Content uploaded by Dhruv Grewal
Author content
All content in this area was uploaded by Dhruv Grewal on Jan 31, 2014
Content may be subject to copyright.
1
Consumer Evaluations of Sale Prices: Role of the Subtraction Principle
ABHIJIT BISWAS
SANDEEP BHOWMICK
ABHIJIT GUHA
DHRUV GREWAL
Abhijit Biswas (A.Biswas@wayne.edu) is the Kmart Endowed Chair and Professor of Marketing
at the Department of Marketing and Supply Chain Management of Wayne State University at
5201 Cass Avenue, Detroit, MI 48202.
Sandeep Bhowmick (Sandeep.Bhowmick@indstate.edu) is an Assistant Professor of Marketing
at Scott College of Business, Indiana State University at 200 North Seventh Street, Terre Haute,
IN 47809.
Abhijit Guha (eg2319@wayne.edu) is an Assistant Professor of Marketing at the Department of
Marketing and Supply Chain Management of Wayne State University at 5201 Cass Avenue,
Detroit, MI 48202.
Dhruv Grewal (dgrewal@babson.edu) is the Toyota Chair in E-Commerce and Electronic
Business and Professor of Marketing at the Department of Marketing of Babson College, Babson
Park, MA 02457.
All authors contributed equally to this article. The authors thank participants at the Pricing
Conference (2011), the American Marketing Association (Winter 2012) and Maastricht
University for helpful comments on prior versions of this article.
Please direct all correspondence to Dhruv Grewal.
2
Consumer Evaluations of Sale Prices: Role of the Subtraction Principle
How exactly does the display location of a sale price, relative to the original price, impact
consumers’ evaluations? Across multiple studies, including field studies with real choices and
studies with non-student samples, this paper shows that consumer evaluations are a function of
the display location of the sale price, but such evaluations are moderated by discount depth. First,
presenting the smaller number to the right (vs. left) makes it easier to initiate the subtraction task,
a phenomenon this paper refers to as the "subtraction principle". Second, given that evaluating
sale prices inherently involves a subtraction task, locating sale prices to the right (vs. left) of the
original price makes it easier to initiate a subtraction task to calculate discount depth, increasing
evaluations for moderate discounts, but not for low discounts. The above effects are potentially
reversed in cases of both very low discounts and exaggerated discounts. The findings in this
paper offer novel and non-intuitive insights into how sale price display locations and discount
depth interact to impact numerical cognitions, processing of sale prices and subsequent
evaluations.
Key words: sale price, sale price display location, discount depth, numerical cognition, price
perceptions.
3
There is no consensus about how retailers should present sale prices. A review of
websites of various online and catalog retailers revealed that some display sale prices to the right
of original prices (e.g. Bodyshop, Saks, Neiman Marcus), while others display sale prices to the
left (e.g. Kmart, REI, Zappos). For example, a 30% discount for a product may be presented
either as [original price of $349.99/sale price $239.99] or as [sale price $239.99/original price
$349.99]. In this paper we examine whether, and how, the location of the sale price relative to
the original price results in different consumer evaluations. Since sale prices constitute an
important marketing cue (Biswas and Blair 1991; Grewal, Marmorstein, and Sharma 1996;
Lichtenstein, Burton, and Karson 1991; Thomas and Morwitz 2009), we believe that academics,
retailers, and public policy makers should all be interested in understanding which sale price
display locations systematically result in better consumer evaluations.
We build from literature in numeric processing (e.g., Beishuizen 1997), subtraction
strategies (Selter 2001) and the ease of verifying equations (Yip 2002), to posit that it is easier to
initiate a subtraction task when the larger number is presented on the left hand side and the
smaller number is presented on the right hand side, a phenomenon we term as the “subtraction
principle”. Consequently we propose that the location of the sale price, relative to the original
price, is likely to influence whether or not consumers actually initiate the subtraction task to
calculate discount depth, and that this difference in the propensity to initiate, in turn, impacts
consumer evaluations. We suggest that, consistent with the subtraction principle, when the sale
price (smaller number) is presented to the right of the original price (larger number), i.e. when an
offer is presented as [original price $349.99/sale price $239.99] more consumers are likely to
initiate the subtraction task and calculate the actual discount depth (i.e. 30%) When an offer is
presented as [sale price $239.99/original price $349.99], fewer consumers are likely to initiate
4
the subtraction task. Further, the (many) consumers who do not initiate the subtraction task will
focus mainly on the sale price, and likely estimate a lower discount depth around 10-12%. This
10-12% estimate is based on prior work by Blair and Landon (1981), wherein participants who
were shown various consumer durables with only the sale price displayed, and then asked to
approximate the discount amount, estimated the average discount level in the 10-12% range.
Thus we posit that for those consumers who do not initiate discount calculations, discount
estimates will be directionally biased toward the 10-12% level, in turn leading to lower
evaluations. The above suppositions are not inconsistent with a wide body of prior work which
indicates that a cue can be processed in a variety of ways (e.g. Chaiken 1980).
The impact of sale prices on consumer evaluations, in general, is also influenced by
discount depth i.e., the original price less the sale price (Grewal et al. 1996). Generally, low
discounts (around 10%) are unlikely to increase evaluations, whereas moderate to large discounts
(30% or more) are likely to increase evaluations (Lichtenstein et al. 1991). In this paper, we
examine how the impact of differences in sale price display location (right vs. left of original
price) on consumer evaluations is moderated by discount depth.
To preview our findings, first we find support for the subtraction principle, which
suggests that displaying smaller numbers to the right of large numbers makes it easier for
consumers to initiate subtraction calculations (Study 1). In the context of sale prices and
consumer evaluations, we show that the location of sale price to the right of the original price
makes it easier to initiate the subtraction task to calculate discount depth, and in case of moderate
discounts this results in higher consumer evaluations (see studies 2A, 2B, 3B, 4A and 4B). Thus,
in the case of moderate discounts, locating sale price to the right of the original price “helps”.
However, initiating this subtraction calculation fails to improve consumers’ evaluations if
5
discount depth is low. In such instances, it matters little whether consumers (i) initiate the
subtraction task and then discover that the discount is low, or (ii) do not initiate the subtraction
task and thus assume the discount as low (see Blair and Landon, 1981; also see studies 2A, 2B
and 4A). Studies 2A, 2B, 3A and 3B examine (i) alternate explanations and (ii) the mechanism
underlying our results. Studies 4A and 4B test external validity by examining (i) effects on non-
student samples and (ii) effects of incorporating real choices. Finally in Study 5, we illustrate
that in some specific cases (e.g., very low discounts – discounts much lower than 10%, and
exaggerated discounts – discounts much greater than 30%), displaying sale prices to the right of
the original price can “hurt” by reducing consumer evaluations. In Figure 1, we present our
conceptual model (Panel A) and the full sequence of related studies and proposed results (Panel
B); the latter panel effectively lays out a roadmap for the paper that follows.
Insert Figure 1 About Here
This research contributes to both the pricing and the numerical cognition literatures. First,
we develop and find support for the subtraction principle. Second, we show how ease of
subtraction impacts sale price evaluations. Third, past literature has typically assumed that
consumers normally calculate discount depth (see Inman, McAlister and Hoyer 1990, and
Mayhew and Winer 1992 for exceptions). Here, for the first time, we propose (and demonstrate)
that differences in sale price display location vis-à-vis the original price impact the likelihood of
consumers calculating discount depth, and consequently impact their evaluations. Fourth, this
study identifies discount depth as a key moderator of the impact of the sale price display location
on consumer evaluations, and identifies specific cases where the retailer may benefit by making
it harder for consumers to calculate discount depth. Finally, this research contributes to practice
by offering specific, actionable policy recommendations about when and how retailers should
6
use different sale price display locations.
Should Differences in Sale Price Display Location Matter?
The marketing literature has long postulated that differences in sale price presentation
formats play a key role in consumers’ perceptions of value (Monroe 2003; Thomas and Morwitz
2009). For example, researchers have specifically investigated the various effects of sale price
presentations, including how the sequence of digits in a price influences consumers’ evaluations
through a left-digit anchoring effect (Thomas and Morwitz 2005), right-digit effect (Coulter and
Coulter 2007), or price precision effect (Thomas, Simon, and Kadiyali 2010). Other researchers
have examined how fonts affect price evaluations, whether in larger or smaller sizes (Coulter and
Coulter 2005) or in different colors, such as red (Puccinelli et al. 2013).
What has remained unexplored is how differences in the location of the sale price (right
vs. left of original price) may impact consumer evaluations. This is an important issue, because
consumer evaluations of offers are affected by the comparison between the sale price and the
original price. That is, consumers gauge the depth of the discount by subtracting the sale price
from the original price (Alba et al. 1999; Grewal et al. 1996), and then develop their evaluations
(e.g., Biswas and Blair 1991; Lichtenstein et al. 1991). To the extent that location of the sale
price makes the subtraction task easier (or alternatively, more difficult) to initiate, perceptions of
the offered discount and consequently, evaluations of the offer might vary.
The key question in this research is whether differences in sale price display location, i.e.
sale prices to the right (vs. left) of the original price impact consumers’ propensity to initiate the
subtraction task. It is generally accepted that consumer inferences are governed by spatial rules
that extract meanings from a specific display location of a stimulus (Valenzuela and Raghubir
2011). Such shared beliefs are found to influence consumers’ willingness to pay, quality
7
inferences, effort, and time estimation (Nelson and Simmons 2009). Nevertheless, existing
research (in pricing and other domains) shows mixed evidence for influence of right versus left
processing. For example, Valenzuela and Raghubir (2011) find that products are judged as more
expensive, the further to the right these products were in a horizontal shelf display. Dehaene
(1989) suggests unconscious recollection of the number line sets our expectations for left to right
processing of numbers. However, Shapiro and MacInnis (1992) find no evidence of any
systematic visual processing in left-right display.
In this paper we posit sale price display locations to the right (vs. left) of the original
price facilitate initiation of the subtraction task. We label this as the subtraction principle.
Building on the subtraction principle, we next examine how the joint effect of sale price location
and discount depth impacts consumers’ evaluations. We develop these two ideas in the sections
that follow.
The Subtraction Principle
Cognitive scientists and psychologists have examined how people handle subtraction
tasks (e.g., Campbell and Xue 2001; Hecht 1999; Imbo, Vandierendonck, and Rosseel 2007),
and show that subtraction computations are difficult and very prone to errors (Baroody 1984;
Fuson, Smith, and LoCicero 1997; Selter 2001). However, is it possible that the perceived
difficulty of the subtraction task is influenced by the presentation format of the two numbers?
