Content uploaded by Philip Gendall
Author content
All content in this area was uploaded by Philip Gendall
Content may be subject to copyright.
MESSAGE FRAMING EFFECTS ON PRICE DISCOUNTING
Philip Gendall, Janet Hoek and Tracy Pope
Massey University, Palmerston North
Abstract
Though the long-term value of price discounting is questionable, it remains a common
marketing practice. When price discounts are implemented there are different ways in which
the same discount can be expressed, or 'framed', and this may affect consumers' reactions.
This paper reports the results of a study designed to examine the effect on consumers' choice
behaviour of framing the same price discount as a percentage or as a dollar amount, for low-
priced and high-priced items.
For 'low-priced' items, cola drinks and potato chips, the framing of a price discount had little
or no effect. Only for potato chips was there any evidence to support the hypothesis that a
percentage discount would be more effective than a cents-off discount and this evidence was
very weak. However, for two 'high-priced' items, computers and stereos, framing a discount
as a dollar value was significantly more effective than expressing the same discount as a
percent off.
The implication for retailers and manufacturers is that, as a general policy, they are better to
express price discounts as dollars or cents off rather than as a percentage, particularly if the
absolute value of the discount is non-trivial.
Introduction
Price discounts are popular because they stimulate immediate purchase of the promoted
product, resulting in a sharp sales spike, their effects are observable and measurable, and,
because they help to increase store traffic, they may aid in maintaining manufacturers'
relationships with retailers, ensuring that the brand is well stocked and has adequate shelf
space. A price discount provides a brand with a feature that enhances its salience and
encourages consumers to consider that brand over others in their brand repertoire (Schultz,
Robinson and Petrison 1998; Ehrenberg, Hammond and Goodhardt 1994).
However, Ehrenberg et al (1994) have shown that price promotions are unlikely to attract new
buyers or result in any long term after-effects, since most of the consumers purchasing a
discounted brand are existing buyers who already have the brand in their consideration set.
Nevertheless, price discounting remains a popular form of instore promotion; in fact there is
evidence that the use of price promotions is increasing (Bearden and Urbany 1998). Thus the
most effective way of implementing a price discount is a relevant issue for most retailers and
many manufacturers, and one aspect of this is the way in which a price discount is expressed,
or 'framed'.
The fact that cognitive judgements are influenced by the way in which decision problems are
framed is well established (see Kahneman and Tversky 1979) and several researchers have
confirmed that the presentation, or framing, of messages about products affects consumers'
purchase intentions or behaviour. For example, Levin et al (1995, cited in Levin and Garth
1998) found that consumers' likelihood of purchasing ground beef was higher when the
ground beef was described (framed) in terms of its percent-lean rather than its percent-fat.
Similarly, Ganzach and Karsahi (1995) reported that message framing affected the behaviour
of credit card holders. A negatively-framed message (the loss incurred by using a cheque
instead of a credit card) produced higher card utilisation and charges than a positively-framed
message (the gains from using a credit card).
The relevance of message framing for price discounting is that there are different ways in
which the same price discounted can be expressed. For example, the same price discount can
be expressed as either a percentage amount off or a dollar amount off. In some cases there are
several ways of expressing the same discount, for example, "50% off", "buy one, get one
free", or "two for the price of one". The effect of these framing alternatives will depend on
how consumers process and interpret them.
Chen, Monroe and Lou (1998) hypothesised that, for high-priced items, consumers will see a
price reduction framed in dollar terms as more significant than when it is expressed in
percentage terms, and that the opposite would be true for low-priced products. Their rationale
was that, for a given percentage discount, the absolute value of the price reduction is higher
for higher-priced products, whereas, for a given absolute price discount, the relative
percentage reduction is higher for lower-priced products. Chen et al tested this proposition in
a study that used a $1595 personal computer and a $7.95 box of floppy disks. The study
confirmed their hypothesis. However, while the framing of the price discounts affected
respondents' evaluation of them, it did not have a significant effect on their purchase
intentions.
Chen et al's study was limited by the fact that it considered only one product in each product
category and made no allowance for brand effects. The research reported here replicates Chen
et al's study but extends it to two product categories at each price level and to three brands
within each product category.
