84
Journal of Marketing
Vol. 71 (January 2007), 84–94
© 2007, American Marketing Association
ISSN: 0022-2429 (print), 1547-7185 (electronic)
Gangseog Ryu & Lawrence Feick
A Penny for Your Thoughts: Referral
Reward Programs and Referral
Likelihood
Because referral reward programs reward existing customers and build the customer base, firms use them to
encourage customers to make recommendations to others. The authors report on four experiments in which they
find that rewards increase referral likelihood. More specifically, they find that rewards are particularly effective in
increasing referral to weak ties and for weaker brands. It is also important who receives the reward. Overall, for
weak ties and weaker brands, giving a reward to the provider of the recommendation is important. For strong ties
and stronger brands, providing at least some of the reward to the receiver of the referral seems to be more effective.
The authors discuss the implications of the results for the design of reward programs.
Gangseog Ryu is Associate Professor of Marketing, College of Business
Administration, Korea University (e-mail: gryu@korea.ac.kr). Lawrence
Feick is Professor of Business Administration, Katz Graduate School of
Business, University of Pittsburgh (e-mail: feick@katz.pitt.edu). The
authors thank the three anonymous JM reviewers for their guidance, Sity
Norani Binte Rohani for her help in stimuli development and data collection,
and Jeff Inman and the Consumer Behavior doctoral seminar participants
at the University of Pittsburgh for their many helpful comments on a
prior draft of this article. The data for Studies 1 and 3 were collected while
the first author was with the National University of Singapore. The SK
Award, given to the first author, provided partial support for this research.
To read and contribute to reader and author dialogue on JM, visit
http://www.marketingpower.com/jmblog.
Word of mouth (WOM), at one time viewed as a
sociological phenomenon to be observed and
described, is increasingly considered a marketing
tool to be managed (Rosen 2000). In initial efforts to manage
WOM, marketers focused on satisfying customers (so
that they would generate positive rather than negative
WOM) and on targeting influential consumers, such as
opinion leaders. Only recently have firms introduced formal
programs that are designed to encourage existing customers
to make product recommendations. In these programs,
firms offer various types of rewards (e.g., vouchers, gifts,
free minutes, miles) when an existing customer attracts a
new customer. Such programs have significant potential to
affect firms’ performance. For example, consider customer
relationship management (CRM), recently a central focus
of marketers’ attention (Hogan, Lemon, and Libai 2003;
Rust, Lemon, and Zeithaml 2004). A core idea in CRM is
that firms need to invest in retaining existing customers not
just in finding new ones. Thus, a referral reward program
can be a key CRM tool because in addition to its potential
to attract new customers, it can improve retention by
rewarding existing customers.
A review of Web search results suggests that the practice
of rewarded referral is widespread. Referral reward programs
exist for a wide variety of goods (e.g., contact lenses
[Smart View Contacts], automobiles [Hanmar Motors], pet
supplies [Neo-Paws]) and services (e.g., airlines [United
Airlines], mobile phone service [Cingular]). They provide
rewards that can be discounts on the product or service
(e.g., discounted future hotel stays [Caesar’s Pocono
Resort], free perfume [Perfumeoutlet.net]) or cash and gifts
unrelated to the product or service category being recommended
(e.g., car dealers [#1 Cochran], real estate agencies
[REMAX]). Researchers expect an increasing use of such
programs because of their targetability and cost effectiveness
compared with more traditional promotional tools
(Mummert 2000).
However, there is limited research on rewarded referral.
Some scholars have examined the impact of various marketing
activities on WOM. For example, Bolton, Kannan, and
Bramlett (2000) show that offering a loyalty program has an
indirect effect on customers’ repatronage behavior and on
their WOM. Blodgett, Hill, and Tax (1997) find that the
way customers are treated has a greater impact on complaining
customers’ negative WOM intention than does
offering discounts. Apart from these indirect approaches to
studying the effects of offering rewards on WOM, two
recent analytical studies offer guidance for developing optimal
referral reward programs. First, on the basis of the
premise that customers make recommendations when they
are delighted, Biyalogorsky, Gerstner, and Libai (2000)
identify conditions under which a referral reward program
is more effective than a price reduction in enhancing the
firm’s profitability. Second, Chen and Shi (2001) use game
theory to demonstrate that the optimal referral reward (cash
versus free products) varies according to market structure
(monopoly versus oligopoly). Overall, however, there has
been almost no empirical work.