We posit that computations generally appear easier if the presentation format conforms to
common norms or conventions. We surmise this based on Yip's (2002) suggestion
that inaccurate equations presented in “normal” format (e.g., 3 + 5 = 7) were easier to verify than
in the reversed format (7 = 5 + 3). It should be noted that Yip (2002) looked only (i) addition and
multiplication, not subtraction, and (ii) at verifying equations, not at the impact of location of a
8
smaller number versus a larger number. Nevertheless, we build on insights gleaned from Yip’s
work, which suggests that “normal” presentation formats might be easier to process.
In general, consumers expect to visualize a series of numbers with smaller numbers to the
left of larger numbers, according to how numbers appear on a number line (Dehaene 1989) and
also according with the right-heavy effect (i.e. that items placed on the right side are perceived as
relatively heavier, Deng and Kahn 2009). By this logic, the “normal” position for a smaller
number should be to the left of the larger number. However, we believe that the more relevant
question here is where consumers expect to see a smaller number (sale price) relative to a larger
number (original price) in the specific context of a subtraction task. Elementary school math
books generally follow a consistent norm, with the smaller number to the right of the larger
number, such that they ask students to calculate 100 – 75 = ? rather than –75 + 100 = ? (also,
such format is similar to calculations used as numerical cognition research exemplars, e.g. Selter
2001). The above discussion suggests that in the context of a subtraction task, individuals expect
the “normal” format as one where the smaller number appears to the right of the larger number.
Further, researchers have found that stepwise calculations are more frequently/
successfully used as a strategy for solving subtraction problems (Beishuizen 1997; Selter 2001).
That is, when calculating [701-698], individuals performed the tasks [701-600], [-90], [-8]. Such
stepwise calculations are easier to do when the larger number is to the left, than to the right, as
then the individual does not have to (mentally) flip the numbers in the head prior to starting the
stepwise calculation. This too suggests that in the context of a subtraction task, in the “normal”
format, individuals expect the smaller number to appear to the right of the larger number. This is
- a priori, and in light of some past findings - non-obvious, and it is especially worth noting that
this is exactly the opposite of how numbers are typically displayed on the “number line”, and
9
exactly the opposite of the findings in prior research (e.g. Valenzuela and Raghubir (2011),
although these findings were not in the context of a subtraction task).
To determine if, in the context of a subtraction task, smaller numbers appearing to the
right of the larger number is viewed as “normal”, we showed 52 undergraduates (Pilot Study 1)
in a U.S. Midwestern university a series of subtraction problems, such as 34582 – 28901 = 5681.
Each subtraction problem appeared with either the smaller number to the right (i.e., normal
format) or the smaller number to the left (i.e., -28901 + 34582 = 5681; reverse format). For each
problem, participants indicated which format they were “more likely to see in daily life” (where
1 = “more likely to see smaller number to left”; 5 = “more likely to see smaller number to
right”). The responses revealed participants felt that they were more likely to see the smaller
number to the right (M = 4.38, one-sample t-test, significantly greater than the midpoint; t(51) =
10.05, p < .05). Thus given that, (i) in the specific context of a subtraction task, presenting
smaller numbers to the right of larger numbers is the norm, and (ii) computations are easier when
presentation formats confirm to norms, we predict the “subtraction principle”, i.e. H1 below:
H
1
: Subtraction tasks are easier to initiate when the smaller number is located to the right (vs.
left) of the bigger number.
Study 1: Testing the Subtraction Principle
In Study 1, we examined H1. Specifically we examined whether presenting smaller
numbers to the right of the larger number, rather than to the left, created a greater propensity to
initiate a subtraction task. In effect, we sought support for the subtraction principle.
Two groups of participants (N = 57, undergraduates in a U.S. Midwestern university)
were given two minutes to solve 16 subtraction problems, each of the type (922 – 345), i.e.
involving two three-digit numbers. The first group calculated problems with the smaller number
shown to the right (922 – 345), and this group we label the “normal presentation group.” The
10
second group assessed problems with the smaller number to the left (–345 + 922), and this group
we label as the “reverse presentation group.” To ensure that the participants took the subtraction
task seriously, we told participants that the person (or persons) who answered the most problems
correctly (in each group) would earn $20. We assumed that participants would attempt to solve
more problems when the task seemed easier. As expected, participants in the normal presentation
group attempted more questions (M = 11.24) than those in the reverse presentation group (M =
9.04; t(55) = 2.48, p < .05).
1
This data supported H
1
and the subtraction principle, as we found
that for a subtraction task, presenting the smaller number to the right of the larger number made
it easier for consumers to initiate a subtraction task. We expect these results to hold for those sale
prices where the (smaller) sale price is to the right of the (larger) original price.
Sale Price Location: The Moderating Role of Discount Depth
In the previous sections we developed the subtraction principle and provided empirical
support (independent of any linkage to sale prices) for the principle. We now proceed to examine
how the subtraction principle may influence sale price evaluations, and how such evaluations
might be moderated by discount depth.
Adaptation-level theory suggests that discount depths that result in prices close to
consumers’ adaptation levels evoke a neutral response (Helson 1964), whereas discount depths
that produce prices markedly different from those adaptation levels evoke stronger responses
(Alba et al. 1999). Along similar lines, assimilation and contrast theory suggests that smaller
discounts get absorbed within consumers’ latitude of acceptable prices, with little impact on their
1
Participants in the normal presentation group also answered more questions correctly (M = 9.62) than did those in
the reverse presentation group (M = 6.18; F(1,55) = 12.2, p < .05). Differences in number of questions answered
correctly can be either due to differences in number of questions attempted (9.04 vs. 11.24) or differences in
propensity to answer questions correctly, or both. From the data, it looks like there was no significant difference in
the propensity to answer questions correctly (test for difference in medians, p > .20), so any differences in the
number of questions answered correctly is mainly due to differences in the number of questions attempted.
11
valuations in comparison with the original price, whereas larger discounts create a salient
contrast with the original price and thus triggers increased evaluations (Sherif 1963). Thus, a
moderate discount depth of 30% is far more effective for increasing consumers’ evaluations, than
a low discount depth of 10% (e.g., Grewal et al. 1996; Lichtenstein et al. 1991).
This is not to say that consumers always need to know discount depth to evaluate a sale
price. In the event that no original price is provided, or consumers do not/ cannot calculate
discount depth, consumers typically assume a discount depth biased in the direction of 10-12%
(Blair and Landon, 1981) and evaluate sale prices using this assumption. Consequently, we
examine how differences in discount depth moderate the impact of the sale price display format
on consumers’ evaluations. At this stage, we consider only moderate discounts (e.g. 30%) versus
low discounts (e.g. 10%); later in this paper, we consider cases where discount depth is (i) much
lower than 10%, and (ii) much higher than 30%.
Two components of a sale price advertisement likely influence consumer evaluations.
First, the sale price, on its own, influences evaluations since it represents the effective price the
consumer pays. Second, and more important from the standpoint of this research, discount depth
influences consumer evaluations, since larger discounts suggests increased deal value. As we
have noted before, calculating discount depth inherently involves a subtraction task and that the
subtraction task is a key determinant of evaluations relating to sale prices. Given the criticality of
this subtraction task, any elements in the sale price display that relate to the subtraction principle,
may in turn drive evaluations. We elaborate on the latter point in the next paragraph.
First, consider a case of moderate discount, i.e. a 30% discount. When sale price is to the
right of the original price, consistent with the subtraction principle, consumers perceive the
subtraction task as (relatively) easier to initiate and consequently calculate discount depth.
12
Consumers identify the discount depth at around the “true” value (about 30%) and thus have
high(er) evaluations. But when sale price is to the left of the original price, consumers perceive
the subtraction task as (relatively) harder to initiate. This inhibits the consumers from initiating
the subtraction task. As a result, many focus just on the sale price and approximate discount
depth at around 10-12% (Blair and Landon 1981), and thus have lower evaluations. In the case of
moderate discounts, therefore, locating sale price to the right (vs. left) leads to relatively higher
evaluations. Potentially, assuming the above chain of logic holds, any differences in evaluations
across the right versus left sale price locations should be mediated by (i) differences in perceived
difficulty of initiating the subtraction task, and (ii) differences in estimated discount depth.
Now consider what happens when the discount depth is low, i.e. a 10% discount. We
follow the same chain of logic outlined earlier, except here we adjust for the lower “true”
discount depth. When sale price is displayed to the right of the original price, more consumers
are likely to initiate the subtraction task to calculate discount depth, but then discover the
discount depth to be low (in line with the “true” 10% discount). This will not raise evaluations.
When sale prices appear to the left of the original price, fewer consumers initiate the subtraction
task. Consumers more likely focus only on the sale price to arrive at their evaluations, and
consistent with Blair and Landon (1981), assume a discount depth of about 10-12%, which is
(again) a low discount depth and will not raise evaluations. Thus, irrespective of whether sale
prices are located to the right or left, discount depth estimates will be around 10-12%, i.e. low,
leading to similar, and low, evaluations. Formally:
H
2
: Discount depth moderates the relationships between the display location of the sale price
(right vs. left of the original price) and value perceptions and purchase intentions. When
the discount is moderate, (a) value perceptions and (b) purchase intentions are higher if
the sale price is located to the right (vs. left) of the original price, but (c) when the
discount is low, value perceptions and purchase intentions are not affected by sale price
location.
13
Study 2A: Examining the Moderating Role of Discount Depth
We next examined H2, i.e., whether presenting the sale price to the right (vs. left) of the
original price leads to increased consumer evaluations, and whether such increases are
moderated by discount depth (low vs. moderate).
Method
We used a 2 (discount depth: moderate vs. low) 2 (sale price display location: left vs.
right of original price) between-subjects design, with random assignment across conditions. In
parallel, we ran a fifth cell, details regarding which are provided later. Data was collected online
from undergraduate students at two large, public U.S. universities in return for course credit. The
final data set included 179 participants across all five conditions.
Participants first saw an advertisement for a BluRay DVD player with an original price of
$349.99. To manipulate discount depth, we showed a sale price of either $239.99 (moderate,
30% discount) or $309.99 (low, 10% discount). Depending on the experimental condition, the
sale price was located either to the left or right of the original price. The dependent variables
were value perceptions (VP) and purchase intentions (PI) (Appendix 1). We also asked two
manipulation check questions: “Was the sale price located to the right or left of the original
price?” [Left; Right], and “Comparing the sale price to the original price of the product
mentioned in the ad, I think the discount offered is of… [1 = extremely low value; 5 = extremely
high value].”
Results
The sale price location manipulation worked as intended. 94.5% correctly identified the
location of the sale price (
2
(1) = 132.59, p < .001) and the 5.5% of participants who incorrectly
14
identified the location of the sale price were not included in further analysis.
2
The 2x2 ANOVA
revealed a significant main effect of discount depth (F(1,153) = 84.35; p < .01) for the second
manipulation check question, but (importantly) no interaction effect (F(1,153) = 0.09; p > .76).