Methodology
A stated-preference choice modelling experiment was conducted for two low-priced and two
high-priced product categories: cola drinks and potato chips, and computers and stereos.
Each product was represented by three brands and three price levels, including equivalent
percent and dollar discounts. The brands and price levels chosen are shown in Table 1.
Table 1 Product attributes and levels tested
Product Attributes Levels
Cola Brands Coca Cola Pepsi Foodtown Cola
Prices $1.00 Save 15c Save 10%
Chips Brands ETA Bluebird Krispa
Prices .80c Save 8c Save 15%
Computers Brands Compaq IBM Advantage
Prices $2,800 Save $420 Save 10%
Stereos Brands Sony Aiwa Phillips
Prices $899 Save $135 Save 15%
The choice sets constructed contained the three brands of each product, each at a specified
price or with the specified price crossed out and a price discount indicated by the words
"SPECIAL OFFER SAVE X ($ or c value, or %)". The same design, with six choice sets,
was used for each product, representing a total of 24 choice sets. However, to overcome the
potential problem of respondent fatigue, two versions of the questionnaire were produced,
each containing only two products (either cola drinks and computers, or potato chips and
stereos) and 12 choice sets.
The sample for the study consisted of 322 shoppers, selected by mall intercept in a Palmerston
North shopping centre in August 2000. Respondents were randomly allocated to one of the
two alternative questionnaire versions, giving sample sizes of 161 respondents for each
product category. Respondents were told to assume they were shopping for the particular
product shown and that the brands on the showcard were the only ones available ; they were
then asked to choose one of the three brands. Age, sex, education, brand awareness and
purchase information were also collected from respondents. This allowed us to establish that
the composition of the sub-samples was similar and to conclude that demographic or
consumption differences between the sub-samples would not affect product comparisons.
Results
The effects of message framing on price discounting were estimated using the multinomial
logit model. To estimate the model, respondents' choices were combined to produce
aggregate frequencies for each brand in each choice set. These aggregate frequencies were
then regressed against the matrix of attribute variables, which included brand, whether or not
the brand was discounted, and the way in which the discount message was framed. Models of
varying degrees of complexity were fitted; the regression coefficients and values for these
models are shown in Tables 2 and 3.
Table 2 Multinomial Logit Regression Results : Low-priced Products
Potato Chips Cola Drinks
Variable1BtVariable2Bt
ETA 0.72 8.0 Coca Cola 1.69 4.8
Bluebird 0.75 8.4 Foodtown Cola
-0.81
-3.5
Discount % 0.60 7.0 Discounted Coca Cola
1.23
5.0
Discount Value 0.45 5.1 Discounted Pepsi 0.23 0.7
Discounted Foodtown
-0.31
-0.7
Model Fit
X
2
, 4
df
= 141.9
(p<.10)
X
2
,6
df
= 1281.5
(p<.05)
Note: 1. Baseline brand is Krispa; baseline discount is No Discount
2. Baseline brand is Pepsi; baseline discount is No Discount.
For potato chips, a discount expressed as a percentage was marginally more attractive than a
discount framed in dollar terms (and both were more attractive than no discount). By contrast,
the type of discount had no effect on respondents' choice of cola drinks, though the effect of
discounting per se varied from brand to brand. Discounting makes Coca Cola significantly
more attractive, Pepsi more attractive (though not significantly) and Foodtown Cola less
attractive (though, again, not significantly).
Table 3 Multinomial Logit Regression Results : High-priced Products
Stereos Computers
Variable1BtVariable2Bt
Sony 0.87 10.0 IBM 1.66 4.5
Aiwa 0.06 0.6 Compaq 1.47 3.9
Discount Value 1.41
13.6
Advantage $ Discount 2.16 6.0
Discount % 0.99
9.3
IBM $ Discount 1.58 7.7
Compaq $ Discount 1.43 6.7
IBM % Discount 1.04 5.7
Compaq % Discount 0.96 4.7
Advantage % Discount 0.92 2.3
Model Fit X2,4df = 354.1
(p<.0001) X2, 8df = 505.7
(p<.05)
Note: 1. Baseline brand is Phillips; baseline discount is No Discount
2. Baseline brand is Advantage; baseline discount is No Discount.
For stereos, the effect of the alternative discount messages was clearly significant: a discount
expressed in dollar terms was more attractive than one expressed in percentage terms, and
both discounts outweighed any brand effect. Similarly, for personal computers there was a
significant difference between the attractiveness of the two discounts in favour of the discount
framed in dollar terms. Unlike the stereos, however, the strength of this effect varied
depending on the brand concerned. For example, the effect was more marked for the
Advantage brand than for either of the other two brands (possibly because Advantage was the
least popular of the three brands tested).
