (same frequency)
62 / Journal of Marketing, January 2010
Please take a couple of minutes to look at the
PDA in the ad on page three before continuing.
Page 3 Regularly $499
Today $409
HP iPAQ 4155 PDA
• The pacesetting iPAQ showcasing Microsoft® Pocket PC Soft-
ware 2003
• 3.5″TFT transreflective color display for easy viewing, both
indoors and outdoors
• 400MHz Intel® Xscale processor
• iPAQ File Store protects critical data in nonvolatile storage
• Integrated Bluetooth, MP3 Stereo
• 64 MB of RAM for applications, files, music and more
At this time you realize that you have to make a deci-
sion to either purchase the PDA at the new price or wait for
another sale or visit another store.
Study 1 Measures
Purchase likelihood at a given price
I am __________ the PDA for $409?
____ likely to buy
____ neither likely or unlikely to buy
____ unlikely to buy
Thought listing
In the space below, please describe your thoughts that
led to this decision.
_____________________________________________
_____________________________________________
Future price expectations
What is your best estimate of what the price of the PDA
will be 1 week from now? $ _____
Anticipated inaction regret
1. If I don’t buy the PDA now, I will regret it later.
Strongly Agree Strongly Disagree
7654321
Store image
1. My image of the store is positive.
Strongly Agree Strongly Disagree
7654321
Chef’s Special
THREE BIG DAYS TO SAVE!
All Glass Wine Stoppers
Regular price $24.95
Buy it on March 20th for $19.95
Buy it on March 21th for $19.95
Buy it on March 22th for $19.95
Janiszewski, Chris and Donald R. Lichtenstein (1999), “A Range
Theory Account of Price Perception,” Journal of Consumer
Research, 25 (March), 353–68.
Kalwani, Manohar U. and Chi Kin Yim (1992), “Consumer Price
and Promotion Expectations: An Experimental Study,” Journal
of Marketing Research, 29 (February), 90–100.
Kalyanaram, Gurumurthy and John D.C. Little (1994), “An
Empirical Analysis of Latitude of Price Acceptance in Con-
sumer Packaged Goods,” Journal of Consumer Research, 21
(3), 408–418.
Kopalle, Praveen K., Carl F. Mela, and Lawrence Marsh (1999),
“The Dynamic Effect of Discounting on Sales: Empirical
Analysis and Normative Pricing Implications,” Marketing Sci-
ence, 18 (3), 317–32.
Krishna, Aradhna (1991), “Effect of Dealing Patterns on Con-
sumer Perceptions of Deal Frequency and Willingness to Pay,”
Journal of Marketing Research, 28 (November), 441–51.
——— (1994), “The Impact of Dealing Patterns on Purchase
Behavior,” Marketing Science, 13 (4), 351–73.
——— and Gita Venkataramani Johar (1996), “Consumer Percep-
tions of Deals: The Biasing Effects of Varying Deal Prices,”
Journal of Experimental Psychology: Applied, 2 (3), 187–206.
Lalwani, Ashok K. and Kent B. Monroe (2005), “A Reexamination
of Frequency-Depth Effects in Consumer Price Judgments,”
Journal of Consumer Research, 32 (3), 480–85.
Lattin, James M. and Randolph E. Bucklin (1989), “Reference
Effects of Price and Promotion on Brand Choice,” Journal of
Marketing Research, 26 (August), 299–310.
Le Boutiller, John, Susanna Shore Le Boutillier, and Scott A. Nes-
lin (1994), “A Replication and Extension of the Dickson and
Sawyer Price-Awareness Study,” Marketing Letters, 5 (1),
31–42.
Mace, Sandrine and Scott A. Neslin (2004), “The Determinants of
Pre- and Postpromotion Dips in Sales of Frequently Purchased
Goods,” Journal of Marketing Research, 41 (August), 339–50.
Mazumdar, Tridib, S.P. Raj, and Indrajit Sinha (2005), “Reference
Price Research: Review and Propositions,” Journal of Market-
ing, 69 (October), 84–102.
Mela, Carl F., Sunil Gupta, and Donald R. Lehmann (1997), “The
Long-Term Impact of Promotion and Advertising on Consumer
Brand Choice,” Journal of Marketing Research, 34 (May),
248–61.