Participants regarded the discounts as we expected (M
low
= 2.48; M
moderate
= 3.64; t(155) = 9.25, p
< .01), with the moderate discount being perceived as larger than the low discount.
For value perceptions (see Figure 2), the ANOVA showed a significant interaction of
discount depth with the sale price display location (F(1,153)
= 15.89, p < .05). In the moderate
discount condition, value perceptions were significantly higher when the sale price appeared to
the right rather than to the left of the original price (M
right
= 4.69, M
left
= 3.69; t(77)
= 4.67, p <
.05), in support of H
2a
. In the low discount condition, value perceptions did not differ
significantly across the two sale price display conditions (M
right
= 3.53, M
left
= 3.82; t(76)
= 1.21,
p > .30), in support of H
2c
.
Insert Figure 2 About Here
The pattern of means for purchase intentions (Figure 2) exactly paralleled that for value
perceptions. The two-way interaction between discount depth and sale price display location was
significant (F(1,153) = 6.48, p < .05). In the moderate discount condition, purchase intentions
were significantly higher when the sale price appeared to the right of the original price (M
right
=
4.19, M
left
= 3.01; t(77)
= 3.18, p < .01), supporting H
2b
. In the low discount condition, purchase
intentions did not differ significantly across the sale price display conditions (M
right
= 2.82, M
left
= 2.89; t(76)
= 0.23, p > .80), supporting H
2c
.
Examining an Alternative Explanation. Study 2A results might also arise because
consumers simply focus on the number to the right-hand side of the advertisement to arrive at
2
Excluding participants who fail manipulation checks is consistent with prior academic research (e.g. Carter and
Gilovich, 2010; pg. 153). Full sample results (provided to reviewers) are similar, and are available from the authors.
15
their evaluations. People generally are right-eye dominant (Porac and Coren 1976) and pay more
attention to objects on the right (Deng and Kahn 2009). If a sale price is presented on the right-
hand side, it is possible that consumers only fixate on the sale price, in which case the
evaluations of the moderate and low discounts could differ because the sale price varies (i.e.,
$239.99 vs. $309.99). When the sale price is to the left, consumers focus only on the original
price, which appears to the right and remains the same ($349.99) across both conditions, leading
to similar evaluations in both.
To test this alternative explanation, we included an additional condition (n=22) in our
experiment, with a sale price of $239.99 on the right (as in the low discount condition) and an
original price of $265.99, so that the discount was only 10% (i.e. a low discount). We contrasted
this cell, with the moderate discount cell with the sale price shown on the right. That is, in this
comparison, both conditions showed the same sale price of $239.99 on the right, but one
indicated a moderate (~30%) discount off the original price of $349.99, whereas the other
indicated a low (~10%) discount off a different original price, of $265.99.
We found significantly higher value perceptions in the moderate discount condition (M =
4.69) than in the low discount condition (M = 3.60; t(59) = 3.8, p < .05), as well as higher
purchase intentions in the moderate versus low discount condition (M
moderate
= 4.19, M
low
= 2.68;
t(59) = 3.08, p < .05). These data do not fit the alternate explanation described above; rather,
they align with the earlier results in Study 2A.
Study 2B: Examining the Mediating Role of Discount Depth Estimates
Study 2A results indicated that discount depth moderates the relationship between the
display location of the sale price and consumer evaluations. In Study 2B, we attempt to replicate
these results. Also, we examine another alternative explanation for our main results, i.e., the
16
possibility that the results in Study 2A were due to differences in the leftmost digit of the two
sale prices [$239.99 vs. $309.99]. Finally, we examine two process check measures, which may
illustrate the mechanism underlying our result. Specifically, we examine (i) consumers’ discount
estimates across right versus left presentation of sale price, which were elicited after participants
responded to the dependent variables, and (ii) response latency measures. Study 2B otherwise
uses methods similar to Study 2A.
Method
Study 2B used a 2(10% vs. 30% discount) x 2(sale price display location: left vs. right of
original price) design, with an additional control cell (explained and discussed later). The data set
included 130 participants across all five conditions. The key dependent variables - value
perceptions and purchase intentions - were similar to Study 2A. The major difference was that
the original price was $329.99, and the sale price was either $296.99 or $230.99. Note that in this
case, the left hand digit [2] across the two sale prices were the exact same, and this addresses a
potential alternative explanation for the difference across the 10% and the 30% conditions in
Study 2A. In addition, we elicited participants’ estimates of discount amount. Further, the
software used (Qualtrix) captured time taken from seeing the stimuli, to the time the participant
responded to the first dependent variable measure.
Results
First, we replicated Study 2A results. The interaction was significant for both value
perceptions (F(1,102) = 4.15, p < .05) and purchase intentions (F(1,102) = 5.89, p < .05). In the
moderate discount condition, both value perceptions and purchase intentions were higher when
the sale price appeared to the right rather than to the left of the original price (VP: M
right
= 4.91,
M
left
= 3.83; t(54) = 3.42, p < .01; PI: M
right
= 4.54, M
left
= 3.26; t(54) = 3.43, p < .01), in support
17
of H
2a
and H2
b
. In the low discount condition, value perceptions and purchase intentions did not
differ significantly across the two sale price display conditions (VP: M
right
= 3.56, M
left
= 3.43;
t(48) = 0.39, p > .69; PI: M
right
= 3.08, M
left
= 3.18; t(48) = -0.24, p > .81), consistent with H2
c
.
Next, in the 2x2 design, we examined dollar value of discount estimates across right
versus left presentation of sale price. In the 30% discount condition, participants estimated a
$91.04 (i.e., 27.59%) discount when sale price was to the right, versus $52.90 (i.e., 16.03%)
discount when sale price was to the left of the original price (t(54)= 5.42, p<.01). The estimates
were again consistent with our earlier assertions. When the sale price was to the right, relatively
more participants initiated calculation of discount depth, and calculated the discount depth as
moderate (M=27.59%, close to 30%). When the sale price was to the left, relatively fewer
participants initiated calculation of discount depth, and relatively more participants assumed a
discount depth around 10-12%. The study estimates of 16.03% were consistent with relatively
more participants imputing a discount around 10-12%, and only a few calculating a discount
depth of 30%. This pattern of discount depth estimates was consistent with the build-up to H2.
In contrast, in the 10% discount condition participants reported discounts of $40.04 (i.e.,
12.13%) versus $38.29 (i.e., 11.6%) in the right versus left sale price display, respectively (t(48)
= 0.31, p>.75), indicating no significant difference across conditions. These estimates were
consistent with our earlier assertions. When the sale price was to the right, relatively more
participants initiated calculation of discount depth, but then identified the discount depth as low
(about 10%). When the sale price was to the left, relatively fewer participants initiated
calculation of discount depth, and relatively more participants assumed a discount depth of 10-
12%. These estimates are in the 10-12% range, consistent with either participants identifying the
discount as low, or participants imputing a discount depth around 10-12%.
18
Next, following the procedures indicated in Preacher, Rucker and Hayes (2007), we ran a
mediated moderation analysis for each dependent variable. With value perceptions and purchase
intentions as the dependent variables, the significant interaction effect of discount level and sale
price display location (β = 0.94, t = 2.04, p < .05 for VP and β = 1.39, t = 2.43, p < .05 for PI)
was fully mediated by discount depth estimates (β = 0.33, t = 0.71, p > .40 for VP and β = 0.48, t
= 0.85, p > .40 for PI), while the direct effect of the mediator remained significant (β = 0.02, t =
3.53, p < .01 for VP and β = 0.03, t = 4.39, p < .01 for PI). Also, the bootstrap confidence
intervals (0.18 to 1.31 for VP and 0.39 to 1.69 for PI) for the mediation effect did not include
zero, implying the effect of mediation was non-zero.
Finally, we examined response latency i.e. how long participants took before responding
to the first dependent variable. We found participants spent less time responding when sale price
was displayed to the left (M
left
= 38.2 seconds) versus right (M
right
= 47.7 seconds; t(99) = 2.15,
p<.05), and there was no interaction effect (F(1,97) = 0.04, p > .80). These findings were
consistent with H
1.
When sale price was displayed on the left, participants were less likely to
initiate the subtraction task, resulting in lower processing time on the average. However, when
sale price was displayed on the right, participants were more likely to initiate the subtraction
task, resulting in higher processing time on the average.
We also examined differences in the frequency distribution of response times, across the
left versus right sale price presentation conditions. When sale price was to the left (vs. right),
relatively more participants responded in less than 25 seconds; possibly indicating the use of a
simple (10-12% discount) heuristic. In contrast, when sale price was to the right (vs. left),
relatively more participants responded in over 50 seconds, possibly reflecting active calculation
of discount depth (detailed analysis of the frequency distribution in the Web Appendix).
19
Additional analysis
As noted earlier, we also included a control cell incorporating a 30% discount, but with
only the sale price shown. By definition, no subtraction was possible in this condition. From
prior literature, we know that many consumers would (in the absence of specific information)
impute a discount around 10-12%. We contrasted the cell above, with the 30% discount cell with
sale price displayed to the left of the original price, where fewer consumers would initiate the
subtraction task to calculate the discount depth, and relatively more consumers would impute
discount depth around 10-12%. Across both cells, value perceptions (M
sale price left of original price
=
3.83, M
no original price
= 3.70) and purchase intentions (M
sale price left of original price
= 3.26, M
no original price
= 3.19) were not significantly different (both t(52) < 0.4, both p > .60). This equivalency, across
a cell where no subtraction was possible, and a cell where subtraction was possible but sale price
was displayed to the left, was consistent with our proposition that participants who saw sale price
to the left were less likely to initiate the subtraction task to evaluate the sale price, and were more
likely to make evaluations based on discount heuristics (as per Blair and Landon 1981). Further,
considering the prior-discussed response latency data, it is unlikely that response latency was
lower when sale price was to the left, because it was easier to initiate the subtraction task. If this
were so, then both value perceptions and purchase intentions should have been higher in the 30%
discount condition with sale price to left versus 30% discount condition with no original price
shown. However, the equivalency across these two cells was more consistent with fewer
participants initiating the subtraction task to calculate the discount depth when sale price was to
the left of the original price.
20
Studies 3A-B: Examining the Role of Perceived Difficulty of Discount Calculations
In Studies 3A and 3B, we examined the role of perceived difficulty of discount
calculations on value perceptions and purchase intentions. In Study 3A we examined a case
wherein all calculations were very simple. In such an instance, most should be able to quickly
calculate discount depth, and so differences in the display location of the sale price should not
matter. In Study 3B, we contrasted a case where we did not explicitly provide discount depth,
with a case where we explicitly provided discount depth. When discount depth is not explicitly
provided, and considering only the 30% discount condition, we expect to replicate the results in
Study 2A and Study 2B; i.e., that evaluations are higher when sale price to the right (vs. left).