Discussion and Conclusions
This study was designed to test the hypothesis that it is more effective to express a price
discount for low-priced products as a percentage off the normal price, rather than as the actual
monetary value of the discount, and vice versa for high-priced products. The results provide
partial support for this hypothesis.
For the two low-priced products tested, the way in which the discount was framed had little
effect. For cola drinks, the discount framing had no effect at all, while for potato chips, the
discount was marginally more effective when expressed as a percentage than as cents amount
off. Though this difference was not significant, it was consistent with the results of Chen et
al's (1998) study and their hypothesis about discount message framing for low-priced
products.
For both high-priced products tested, the price discount was significantly more attractive
when expressed as a dollar amount off than as a percentage amount off. This is consistent
with Chen et al's (1998) results and provides support for their hypothesis about discount
message framing for more expensive products.
One explanation for the difference in results for potato chips and cola drinks is that
respondents were influenced by the relative absolute size of the discount in each case. For the
potato chips, the absolute value of the cents off amount was less than the absolute value of the
percentage amount off (8 cents versus 10%). By contrast, for the cola drinks, the cents-off
amount was greater than the absolute value of the percent-off amount (15 cents versus 10%).
If respondents reacted to these numbers, it would explain the greater attractiveness of the
percent discount for potato chips. But the same reasoning would suggest that the cents-off
discount should have been more attractive for the cola drinks, and it was not. (For cola
drinks, the utility of the Coca Cola brand effectively swamped all other effects.)
There are some limitations to this research. First, stereos and computers are infrequently-
bought products. If a respondent was in the market for either of these products, they would be
likely to have more knowledge about the technical features and quality of the different brands,
and brand may be more important than suggested in our experiment. Nevertheless, for
computers at least, there was a strong interaction between brand and price discount, which
suggests that respondents responded to brand as well as price and discount level in this
product category.
Second, although choice modelling experiments simulate reality, they do not replicate actual
purchase situations. In reality, discounted brands may be promoted with special signs to
emphasise the fact that they are discounted, and respondents may react differently to such
stimuli than to the presentation of the discounts in our experiment (though this should not
have affected the comparison of the alternative message framing approaches).
Despite these limitations, the practical implication for retailers and manufacturers using price
discounting seems clear. As a general policy, it is better to express price discounts as dollars
or cents off rather than as a percentage off. For high-priced products this conclusion is
unequivocal, for low-priced products it is less well-supported, but since there is little evidence
that percentage discounts are more effective, manufacturers and retailers can be reasonably
confident in using cents-off discounts for these products as well.
References
Bearden, W O and Urbany, J E (1998). Introduction to the special issue. Journal of Retailing,
74(3), 305-309.
Chen, S S, Monroe, K B and Lou, Y (1998). The effects of framing price promotion
messages on consumers' perceptions and purchase intentions. Journal of Retailing, 74(3),
353-372.
Ehrenberg, Andrew C, Hammond, Kathy and Goodhardt, Gerald (1994). The after-effects of
price-related consumer promotions. Journal of Advertising Research, 34, 11-21.
Ganzach, Y and Karsahi, N (1995). Message framing and buyer behaviour: A field
experiment. Journal of Business Research, 32, 11-17.
Kahneman, D and Tversky, A (1979). Prospect theory: An analysis of decision under risk.
Econometrica, 47, 263-291.
Levin, I P and Gaeth, G F (1988). How consumers are affected by the framing of attribute
information before and after consuming the product. Journal of Consumer Research, 15, 374-
378.
Schultz, D, Robinson, W and Petrison, L (1998). Sales promotion essentials: The 10 basic
sales promotion techniques…and how to use them. (3rd ed). Chicago: NTC Business Books.