———, Kamel Jedidi, and Douglas Bowman (1998), “The Long-
Term Impact of Promotions on Consumer Stockpiling Behav-
ior,” Journal of Marketing Research, 35 (May), 250–62.
Meyer, Robert J. and Joao Assuncao (1990), “The Optimality of
Consumer Stockpiling Strategies,” Marketing Science, 9 (1),
18–41.
Monroe, Kent B. (2003), Pricing: Making Profitable Decisions, 3d
ed. New York: McGraw-Hill/Richard D. Irwin.
——— and R. Krishnan (1985), “The Effect of Price on Subjec-
tive Product Evaluations,” in Perceived Quality: How Con-
sumers View Stores and Merchandise, M. Jacob Jacoby and
Jerry Olson, eds. Lexington, MA: Lexington Books, 209–232.
Pauwels, Koen, Dominique M. Hanssens, and S. Siddarth (2002),
“The Long-Term Effects of Price Promotions on Category Inci-
dence, Brand Choice, and Purchase Quantity,” Journal of Mar-
keting Research, 39 (November), 421–39.
Sevdalis, Nick, Nigel Harvey, and Michelle Yip (2006), “Regret
Triggers Inaction Inertia—but Which Regret and How?” The
British Psychological Society, 45 (4), 839–53.
Simonson, Itamar (1992), “The Influence of Anticipating Regret
and Responsibility on Purchase Decisions,” Journal of Con-
sumer Research, 19 (1), 105–118.
Slonim, Robert and Ellen Garbarino (1999), “The Effect of Price
History on Demand as Mediated by Perceived Price Expensive-
ness,” Journal of Business Research, 45 (1), 1–14.
Ending a Price Promotion / 63
REFERENCES
Ailawadi, Kusum L., Karen Gedenk, Christian Lutzky, and Scott
A. Neslin (2007), “Decomposition of the Sales Impact of
Promotion-Induced Stockpiling,” Journal of Marketing
Research, 44 (August), 450–67.
Alba, Joseph W., Carl F. Mela, Terence A. Shimp, and Joel E.
Urbany (1999), “The Effect of Discount Frequency and Depth
on Consumer Price Judgments,” Journal of Consumer
Research, 26 (2), 99–114.
Ariely, Dan (1998), “Combining Experiences over Time: The
Effects of Duration, Intensity Changes and On-line Measure-
ments on Retrospective Pain Evaluations,” Journal of Behav-
ioral Decision Making, 11 (March), 19–45.
Baron, Reuben M. and David A. Kenny (1986), “The Moderator-
Mediator Variable Distinction in Social Psychological
Research: Conceptual, Strategic, and Statistical Considera-
tions,” Journal of Personality and Social Psychology, 51 (6),
1173–82.
Bello, L.L. and N. Al-Hammad (2006), “Pattern of Fluid Con-
sumption in a Sample of Saudi Arabian Adolescents Aged 12-
13 Years,” International Journal of Pediatric Dentistry, 16 (3),
168–73.
Blattberg, Robert C., Richard Briesch, and Edward J. Fox (1995),
“How Promotions Work,” Marketing Science, 14 (3), G122–32.
Briesch, Richard A., Lakshman Krishnamurthi, Tridib Mazumdar,
and S.P. Raj (1997), “A Comparative Analysis of Reference
Price Models,” Journal of Consumer Research, 24 (2),
202–214.
Buyukkurt, B. Kemal (1986), “Integration of Serially Sampled
Price Information: Modeling and Some Findings,” Journal of
Consumer Research, 13 (3), 357–73.
DelVecchio, Devon, H. Shanker Krishnan, and Daniel C. Smith
(2007), “Cents or Percent? The Effects of Promotion Framing
on Price Expectations and Choice,” Journal of Marketing, 71
(July), 158–70.
Desiraju, Romareo and Steven M. Shugan (1999), “Strategic Ser-
vice Pricing and Yield Management,” Journal of Marketing, 63
(January), 44–56.
Dickson, Peter R. and Alan G. Sawyer (1990), “The Price Knowl-
edge and Search of Supermarket Shoppers,” Journal of Market-
ing, 54 (July), 42–53.