When discount depth is explicitly provided, there is no need to perform calculations, and so
evaluations should be similar irrespective whether sale price is to right or left. Full details are
provided for Study 3A; however, for brevity-related reasons here we only provide key results of
Study 3B (full details are provided in the Web Appendix.)
Study 3A
Method. We collected data online from mTurk participants (N = 164).The study was
exactly similar to Study 2B, with two exceptions. First, we used an original price of $299.99, and
sale prices of $269.99 and $209.99. Thus, in this study, “discount depth” calculations were very
simple, with [(300 - 270)/300 = 10%] and [(300 - 210)/300 = 30%]. Second, we collected
information only about value perceptions and purchase intentions (also, the Qualtrix software
collected information about response latency).
Results. We found only a main effect of discount depth (VP: F(1,160) = 34.79, p< .01;
PI: F(1,160) = 28.69, p< .01). Unlike in Study 2B, there were no interaction effects (both
variables, F(1,160) < .23, p> .60). In the 30% discount condition, there were no significant
21
differences in evaluations across sale price right (vs. left) conditions (VP: 5.33 vs. 5.27; PI: 4.79
vs. 4.81; in all cases, t(77) < 1, p > .7). Similarly, in the 10% discount condition, there were
again no significant differences in evaluations across sale price right (vs. left) conditions (VP:
4.37 vs. 4.24; PI: 3.54 vs. 3.33; in all cases, t(83) < 1, p > .5). Also, unlike Study 2B, we found
no effect of left versus right presentation of sale prices in terms of response latency (M
left-right
=32.03-31.20; t(162) = 0.16, p> .80), implying that participants are equally likely to initiate the
subtraction task in both presentation conditions. Thus, when discounts are (very) easy to
calculate, all consumers calculate discounts irrespective of differences in sale price display
location. This indicates that discount calculations, and the associated difficulty in computing
such discounts, were at the heart of Study 2A and Study 2B results.
Study 3B
In Study 3B, participants were exposed to just the moderate discount (30% discount)
condition, with sale price either displayed to the right or the left of the original price. Also, in
each sale price display condition, half of the participants were provided with the specific
discount percentage (30%), whereas the rest were not. The primary measures of interest related
to value perceptions, purchase intentions and “perceived difficulty” of the computation task.
As expected, ANOVAs indicated significant interactions between the percentage discount
cues and sale price display location for both value perceptions (F(1,157)
= 6.76, p < .01) and
purchase intentions (F(1,157)
= 4.12, p < .05). Further, without the percentage discount, just as in
Study 2A, evaluations were significantly higher when sale price was to the right (vs. left) of the
original price. But when the percentage discount was explicitly provided, value perceptions and
purchase intentions were similar, regardless of the sale price display location. Also as expected,
the “perceived difficulty” variable fully mediated the above moderation. Full details are provided
22
in the Web Appendix.
Discussion. Study3A and Study 3B provide convergent support for the fact that
performing calculations, and the associated difficulty levels with doing so, underlies our results.
In Study 3A, where calculations were extremely simple, there is no difference in evaluations
across sale price being located right (vs. left). In Study 3B, in the case where the discount depth
was explicitly provided and so there was no need to do calculations, there was (again) no
difference in evaluations across sale price being located right (vs., left). But when ( in Study 3B)
discount depth was not provided, we replicated earlier results (that evaluations were higher when
sale price was located to the right vs. left), and differences in evaluations were mediated by
“perceived difficulty” of calculations, implying that differences in sale price display location
(right vs., left), consistent with Study 1 and the subtraction principle, led to differences in the
perceived difficulty of engaging in the subtraction task associated with calculating discount
depth.
Studies 4 A-B: Generalizing the Results
We next examine external validity of our primary findings, by examining if the results of
Studies 2A and 2B extend to (i) non-student samples (Studies 4A and 4B), and (ii) real choices
(Study 4B).
Study 4A
Method .We re-ran Study 2B (similar 2x2 research design, but here we used a 32-inch
HDTV) on a non-student sample (N=172). This was a paper-and-pencil study. Participants first
saw an advertisement with an original price of $329.99, and a sale price of either $230.99
(moderate, 30% discount) or $296.99 (low, 10% discount). They responded to questions
pertaining to: (i) value perceptions and purchase intentions, (ii) location of sale price, as a
23
manipulation check, and (iii) demographics. We excluded 3.5% of the participants who
incorrectly identified the location of the sale price. As regards participant profile, (i) 46.7% were
males (ii) 40.3% were over 35 years, and (iii) 38.2% reported earnings of over $50,000 per
annum.
Results. We replicated the results of Study 2B. Briefly, in the moderate discount
condition, value perceptions and purchase intentions were significantly higher when the sale
price appeared to the right (vs. left) of the original price (VP: M
right
= 4.93, M
left
= 3.79; t(81)
=
4.60, p < .01; PI: M
right
= 4.81, M
left
= 3.23; t(81)
= 4.97, p < .01; see Figure 3), in support of H
2a
and H
2b
. In the low discount condition, value perceptions and purchase intentions did not differ
significantly across sale price display locations (VP: M
right
= 3.72, M
left
= 3.96; t(81)
= 0.91, p >
.30; PI: M
right
= 3.15, M
left
= 2.92; t(81)
= 0.83, p > .40), in support of H
2c
.
Insert Figure 3 About Here
Study 4B
The purpose of this study was two-fold. First, we sought to replicate the key results from
Studies 2A and 2B, using another non-student sample. Second, we also incorporated a binary
dependent variable, reflecting choice between two items (the previous studies reported
evaluations for a single item, i.e., value perceptions and purchase intentions), where participants
would actually receive the chosen item. As a consequence, the choice was real and specifically
reflected whether the relative choice share of the focal product would increase if the sale price
was shown on right as opposed to the left.
Method. Participants (N=73) were adults who signed up for a cooking class in a gourmet
food store in an upscale suburb of a major US city. Of the participants who explicitly provided
gender details (n=62), 75.8% were female. All participants provided age range details, as
24
follows: < 30 years = 6%, 30-39 years = 17%, 40-49 years = 22%, > 50 years = 55%.
Prior to the start of the class, the owner-instructor introduced the researchers and
requested the attendees to participate in the study. Participants were then handed a booklet
wherein, on the cover page, they were asked to provide purchase intention judgments about
cooking-related household items. On the first page, a single food processor was displayed
(original price = $184.99, sale price = $129.99, about 30% discount) in catalog-style format.
Participants were shown sale price located either to the left or right (two cell, between-subjects
design), and asked to indicate their purchase intentions (scale similar to Study 2A).
On the second page, two good-quality wooden spoons, imported from Western Europe,
were displayed, again in catalog-style format. Spoon “A” had a sale price of $8.49 (no original
price shown). Spoon “B” had an original price of $12.29 and a sale price of $8.59, i.e. about a
30% discount. In the case of spoon “B’, participants were (similarly) shown sale price either to
the left or right. Participants were asked to pick which spoon they preferred, and were explicitly
told that this spoon would be given to the participant. Finally, on the last page, we asked for
demographic data.
Results. For the food processor, purchase intentions were higher when sale price was
shown to the right versus left (M
right
= 4.50, M
left
= 3.62, t(71) = 2.18, p < .05). As regards the
wooden spoons, relative choice of focal spoon B (vs. spoon A) was higher when sale price was
shown to the right versus left (M
right
= 38.9%, M
left
= 18.9%, χ
2
(1) = 3.55, p = 0.06). These results
were consistent with those of Studies 2A and 2B, and thus generalize to non-student samples,
and also generalize to real choices.
25
Moderating Role of Very Low and Exaggerated Discounts
Motivation
Our findings consistently reveal that displaying the sale price to the right of the original
price facilitates initiation of the subtraction task. Also, since initiating a subtraction task leads to
calculation of discount depth, such initiations “help”, i.e. increase evaluations particularly for
moderate discounts (Studies 2A, 2B, 3B, 4A and 4B). Furthermore, placing sale price to the right
does not “hurt” when discounts are low (Studies 2A, 2B and 4A). These results make it tempting
to conclude that retailers should always display sale prices to the right of the original price.
In our next study, we examine whether displaying sale price to the right of the original
price might “hurt”, i.e. decrease consumer evaluations. In the course of such examination, we
consider two discount depth exemplars that we have not examined prior in the context of sale
price display location. Specifically, we examine the effects of (i) very low discount (much lower
than 10%), and (ii) exaggerated discount (much higher than 30%). We propose that the effects of
displaying sale price to the right (vs. left) of the original price might “hurt” when the discounts
are either very low or exaggerated. This is likely because upon calculating the discount depth,
revelation of either a very low, or an exaggerated discount may lead to negative inferences by
consumers, in turn decreasing consumers’ evaluations. We also propose that the mechanisms
underlying these reversals should vary for the two types of discounts, as arguments primed by
exaggerated discounts should differ from those primed by very low discounts. Our argument is in
line with the Wegener et al. (2010), in that different anchors result in the “selective accessibility”
of information, so that “anchor-consistent knowledge” provides possible explanations for a
judgment target. In our context for example, varying discount levels might result in different
anchors and hence invoke separate arguments. These expectations are also supported by
26
Schwarz’s (2006) “situated cognition” perspective where human cognition is viewed to serve
instrumental roles in facilitating people’s interactions with the world, and hence such cognition is
suggested as being highly adaptive to specific contexts.
Very Low Discounts
Very low discounts may trigger negative beliefs about the retailer, such as the retailer is
trying to be opportunistic, or aiming to manipulate consumers by advertising a sale that is not
“truly” a sale (Simonson, Carmon, and O’Curry 1994). Consumers generally resent such
transparent attempts at motivating purchase and manipulating choice, view such tactics as
manipulative, and even “punish” offending retailers by restricting their purchases (i.e.,
“consumers use this as a reason not to choose the promoted brand,” Simonson et al., 1994, p. 31;
see also Brehm 1966). If the consumer believes that the retailer is merely trying to motivate
purchase, without offering anything substantive in return, those beliefs arouse resentment and
negatively affect consumers’ evaluations.
Consider a sale price presented to the right of the original price when the discount is very
low. The sale price’s display location on the right makes it easier to undertake the subtraction
task (Study 1). So more consumers are likely to initiate the subtraction task and identify the
discount depth as “very low”, which (i) does not increase evaluations, but (ii) may signal that the
retailer is merely being opportunistic - i.e., not offering a “true sale”. This negative attribution
would (strongly) decrease evaluations and consequently, consumers would be less likely to
purchase the product. However, when the sale price is presented to the left, fewer consumers are
likely to initiate the subtraction task, resulting in fewer consumers identifying the discount depth
as “very low”. In parallel, the relatively more consumers who do not initiate the subtraction task
will base their value judgments primarily on the sale price, and infer a discount depth around 10-
27
12%, which should not prompt perceptions of retailer opportunism. On balance, for very low
discounts, we posit that a sale price presented to the left of the original price is less likely to
adversely impact consumers’ evaluations, compared to a sale price presented to the right. We
thus hypothesize:
H
3
Discount depth moderates the relationships between the display location of sale price (to
the right versus left of the original price) and value perceptions and purchase intentions.