This article reports the results of four laboratory experiments
that investigate the impact of referral reward programs
on referral likelihood. Across our studies, we examine
the effect of the presence or absence of a reward, reward
size, and reward recipient (i.e., does the existing customer,
Referral Reward Programs and Referral Likelihood / 85
does the new customer, or do both get the reward?). We also
examine the impact of the relationship between the recommender
and the receiver of the recommendation (i.e., strong
or weak ties) and of brand strength (i.e., stronger or weaker
brands). Our results have implications for the design of
reward programs and for WOM theory.
Conceptual Background
We adopt an exchange theory framework for examining
consumers’ responses to referral reward programs. From
this perspective, the customer’s decision about whether to
engage in WOM depends on the perceived costs and benefits
of the exchange (Frenzen and Nakamoto 1993;
Gatignon and Robertson 1986; Walster, Walster, and
Berscheid 1978). Research on WOM identifies several (primarily
psychological or social) benefits of or motivations
for transmitting WOM (Arndt 1967; Dichter 1966;
Gatignon and Robertson 1986). For example, consumers
may use WOM in an attempt to reduce postpurchase anxiety
or dissonance by talking about their product experiences.
In addition, consumers may use WOM as a way to
manage others’ impressions of them. Finally, WOM can be
a means of expressing concern about others and helping
them make better choices. Providing WOM also involves
costs (Gatignon and Robertson 1986). The most obvious
cost is the effort and time spent communicating. In addition,
because of the norm of reciprocity, the recommender may
feel obliged to be a “good listener” in future communication.
Finally, there is the risk that if the receiver is dissatisfied
with a purchase that results from the recommendation,
the relationship will suffer (Folkes 1984).
With referral reward programs, exchange is more complex.
As with natural WOM, there is the communication
between the recommender and the receiver of a referral. If
the referral successfully leads the receiver to purchase the
product, another exchange will take place between the
receiver and the company. Finally, the recommender
receives a reward from the company in return for the referral.
Because obtaining the reward is a result of the purchase,
the receiver (new customer) provides a benefit (albeit indirectly)
to the recommender (existing customer).
We expect that this added exchange complexity affects
how consumers perceive a referral’s costs and benefits. The
obvious additional benefit is the (economic) gain from the
referral. On the cost side, if the receiver of the rewarded
referral is dissatisfied with the product, there is increased
(social) risk that the receiver might attribute his or her dissatisfaction
to the recommender. Self-perception theory
(Bem 1965, 1972) suggests that people try to understand the
causes of their own behavior. If a referral is rewarded, consumers
may perceive the referral as being driven by the
reward rather than by intrinsic motivations. As a result, recommenders
may feel as if they “sold” their recommendation,
a perception unlikely to be consistent with their selfimage.
For a firm to consider a referral reward program effective,
the program should show results beyond what would
occur naturally. In other words, the marginal gain from the
reward program should be large enough to compensate for
its cost. Thus, natural WOM is the appropriate comparison
in the evaluation of the effects of referral reward programs.
Study 1
In Study 1, we investigate whether the presence or absence
of a reward and reward size influence satisfied customers’
referral likelihood. In addition, we examine the moderating
effects of strong versus weak ties between the recommender
and the receiver and of the strength of the brand offering the
reward.
Referral Rewards and Tie Strength
Unlike more traditional sales promotion and customer loyalty
programs that involve only the customer and company,
referral reward programs also have implications for other
consumers. Consumers consider the value of potential gains
and costs for themselves and for the other consumer in
rewarded referral. The nature of the recommender–receiver
relationship influences these perceptions of costs and benefits.