Grewal, Dhruv, R. Krishnan, Julie Baker, and Norm Borin (1998),
“The Effect of Store Name, Brand Name, and Price Discounts
on Consumers’ Evaluations and Purchase Intentions,” Journal
of Retailing, 74 (3), 331–52.
———, Howard Marmorstein, and Arun Sharma (1996), “Com-
municating Price Information Through Semantic Cues: The
Moderating Effects of Situation and Discount Size,” Journal of
Consumer Research, 23 (2), 148–55.
Hays, Constance L. (1999), “Variable-Price Coke Machine Being
Tested,” The New York Times, (October 28), C1.
Helson, Harry (1964), Adaptation-Level Theory. New York:
Harper & Row.
Hsee, Christopher K., Robert P. Abelson, and Peter Salovey
(1991), “The Relative Weighting of Position and Velocity in
Satisfaction,” Psychological Science, 2 (June), 263–66.
Inman, J. Jeffrey, James S. Dyer, and Jianmin Jia (1997), “A Gen-
eralized Utility Model of Disappointment and Regret Effects
on Post-Choice Valuation,” Marketing Science, 16 (2), 97–111.
——— and Marcel Zeelenberg (2002), “Regret in Repeat Pur-
chase Versus Switching Decisions: The Attenuating Role of
Decision Justifiability,” Journal of Consumer Research, 29 (1),
116–28.
Jacobson, Robert and Carl Obermiller (1990), “The Formation of
Expected Future Price: A Reference Price for Forward-Looking
Consumers,” Journal of Consumer Research, 16 (4), 420–31.
——— and ——— (2001), “Product Aversion Following a
Missed Opportunity: Price Contrast or Avoidance of Antici-
pated Regret?” Basic and Applied Social Psychology, 23 (3),
149–56.
———, ———, and Erin E. Tuttle (1995), “Inaction Inertia: Fore-
going Future Benefits as a Result of an Initial Failure to Act,”
Journal of Personality and Social Psychology, 68 (5), 793–803.
Vanhuele, Marc and Xavier Drèze (2002), “Measuring the Price
Knowledge Shoppers Bring to the Store,” Journal of Market-
ing, 66 (October), 72–85.
Winer, Russell (1985), “A Price Vector Model of Demand for Con-
sumer Durables: Preliminary Developments,” Marketing Sci-
ence, 4 (Winter), 74–90.
Zeelenberg, Marcel and Rik Pieters (1999), “Comparing Service
Delivery to What Might Have Been: Behavioral Responses to
Regret and Disappointment,” Journal of Service Research, 2
(August), 86–97.
——— and Marijke van Putten (2005), “The Dark Side of Dis-
counts: An Inaction Inertia Perspective on the Post-Promotion
Dip,” Psychology & Marketing, 22 (8), 611–22.
64 / Journal of Marketing, January 2010
Sobel, Michael E. (1982), “Asymptotic Intervals for Indirect
Effects in Structural Equations Models,” in Sociological
Methodology, S. Leinhart, ed. San Francisco: Jossey-Bass,
290–312.
Sun, Baohong, Scott A. Neslin, and Kannan Srinivasan (2003),
“Measuring the Impact of Promotions on Brand Switching
When Consumers Are Forward Looking,” Journal of Market-
ing Research, 40 (November), 389–405.
Taylor, Kimberly A. (1997), “A Regret Theory Approach to
Assessing Consumer Satisfaction,” Marketing Letters, 8 (2),
229–38.
Tsiros, Michael (2009), “Releasing the Regret Lock: Consumer
Response to New Alternative After a Sale,” Journal of Con-
sumer Research, 35 (April), 1039–1059.
——— and Vikas Mittal (2000), “Regret: A Model of Its
Antecedents and Consequences in Consumer Decision Mak-
ing,” Journal of Consumer Research, 26 (4), 401–417.
Tykocinski, Orit E. and Thane S. Pittman (1998), “The Conse-
quences of Doing Nothing: Inaction Inertia as Avoidance of
Anticipated Counterfactual Regret,” Journal of Personality and
Social Psychology, 75 (3), 607–616.