When the discount is moderate, both (a) value perceptions and (b) purchase intentions are
higher if sale price is located on the right, but when the discount is very low, (c) value
perceptions and (d) shopping intentions are lower if the sale price is located on the right
instead of the left.
Exaggerated Discounts
In the case of exaggerated discounts, consumers may see a high offer value, even if they
“discount the discount” (Gupta and Cooper 1992; Urbany et al., 1988). However, in certain
instances, consumers also may perceive that the product is of lower quality – essentially because
of the “price-quality” heuristic (Dodds, Monroe and Grewal 1991; Grewal and Compeau 2006;
Lichtenstein and Burton 1989). Consequently, when consumers see an exaggerated discount,
which is unlike what they see typically, or which is for an unknown product, they may perceive
the exaggerated discount (i.e., very low selling price) to reflect poor product quality (Dodson,
Tybout, and Sternthal 1978). Consequently, despite the exaggerated discount, consumers may be
less likely to buy the product (in some cases) due to negative quality perceptions.
Consider a sale price presented to the right of the original price when the discount is
exaggerated. The sale price’s location on the right makes it easier to undertake the subtraction
task (Study 1). So more consumers likely initiate the subtraction task and identify the discount
depth as “exaggerated”, which (i) increases evaluations, but (ii) may signal problems with
product quality. In some specific instances, negative attributions about quality may overwhelm
any benefit of exaggerated discounts, thus decreasing overall evaluations and lowering likelihood
28
of purchase. However, when the sale price is presented to the left, fewer consumers are likely to
initiate the subtraction task, resulting in fewer consumers identifying the discount depth as
“exaggerated”. In parallel, the relatively more consumers who do not initiate the subtraction task
will base their evaluations primarily on the sale price, and their inferred savings will be biased in
the direction of the default 10-12% discount, mitigating any negative quality attributions. On
balance, for exaggerated discounts, we posit that a sale price presented to the right of the original
price is more likely to adversely impact consumers’ evaluations, compared to a sale price
presented to the left. We thus hypothesize:
H
4
Discount depth moderates the relationships between display location of sale price (to the
right versus left of the original price) and value perceptions and purchase intentions.
When the discount is moderate, (a) value perceptions and (b) purchase intentions are
higher if the sale price is located on the right, but when the discount is exaggerated, (c)
value perceptions and (d) purchase intentions are lower if the sale price is located on the
right rather than the left.
Study 5: Examining Effect of Very Low and Exaggerated Discounts
Method
We used a 3 (discount depth: very low, moderate, or exaggerated) 2 (sale price display
location: left vs. right of original price) between-subjects design, with random assignment across
conditions. Data was collected online from undergraduate students (N = 228) from two public
U.S. universities in return for course credit.
Participants were asked to imagine that they were responsible, on behalf of their family,
for buying a Turkish carpet as a wedding present for a cousin. The Turkish carpet purchase was
from a relatively unknown online retailer, about whom little prior information was available. We
chose the above-described context because we anticipated that participants generally would not
have prior exposure to this domain, nor would they likely be able to distinguish good retailers
29
from bad. In this context, consumers thus would need to rely more on price, and less on their
prior knowledge, to determine retailer intentions, product quality, and purchase decisions.
Participants viewed an advertisement for a Turkish carpet, described as follows: “This is a
colorful Turkish carpet incorporating stylized motifs and a geometric design. It is a detailed yet
understated carpet that will add a touch of sophistication to any room.”
Pretests. We ran two pretests to select the very low and exaggerated discounts. Two
groups of participants (discount levels: very low - N=31, and exaggerated/ very high - N=41)
were instructed to assume that they were shopping online for Turkish carpets. They indicated the
discount level at which they would start getting concerned about why the discount was very low
or very high, respectively. Next, they chose the motivation (from a list) for the retailer to offer a
very low (or very high) discount.
For the very low discount group, 90% of the participants indicated that they would be
concerned if the discount was 4%. Furthermore, a significant majority of participants (M =
80.6%, binomial z = 3.24, p < .05) indicated that offering a sale price of such a low level was
likely motivated by the retailer being opportunistic and trying to trick consumer into buying the
carpet. Such response patterns were consistent with our arguments leading into H
3
. In a
confirmatory pretest (N=35), respondents indicated their view about the 4% discount by selecting
one of the following options: (i) trick to attract consumers or (ii) generally, small discounts are
irritating. Similar to the first pretest result, 71.4% participants (significant choice share, binomial
z = 2.36, p < .05) viewed the 4% discount as a trick to attract consumers.
For the exaggerated discount group, 90% participants indicated that they would be
concerned if the discount was 85% and a significant majority of the participants (M = 78.0%,
binomial z = 3.44, p < .05) indicated that their main concern was that the retailer was trying to
30
hide poor quality, consistent with our arguments for H
4
. In a confirmatory pretest (N=29),
respondents were asked to indicate their view of this 85% discount by selecting one of the
following options: (i) low quality carpet, (ii) carpet not selling well, (iii) outdated model. 58.6%
participants (significant choice share, binomial z = 2.6, p < .05) indicated that they had quality
concerns, similar to the earlier pretest results.
Stimulus. The original price of the Turkish carpet was $976. To manipulate discount
depth, we displayed sale prices of $939 (very low discount, 4%), $679 (moderate discount, 30%,
approximately the same as in Studies 2-4), or $139 (exaggerated discount, 85%). Depending on
the experimental condition, the sale price was displayed to either the left or the right of the
original price. Immediately after exposure to the treatment, the participants responded to scales
measuring value perceptions and purchase intentions. Only after obtaining measures of the main
dependent variables, we assessed participants’ perceptions of retailer opportunism using a three-
item, seven-point scale (Dutta, Biswas and Grewal, 2011), and product quality perceptions using
a two-item, seven-point scale (see Appendix 1). We also included two distractor items in this
section, the responses to which were not expected vary across conditions
3
. Finally, we asked the
two manipulation check questions to assess what participants thought about the discount depth,
and whether they recalled if the sale price was located to the right or left of the original price.
Implicitly, the Qualtrix software measured response latency, i.e., the time before participants
responded to the first dependent variable.
Study 5 differs from previous study formats in two important respects. First, we
examined the effects of two extreme discount conditions, in addition to the moderate discount
3
The items were: “The retailer is probably offering a discount for a special occasion,” and “The retailer is offering a
discount as part of an event promoting ethnic goods”. The ANOVA results showed no significant interaction effects,
and the individual contrasts showed no significant differences across discount levels (all cases, p
s
> .05), implying
that these were not correlated with the dependent variables.
31
condition. Unlike Study 2A, we did not examine the low discount condition. Second, we
included two process-check variables (retailer opportunism and quality perceptions), which we
hoped should vary in such a way that they explain the dependent variables. Note that these
process check variables were measured after we elicited the dependent variables, and so there
should be no demand effect, or priming effects, on responses to the dependent variable measures.
Finally, we did not include the 1.7% of participants who incorrectly indicated whether the sale
price was on the right or left of the original price.
Results
A 2 3 ANOVA indicated the absence of either an interaction effect (F(2,217)
= 0.14, p
> .81) or main effect (F(1,217)
= 0.30, p > .50) of sale price display location on the manipulation
check variable related to discount depth. However, the effect of discount depth was significant
(F(2,217)
= 113.45, p < .01). Participants perceived the various discounts as expected (M
very low
=
2.05; M
moderate
= 3.16; M
exaggerated
= 4.02; moderate vs. very low: t(149) = 8.85, p < .01;
exaggerated vs. very low: t(148) = 14.43, p < .01; exaggerated vs. moderate: t(143) = 6.48, p <
.01).
The response latency results were also as expected, showing a significantly lower
response latency value when sale price was displayed to the left versus right (M
left
= 31.06
seconds, M
right
= 40.45 seconds, t (222) = -3.87, p < .01) and no interaction effect (F(2,218) =
0.87, p> .40). Given these response latency measures, we reasoned (as in Study 2B) that fewer
participants likely initiated the subtraction task when the sale price was displayed to the left. In
addition, like in Study 2B, we also examined differences in the frequency distribution of
response times. The pattern of results was exactly similar to the pattern evidenced in Study 2B.
Detailed results are provided in the Web Appendix.
32
In line with H
3
and H
4
, the ANOVA for value perceptions showed that the interaction
between discount depth and sale price display location (right vs. left) was significant (F(2,218)
=
18.73, p < .01). According to the individual contrasts in Figure 4, in the moderate discount
condition, value perceptions were significantly higher when sale price was located to the right of
the original price (M
right
= 4.66, M
left
= 3.85; t(72)
= 3.59, p < .01), which wholly replicated the
results for the moderate discount condition in Study 2A. In contrast, in the very low discount
condition, value perceptions were significantly lower with a sale price located to the right of the
original price (M
right
= 2.49, M
left
= 3.59; t(76)
= 4.34, p < .01), in support of H
3
. Finally, in the
exaggerated discount condition, value perceptions (again) were significantly lower when the sale
price was located to the right of the original price (M
right
= 3.90, M
left
= 5.29; t(70)
= 4.12, p <
.01), which supported H
4
.
Insert Figure 4 About Here
The results for purchase intentions mirrored the results for value perceptions. An
ANOVA showed a significant interaction between discount depth and sale price display location
(F(2,218)
= 15.35, p < .01). As in Figure 4, purchase intentions were significantly higher in the
moderate discount condition when the sale price was located to the right (M
right
= 4.31, M
left
=
3.21; t(72)
= 3.18, p < .01), but significantly lower in the very low discount condition (M
right
=
2.21, M
left
= 2.99; t(76)
= 2.95, p < .01; supporting H
3
), as well as significantly lower in the
exaggerated discount condition (M
right
= 3.47, M
left
= 4.94; t(70)
= 3.71, p < .01; supporting H
4
).
Thus, in addition to reconfirming H
2a
and H
2b
, our findings offer support for H
3
and H
4
.
In contrast with the moderate discount condition, displaying the sale price to the right of
the original price “hurt”, by leading to lower value perceptions and purchase intentions, in both
the very low and exaggerated discount conditions, effectively reversing the (moderate discount)
33
results in Studies 2A, 2B, 3B and 4A. Although H
3
and H
4
suggest similar patterns of means, the
underlying mechanisms were hypothesized as being different; so we investigated, via mediation
analyses, if our data illustrated the presence of two distinct mechanisms.
Examining Process in the Very Low Discount Condition. In setting up H
3
, we argued that
differences in evaluations across very low discount conditions primarily reflect differences in
consumers’ perceptions of retailer opportunism. This argument was supported. Consumers’
perceptions of retailer opportunism were significantly higher when the sale price was displayed
on the right (M
right
= 5. 81, M
left
= 3.61; t(76)
= 7.39, p < .01). No such differences in evaluations
emerged in the moderate discount condition (M
right
= 4.37, M
left
=4.32; t(72)
=0.12, p> .90).
Furthermore, quality perceptions were similar irrespective of the location of the sale price, and in
both the very low (M
right
= 4.04, M
left
= 4.37; t(75)
= 0.97, p > .30) and moderate (M
right
= 3.71,
M
left
= 3.59; t(71)
= 0.44, p > .60) discount conditions, suggesting that differences across the very
low and the moderate discount conditions were not driven by quality-related differences.
Next, following Preacher and Hayes (2004), we examined if, in the very low discount
condition, retailer opportunism mediated the effects of sale price location. With value
perceptions as the dependent variable, the significant effect of differences in sale price display
location (β = -1.1, t = -4.34, p < .01) was completely eliminated (β = -0.52, t = -1.63, p > .10)
when “retailer opportunism” was included as the mediator, while the direct effect of “retailer
opportunism” remained significant (β = -0.26, t = -2.79, p < .01). We found similar results for
purchase intentions. The significant effect of differences in sale price display location (β = -0.78,
t = -2.95, p < .01) was completely eliminated (β = -0.20, t = -0.61, p > .50) by retailer
opportunism, while the direct effect of the mediator remained significant (β = -0.26, t = -2.68, p
< .01). The bootstrap confidence intervals for the mediation, for both value perceptions (-1.15 to
34
-0.06) and purchase intentions (-1.19 to -0.13) did not include zero, implying significant
mediation. These results were wholly consistent with the theory we espoused to develop H
3
.
4
It is worth examining why we found full mediation. Consider the cell with the sale price
located to the right. Here many are likely to initiate the subtraction task and identify the discount
as 4%. In turn, the 4% discount depth prompts perceptions of opportunism. However, when sale
price is located to the left, few are likely to initiate the subtraction task; and so many will just
proceed to assume a discount of about 10-12%. Overall, (i) neither the 4% nor the 10% discount
increases consumer evaluations, but (ii) the 4% discount prompts perceptions of retailer
opportunism. So the only difference between the two cells is related to perceptions of retailer
opportunism, and hence this variable fully mediates differences in evaluations.
Examining Process in the Exaggerated Discount Condition. With H
4
, we theorized
(different from H
3
) that differences across sale price display location in the exaggerated discount
condition would be driven primarily by differences in perceptions of quality, and not perceptions
of retailer opportunism. We argued that sale price displayed to the right of the original price
should prompt more participants to initiate the subtraction task and identify the discount as
exaggerated, leading them to suspect the quality offered. We found support for this mechanism
as well. Quality perceptions were significantly lower when the sale price appeared on the right of
the original price (M
right
= 2.39, M
left
= 4.15; t(70)
= -4.94, p < .01), but those differences
disappeared in the moderate discount condition (M
right
= 3.71, M
left
= 3.59; t(71)
= 0.44, p > .60).
4
Study 5A: We also directly tested the mechanisms underlying the results in study 5, by conducting a study similar
to Thomas and Morwitz (2009). We used 2 (credibility cue: absent or present) 2 (sale price display location: left or
right of original price) between-subjects design (N=112). The credibility cue was presented to nullify perceptions of
retailer opportunism. In all cases, the discount depth was very low, i.e. 4%. In the credibility cue absent condition,
value perceptions and purchase intentions were significantly lower when the sale price appeared to the right (vs. left)
of the original price, consistent with the results in Study 5. In contrast, in the credibility cue present condition, given
that perceptions of retailer opportunism were suppressed, evaluations did not differ across the sale price display
location conditions, and were consistent with the low discount condition results in Study 2A. These results are
consistent with the mechanism we proposed as underlying H
3
. Full details are available on request.
35
Furthermore, perceptions of retailer opportunism were similar, irrespective of the location of the
sale price (exaggerated: M
right
= 4.11, M
left
= 4.08; t(70)
= 0.08, p > .90; moderate: M
right
= 4.37,
M
left
= 4.32; t(72)
= 0.12, p > .90), suggesting that differences across the exaggerated and the
moderate discount conditions were not driven by opportunism-related thoughts. This pattern of
results is wholly consistent with the theory we espoused to develop H
4
, as well as indicative of a
very different mechanism vis-à-vis the very low discount condition (i.e., H
3
).
Following Preacher and Hayes (2004), we examined if, in the exaggerated discount
condition, perceived quality differences mediated the effects of sale price display location. We
found that the significant main effects of differences in sale price location on value perceptions
(β = -1.39, t = -4.11, p< .01) and purchase intentions (β = -1.46, t = -3.71, p< .01) were (partially)
eliminated with quality perception included as the mediator in the model (β = -0.84, t = -2.25, p>
.01 but p< .05 for value perceptions, and β = -0.79, t = -1.82, p> .05 for purchase intentions),
while the main effect of the mediator remained significant (β = -0.31, t = -2.89, p< .01 for value
perceptions, and β = -0.38, t = -3.05, p< .01 for purchase intentions). The bootstrap confidence
interval for mediation, for both value perceptions (-1.08 to -0.20) and purchase intentions (-1.33
to -0.22) did not include zero, implying significant (partial) mediation.
5
This analysis was also
consistent with the theory we espoused to develop H
4
.
6
This partial mediation result is consistent with the mechanism we explained earlier.
5
As indicated in Appendix 1, the measures relating to perceived quality reference the price discount, and above we
show how these quality measures mediates differences in evaluations. In response to a query raised by one of the
reviewers, we ran a separate study (not reported in the paper) to show that this mediation result sustains even if we
use (other) quality measures that do not reference the price discount.
6
Study 5B: We again directly tested the mechanisms underlying the results in study 5, by conducting a study similar
to Thomas and Morwitz (2009). We used a 2 (quality cue: absent or present) 2 (sale price display location: left or
right of original price) between-subjects designs (N=86), to directly test the mechanism underlying the results of
study 5. The quality cue was presented to nullify perceptions of inferior product quality. In the quality cue absent
condition, given that the cue suppressed perceptions of quality suspicions, consistent with both H
4
and the results
from Study 5, value perceptions and purchase intentions were significantly lower when the sale price appeared to the
right of the original price. In contrast, in the quality cue present condition, evaluations were significantly higher with
a sale price to the right. These findings were consistent with the moderate discount condition results in Study 2A,
and were also consistent with our proposed mechanism underlying H
4
. Full details are available on request.
36
When the sale price is located to the right, many are likely to initiate the subtraction task and
identify the discount as 85%. This large discount can (i) increase evaluations, but (ii) also prompt
perceptions of reduced quality, a negative attribution which decreases evaluations. So, in case of
the large discount, the reduced evaluations reflect a tradeoff between (i) increased evaluations
due to the discount, and (ii) reduced evaluations due to perceptions of reduced quality. Hence
perceptions of reduced quality only partially mediate any differences in evaluations.
Discussion
Retailers use a variety of display locations to present sale prices. Specifically, a survey of
online and catalog retailers found retailers displaying sale prices to the right of, as well as to the
left of, the original price. An open question therefore is whether some sale price display locations
are more effective than others in influencing consumer evaluations. We examined right versus
left sale price display locations, and we found convergent and reinforcing evidence that (i) sale
prices displayed to the right (vs. left) of the original price - consistent with the subtraction
principle - facilitate initiation of the subtraction task necessary to calculate discount depth, and
(ii) whether or not initiation of this subtraction task increases consumer evaluations, is contingent
on discount depth.
This paper makes five key contributions. First, we provide evidence that supports the
subtraction principle, i.e. that subtraction is harder to initiate when the smaller number is placed
before the larger number. This is novel to both the numerical cognition literature and the
marketing literature. Second, we show that the subtraction task plays a key role in the evaluation
of sale prices, via influencing the propensity to actually calculate discount depth. Third, we show
that not all consumers will initiate the subtraction task, and whether or not consumers initiate the
subtraction task is dependent on the sale price display location. Fourth, we show that whether or
37
not initiation of the subtraction task to calculate discount depth impacts sale price evaluations, is
contingent on discount depth. For example, as shown in Studies 2A, 2B, and 4A, when discount
depth is low, using a sale price display location that facilitates the initiation of the subtraction
task to calculate discount depth does not impact evaluations. Even if consumers calculate the
discount depth, they identify discount depth as so low that it does not increase their evaluations.
However, when discount depth is moderate (see Studies 2A, 2B, 3B, 4A and 4B, and Appendix
2), using a sale price display location that facilitates initiation of the subtraction task causes more
consumers to identify this larger discount, such that more consumers have higher evaluations.
Such results are mediated by perceived difficulty/ ease of initiating the subtraction task (Study
3A and Study 3B). Finally, we show some specific cases where the above results reverse, and
also show the associated process.
This multi-study research article presents a unique opportunity to utilize meta-analytic
procedures to assess the robustness of the results. Using procedures suggested by Rosenthal and
Rosnow (2008), first we calculated the effect sizes associated with the key planned contrast (i.e.,
sale price on the left vs. right, for the moderate discount, see Appendix 2 for a table documenting
such results) (VP: S2A: η = .47, S2B: η = .42, S3B: η = .38, S4A: η = .46; S5: η = .39, S5R
(almost identical to study 5, not reported in the paper): η = .53; PI: S2A: η = .34, S2B: η = .42,
S3B: η = .33, S4A: η = .48; S4b: η = .25; S5: η = .35, S5R: η = .44), suggesting that these effect
sizes are not significantly different (VP: χ
2
(5) = 1.66, ns; PI: χ
2
(6) = 3.71, ns). The average
weighted η was .44 (VP) and .37 (PI), and the overall relationship between the two variables was
significant at p < .001. Second, we used Rosenthal and Rosnow’s (2008) file drawer technique to
determine that it would take over 180 null studies to reduce the significance of the results to the
.05 level, suggesting that the results in this paper are robust.
38
Managerial Implications
Retailers seek methods and tactics to increase the impact of sale prices, and such retailer
tactics have widespread implications for consumers, retailers, academics, and public policy
makers. As a result, expansive research has focused specifically on either the optimal discount
depth, or optimal sale price display locations. Relatively little work has linked these concepts, to
determine whether and how the optimality of a certain sale price display locations might be
contingent on discount depth.
Retailers typically want to present prices in ways that increase consumers’ evaluations.
Our research suggests that when they offer moderate discounts, retailers should use the sale price
display location to facilitate consumers’ initiation of the subtraction task. If consumers initiate
the subtraction task and identify the moderate discount, those consumers develop higher
evaluations. These findings offer a temptation to suggest that retailers should always display sale
prices to the right of original prices. With Study 5, we challenged this premise and show that at
both very low discount depths and exaggerated discount depths, retailers should use sale price
display locations that hinder initiation of the subtraction task. This is because when consumers
calculate these discount depths, they may suspect the retailer either of opportunistic motives (in
case of very low discounts), or question product quality (in case of exaggerated discounts). In
either case, consumers’ evaluations diminish. The complex connections across sale price display
locations and discount depth thus reveal non-obvious, key implications for retailers.
This research also offers clear recommendations by helping retailers understand which
sale price display location leads to the highest consumer evaluations under different discount
conditions. If a retailer is able to change its sale price display location often, contingent on
whether it offers moderate, very low, or exaggerated discounts, it should follow the guidelines
39
we have established to design its sale price display locations and increase consumers’ value
perceptions and purchase intentions. However, if a retailer is locked into a certain sale price
display location, it still can gain insights from our study. We present two exemplar applications,
though clearly, more applications are possible.
If the retailer’s sale price display location hinders the subtraction task (e.g., sale price
displayed to the left of the original price), it will suffer from lower value perceptions and
purchase intentions, even when it offers moderate discounts. Because this retailer recognizes the
root cause of this problem, it can provide other information to mitigate the negative effect, such
as the percentage discount information we considered in Study 3B.
If, however, the retailer’s sale price display location encourages the subtraction task, it
should, in certain cases, either avoid offering very low discounts or exaggerated discounts, which
lower value perceptions and purchase intentions (Study 5), or mitigate the problem by providing
external indicators such as credibility cues or quality cues (e.g., cues used in Studies 5A and 5B).
From a public policy perspective, both consumers and policy makers should recognize
that retailers can present sale prices to influence or potentially mislead consumers. For example,
in the case of very low discounts, retailers might choose to post sale prices to the left of the
original price, so that fewer consumers initiate the subtraction task to calculate the actual
discount depth. Such a practice is not illegal, but it may reduce market efficiency and consumer
surplus (see concerns espoused in Carlin 2009). Such insights may be useful for both consumer
education and to formulate effective regulatory policy.
Future Research
We realize that spatial representation is not just confined to the horizontal dimension
(Meier and Robinson 2004; Schubert 2005; Schwarz and Keus 2004). Consistent with the top-
40
heavy effect (similar to the right-heavy effect, Deng and Kahn 2009), top numbers are expected
to indicate larger magnitudes. Also, people’s eyes shift to lower locations when they expect to
find smaller (vs. larger) numbers (vertical SNARC effect, Ito and Hatta 2004). Therefore,
displaying the smaller number below the larger number is likely the norm, and because of this,
initiating subtraction tasks may be easier when the smaller number is presented below the larger
number. It would be useful to examine whether we can generalize the subtraction principle
across other sale price presentation formats, e.g. sale prices presented above versus below the
original price.
7
In this paper we considered only sale prices, which always involve a subtraction task (i.e.,
original price minus sale price), and not prices in general. Further research could examine if
conceptually similar results arise when we consider price formats that involve addition tasks, e.g.
in contexts like partitioned pricing. Finally, Dutta et al. (2011) demonstrated how the effects of
pricing tactics (e.g., low price guarantees) on consumers’ behaviors are moderated by
consumers’ regulatory focus (e.g., prevention vs. promotion focus). It would be useful to assess
if regulatory focus also plays a moderating role in influencing the effects of sale price location.
Finally, further research might investigate the effects of other dimensions of sale price display,
such as font color, font salience, or consider the role of additional moderator variables, such as
individual differences, goals, etc.
7
Results of two studies conducted by the authors suggest that locating the sale price below the original price is more
effective than above the reference price. Interested readers can obtain these results by contacting the authors.
41
References
Alba, Joseph. W., Carl F. Mela, Terrence 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.
Baroody, Arthur (1984), “Children's Difficulties in Subtraction: Some Causes and Questions,”
Journal for Research in Mathematics Education, Vol. 15, (3), 203-213.
Beishuizen, Meindert, C. M. Van Putten and F. Van Mulken (1997), “Mental Arithmetic and
Strategy Use with Indirect Number Problems up to One Hundred,” Learning and
Instruction, 7(1), 87-106
Biswas, Abhijit and Edward A. Blair (1991), "Contextual Effects of Reference Prices in Retail
Advertisements," Journal of Marketing, 55(3), 1-12.
Blair, Edward A. and E. Laird Landon, Jr. (1981), "The Effects of Reference Prices in Retail
Advertisements," Journal of Marketing, 45 (Spring), 61-69.
Brehm, Jack W. (1966). A Theory of Psychological Reactance. New York: Academic Press.
Campbell, Jamie I. D. and Quilin Xue (2001), "Cognitive Arithmetic Across Cultures," Journal
of Experimental Psychology: General, 130 (2), 299-315.
Carlin, Bruce I. (2009), "Strategic Price Complexity in Retail Financial Markets," Journal of
Financial Economics, 91(3), 278-287.
Carter, Travis and Thomas Gilovich (2010), “The Relative Relativity of Experiential and
Material Purchases,” Journal of Personality and Social Psychology, 98(1), 146−159.
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 (5), 752-6.
42
Coulter, Keith. S., and Robin A. Coulter (2005), "Size Does Matter: The Effects of Magnitude
Representation Congruency on Price Perceptions and Purchase Likelihood," Journal of
Consumer Psychology, 15 (1), 64-76.
Coulter, Keith. S., and Robin A. Coulter (2007), "Distortion of Price Discount Perceptions: the
Right Digit Effect," Journal of Consumer Research, 34 (2), 162-173.
Dehaene, Stanislas (1989), "The Psychophysics of Numerical Comparison: A Reexamination of
Apparently Incompatible Data," Perception and Psychophysics, 45 (6), 557–66.
Deng, Xiaoyan and Barbara E. Kahn (2009), "Is Your Product on the Right Side? The “Location
Effect” on Perceived Product Heaviness and Package Evaluation," Journal of Marketing
Research, 46(6), 725-738.
Dodds, William B., Kent B. Monroe and Dhruv Grewal (1991), "Effects of Price, Brand, and
Store Information on Buyers' Product Evaluations," Journal of Marketing Research, 28
(August), 307-19.
Dodson, Joe A., Alice M. Tybout and Brian Sternthal (1978). "Impact of Deals and Deal
Retraction on Brand Switching," Journal of Marketing Research, 15(1), 72-81.
Dutta, Sujay, Abhijit Biswas and Dhruv Grewal (2011), “Regret from Post-Purchase Discovery
of Lower Market Prices: Do Price Refunds Help?” Journal of Marketing, 75 (6), 124-
138.
Fuson, Karen C., Steven T. Smith, and Ana Maria LoCicero (1997), "Supporting Latino First-
Graders' Ten-Structured Thinking in Urban Classroom," Journal for Research in
Mathematics Education, 28, 738-66.
Grewal, Dhruv and Larry Compeau (2006), “Consumer Responses to Price and Its Contextual
Information Cues: A Synthesis of Past Research, a Conceptual Framework, and Avenues
43
for Further Research,” Review of Marketing Research, Volume 3, Naresh Malhotra (ed.),
M.E. Sharpe, 109-131.
Grewal, Dhruv, Howard Marmorstein, and Arun Sharma (1996), "Communicating Price
Information through Semantic Cues: The Moderating Effects of Situation and Discount
Size," Journal of Consumer Research 23, 148–155.
Gupta, Sunil and Lee G. Cooper (1992), “The Discounting of Discounts and Promotional
Thresholds,” Journal of Consumer Research 19, 401–411.
Hecht, Steven A. (1999), "Individual Solution Processes while Solving Addition and
Multiplication Math Facts in Adults," Memory & Cognition 27 (6), 1097-1107.
Helson, Harry (1964), Adaptation-Level Theory. New York: Harper & Row.
Imbo, Ineke, Ande Vandierendonck, and Yves Rosseel (2007), "The Influence of Problem
Features and Individual Differences on Strategic Performance in Simple Arithmetic,"
Memory & Cognition, 35(3), 454-463.
Inman, Jeffrey J., Leigh McAlister and Wayne Hoyer (1990), “Promotion Signal: Proxy for Price
Cut?” Journal of Consumer Research 17, 74-81
Ito, Yasuhiro and Takeshi Hatta (2004), "Spatial Structure of Quantitative Representation of
Numbers: Evidence from the SNARC Effect," Memory & Cognition 32, 662-673.
Lichtenstein, Donald R. and Scot Burton (1989), “The Relationship between Perceived and
Objective Price-Quality,” Journal of Marketing Research 26, 429-443
Lichtenstein, Donald R., Scot Burton, and Eric J. Karson (1991), "The Effect of Semantic Cues
on Consumer Perceptions of Reference Price Ads," Journal of Consumer Research 18,
380–391.
44
Mayhew, Glenn E. and Russell Winer (1992), “An Empirical Analysis of Internal and External
Reference Prices Using Scanner Data,” Journal of Consumer Research 19, 62-70
Meier, Brian P. and Michael D. Robinson (2004), "Why the Sunny Side Is Up: Associations
Between Affect and Vertical Position," Psychological Science, 15 (4), 243–47.
Monroe, Kent B. (2003), “Behavioral Foundations for Pricing Management,” in Pricing: Making
Profitable Decisions, Burr Ridge, IL: McGraw-Hill/Irwin.
Nelson, Leif D. and Joseph P. Simmons (2009), "On Southbound Ease and Northbound Fees:
Literal Consequences of the Metaphoric Link Between Vertical Position and Cardinal
Direction," Journal of Marketing Research, 46 (6), 715-724.
Porac, Clare and Stanley Coren (1976), "The Dominant Eye," Psychological Bulletin, 83(5),
880-897.
Preacher, Kristopher J., and Andrew F. Hayes (2004), “SPSS and SAS Procedures for Estimating
Indirect Effects in Simple Mediation Models,” Behavior Research Methods, Instruments,
& Computers, 36 (4), 717-731.
Preacher, Kristopher J., Derek D. Rucker, and Andrew F. Hayes (2007), “Addressing Moderated
Mediation Hypotheses: Theory, Methods, and Prescriptions,” Multivariate Behavioral
Research, 42 (1), 185-227.
Puccinelli, Nancy M., Rajesh Chandrashekaran, Dhruv Grewal, and Rajneesh Suri (2013),
“Are Men Seduced by Red? The Effect of Red Versus Black Prices on Price
Perceptions,” Journal of Retailing, (forthcoming).
Rosenthal, Robert and Ralph L. Rosnow (2008), Essentials of Behavioral Research: Methods
and Data Analysis (3rd ed.). Boston, MA: McGraw-Hill.
45
Schubert, Thomas W. (2005), “Your Highness: Vertical Positions as Perceptual Symbols of
Power,” Journal of Personality and Social Psychology, 89 (1), 1–21.
Schwarz, Wolfgang and Inge M. Keus (2004), "Moving the Eyes along the Mental Number Line:
Comparing SNARC Effects with Saccadic and Manual Responses," Attention, Perception
& Psychophysics, 66 (4), 651-664.
Schwarz, Norbert (2006), “Feelings, Fit, and Funny Effects: A Situated Cognition Perspective,”
Journal of Marketing Research, 43 (February), 20-23.
Selter, Christoph (2001), “Addition and Subtraction of Three-Digit Numbers: German
Elementary Children's Success, Methods and Strategies.” Educational Studies in
Mathematics, Vol. 47 (2), 145-173.
Shapiro, Stewart and Deborah J. MacInnis (1992), "Mapping the Relationship between
Preattentive Processing and Attitude", in Advances in Consumer Research eds. John F.
Sherry, Jr. and Brian Sternthal, Provo, UT: Association for Consumer Research, 19, 505-
513.
Sherif, Carolyn W. (1963), "Social Categorization as a Function of Latitude of Acceptance and
Series Range," Journal of Abnormal and Social Psychology, 67 (2), 148-156.
Simonson, Itamar, Ziv Carmon, and Suzanne O'Curry (1994), "Experimental Evidence on the
Negative Effect of Product Features and Sales Promotions on Brand Choice," Marketing
Science, 13(1), 23-40.
Thomas, Manoj and Vicki Morwitz (2005), "Penny Wise and Pound Foolish: The Left Digit
Effect in Price Cognition," Journal of Consumer Research, 32 (1), 54-64.
46
Thomas, Manoj and Vicki Morwitz (2008), "The Ease of Computation Effect: The Interplay of
Metacognitive Experiences and Naïve Theories in Judgments of Price Differences,"
Journal of Marketing Research, 46 (February), 81-91.
Thomas, Manoj, Daniel H. Simon, and Vrinda Kadiyali (2010), "The Price Precision Effect:
Evidence from Laboratory and Market Data," Marketing Science, 29(1), 175-190.
Urbany, Joel E., William O. Bearden, and Dan C. Weilbaker (1988), "The Effect of Plausible
and Exaggerated Reference Prices on Consumer Perceptions and Price Search," Journal
of Consumer Research, 15 (June), 95–110.
Valenzuela, Ana and Priya Raghubir (2011), “Are Top-Bottom Inferences Conscious and Left-
Right Inferences Automatic? Implications for Shelf Space Positions,” Unpublished
Working Paper, New York University.
Wegener, Duane T., Richard E. Petty, Kevin L. Blankenship, and Brian Detweiler-Bedell (2010),
"Elaboration and Numerical Anchoring: Implications of Attitude Theories for Consumer
Judgment and Decision Making," Journal of Consumer Psychology, 20 (1), 5-16.
Yip, Michael C. W. (2002), "Presentation Effects on Arithmetic Problem Solving," Psychologia,
45, 90-97.
47
Figure 1: Conceptual Framework and Roadmap
Display Location x
Moderate
Discount
Display Location x
Low Discount
Display Location x
Extremely Low
Discount
Display Location x
Exaggerated
Discount
Sale
price
left
Sale
price
right
Sale
price
left
Sale
price
right
Sale
price
left
Sale
price
right
Retailer
Opportunism
Quality
Concerns
x Credibility
Enhancement
x Quality
Enhancement
Studies 2A/ 2B/ 3B/ 4A/ 4B/ 5
Studies 2A/ 2B/ 3A (Easy calculations)/ 4A
Studies 5/ 5A
Studies 5/ 5B
Display Location of Smaller #
(or Sale Price) relative to Larger #
(or Original Price)
• Right vs. Left (#s S1)
Evaluations
Value Perceptions (VP)
Purchase Intentions (PI)
Choice
Discount Size
(Moderate/ Low/ Very
Low/ Exaggerated)
Perceived
Difficulty/
Estimated
Discount
Panel A
Panel B
Study 5A
Study 5B
Perceived
Difficulty/
Estimated
Discount
Sale
price
left
Study 3A (Easy calculations)/ 3B (Percent discount
provided/ No calculations required)
Sale
price
right
48
Figure 2: Impact of Discount Depth and Right versus Left Sale Price Display Location
(Study 2A)
3.82
3.69
3.53
4.69
1
2
3
4
5
6
7
LOW discount MODERATE discount
Sale price to LEFT
Sale price to RIGHT
Value Perceptions
2.89
3.01
2.82
4.19
1
2
3
4
5
6
7
LOW discount MODERATE discount
Sale price to LEFT
Sale price to RIGHT
Purchase Intentions
49
Figure 3: Impact of Discount Depth and Right versus Left Sale Price Display Location
(Study 4A – Non Student Sample)
3.96
3.79
3.72
4.93
1
2
3
4
5
6
7
LOW discount MODERATE discount
Sale price to LEFT
Sale price to RIGHT
Value Perceptions
2.92
3.23
3.15
4.81
1
2
3
4
5
6
7
LOW discount MODERATE discount
Sale price to LEFT
Sale price to RIGHT
Purchase Intentions
50
Figure 4: Impact of Discount Depth and Right versus Left Sale Price Display Location
(Study 5)
3.59
3.85
5.29
2.49
4.66
3.9
1
2
3
4
5
6
7
VERY LOW
discount
MODERATE
discount
EXAGGERATED
discount
Sale price to LEFT
Sale price to RIGHT
Value Perceptions
2.99
3.21
4.94
2.21
4.31
3.47
1
2
3
4
5
6
7
VERY LOW
discount
MODERATE
discount
EXAGGERATED
discount
Sale price to LEFT
Sale price to RIGHT
Purchase Intentions
51
Appendix 1
Value Perceptions: Four strongly disagree (1) to strongly agree (7) items were used. Reliabilities were .86 (S2A),
.93 (S2B), .90 (S3A), .88 (S3B), .85 (S4A) and .85 (S5).
The [product] offered by the retailer provides an excellent buy for the money.
The advertised offer represents an extremely large savings.
The price charged by the retailer for the [product] is extremely fair.
The price at which the [product] is offered by the retailer provides an extremely good value.
Purchase Intentions: Three very low (1) to very high (7) items. Reliabilities were .94 (S2A), .93 (S2B), .95 (S3A),
.92 (S3B), .94 (S4A), .96 (S4B) and .94 (S5).
The likelihood that I would buy the [product] at the sale price is-
The probability that I would consider buying the [product] is-
My willingness to buy the [product] at the sale price is-
Difficulty of Subtraction Task: Four strongly disagree (1) to strongly agree (7) items were used (all reverse scaled)
(reliability .91; S3B):
I found it easy to figure out the percentage discount.
The retailer made it easy to figure out the percentage discount.
The price display made it easy to figure out the percentage discount.
A typical consumer would have found it easy to figure out the percentage discount.
Web Appendix - Math Self-Concept: Four strongly disagree (1) to strongly agree (7) items were used (all reverse
scaled) (reliability .96; S3B):
I always believed that mathematics is one of my favorite subjects.
I learn mathematics quickly.
I understand even the most difficult concepts in my mathematics class.
I have always received good grades in mathematics.
Web Appendix - Preference for Numerical Information: Eight strongly disagree (1) to strongly agree (7) items
(reliability .90; S3B):
I enjoy work that requires the use of numbers.
I find it satisfying to solve day-to-day problems involving numbers.
Numerical information is very useful in daily life.
I like to make calculations using numerical information.
I like to go over numbers in my mind.
I prefer not to pay information to decisions involving numbers.
I don’t like to pay attention to issues involving numbers.
I don’t find numerical information relevant for most situations.
Quality Perceptions: Two strongly disagree (1) to strongly agree (7) items were used (reverse scaled) (correlation
.74; S5):
The retailer offered the discount to stimulate sales of a not-so-good Turkish carpet.
Given the discount level offered, there must be something wrong with the Turkish carpet.
Retailer Opportunism: Three strongly disagree (1) to strongly agree (7) items were used (reliability .92; S5):
The retailer advertises sale prices to attract consumers despite knowing that the sale prices are not very
attractive.
The retailer uses its sale prices as a trick to attract people.
I feel that the retailer is trying to take advantage of the value people place on merely seeing sale prices.
52
Appendix 2
Study
Value Perceptions
Purchase Intentions
Notes
2A
30%: L = 3.69 vs. R = 4.69*
10%: L = 3.82 vs. R = 3.53
30%: L = 3.01 vs. R = 4.19*
10%: L = 2.89 vs. R = 2.82
H2 supported
2B
30%: L = 3.83 vs. R = 4.91*
10%: L = 3.43 vs. R = 3.56
30%: L = 3.26 vs. R = 4.54*
10%: L = 3.18 vs. R = 3.08
H2 supported
3A
30%: L = 5.23 vs. R = 5.27
10%: L = 4.37 vs. R = 4.24
30%: L = 4.79 vs. R = 4.81
10%: L = 3.54 vs. R = 3.33
H2 not supported when
calculations are very simple
3B
No cue: L = 3.71 vs. R = 4.69*
With cue: L = 4.88 vs. R = 4.87
No cue: L = 3.50 vs. R = 4.35*
With cue: L = 4.41 vs. R = 4.42
H2 supported, BUT
H2 not supported when
calculations unnecessary
4A
30%: L = 3.79 vs. R = 4.93*
10%: L = 3.96 vs. R = 3.72
30%; L = 3.23 vs. R = 4.81*
10%: L = 2.92 vs. R = 3.15
H2 supported with
nonstudent sample
4B
30%/L = 3.62 vs. 30%/R = 4.50*
Preferring focal spoon:
L=18.9% vs. R = 38.9% #
H2 supported with
nonstudent sample + in a
real choice
5
4%: L = 3.59 vs. R = 2.49*
30%: L = 3.85 vs. R = 4.66*
85%: L = 5.29 vs. R = 3.90*
4%: L = 2.99 vs. R = 2.21*
30%: L = 3.21 vs. R = 4.31*
85%: L = 4.94 vs. R = 3.47*
H3 and H4 supported
5A
No cue: L = 3.84 vs. R = 3.11*
With cue: L = 3.94 vs. R = 3.97
No cue: L = 3.10 vs. R = 2.16*
With cue: L = 3.36 vs. R = 3.58
H3 supported, BUT
H3 not supported when
retailer opportunism
perceptions reduced
5B
No cue: L = 4.90 vs. R = 3.88*
With cue: L = 3.95 vs. R = 4.99*
No cue: L = 4.75 vs. R = 3.69*
With cue: L = 3.60 vs. R = 4.79*
H4 supported, BUT
H4 not supported when poor
product quality perceptions
reduced
Key: 30%: L → mean value for 30% discount depth, sale price to left of original price
* → p < .05 # → p = .06