Research shows that tie strength is an important property
of relationships in determining how social context
affects referral (Brown and Reingen 1987; Reingen and
Kernan 1986). Tie strength varies from strong primary, such
as spouse or close friends, to weak secondary, such as
seldom-contacted acquaintances (Reingen and Kernan
1986).
With strong ties, people tend to have communal relationships
in which they feel general concern about the other
person’s welfare. They respond to the other’s needs but do
not expect anything in return (Clark 1984; Clark, Mills, and
Powell 1986; Frenzen and Nakamoto 1993). Conversely,
with weak ties, people typically have exchange relationships,
driven primarily by self-interest. In such relationships,
participants do not feel any special responsibility for
the other person and try to maximize their own outcomes
and minimize their costs. With weak ties, reciprocity is
important; people expect to get back what they put in.
People prefer an equitable (balanced) exchange, and if it is
unbalanced, they feel distress and try to adjust. In such
cases, they try to reduce what they give or increase what
they receive to achieve equilibrium (Walster, Berscheid, and
Walster 1973; Walster, Walster, and Berscheid 1978).
Research on naturally occurring WOM finds that consumers
are more likely to make a referral to a strong tie than
to a weak tie (e.g., Brown and Reingen 1987; Frenzen and
Nakamoto 1993), perhaps because their communal orientation
toward strong ties motivates them to share the pleasure
that they have received from using a product. Furthermore,
people know much more about the needs and preferences of
strong ties because they are in frequent contact with them
(Granovetter 1973) and keep track of their needs (Clark,
Mills, and Powell 1986). Greater knowledge about preferences
is likely to make consumers feel more comfortable
about sharing experiences and make the information more
useful, especially for products from high-preferenceheterogeneity
categories (Feick and Higie 1992).
What changes are expected when a reward is offered for
a referral? We consider strong ties. On the benefit side, a
reward should have little additional impact. Consumers help
86 / Journal of Marketing, January 2007
without expecting a reward; indeed, helping is its own
reward (Beach and Carter 1976; Clark, Mills, and Powell
1986). Frenzen and Nakamoto (1993) find that consumers
are likely to share all types of information with strong ties
(including both high- and low-value information). On the
cost side, there is the potential social risk of negatively
affecting a relationship if an economically driven referral
does not work out. Thus, with strong ties, because the marginal
benefit is small and because of potential (social) costs,
we expect little impact of a reward on referral likelihood.
In contrast, for weak ties, equity theory suggests that a
recommender will regard a referral as a favor done for the
receiver and/or the company. Thus, referrals yield inequity
because an input is increased without a simultaneous
increase in output (Walster, Berscheid, and Walster 1973;
Walster, Walster, and Berscheid 1978). If the recommender
receives a reward, the level of inequity is reduced, resulting
in movement toward equilibrium. Furthermore, unless the
economic value of the reward is large, it seems unlikely that
a concern about “being bought” will be important to the
recommender in interacting with a weak tie. Thus, with
weak ties (in exchange relationships), we expect that consumers
will be more influenced by economic motives and
less concerned about the social or psychological risks of
rewarded referral (Frenzen and Nakamoto 1993; Törnblom
and Nilsson 1993). Thus:
H1: The impact of offering a reward on referral likelihood
depends on tie strength. The presence of a reward (compared
with no reward) increases referral likelihood more
for weak ties than for strong ties.
Referral Rewards and Brand Strength
Brand features also should moderate the effect of rewards
on referral likelihood. We focus on brand strength. Following
Keller’s (1993) work, we conceptualize strong brands as
enjoying high brand awareness and well-established brand
associations. Research shows that consumers respond to
stronger and weaker brands differently. For example, a
given price reduction will have a greater sales impact when
it is applied to a higher-quality (stronger) brand than when
it is applied to a lower-quality (weaker) brand (e.g., Blattberg
and Wisniewski 1989; Heath et al. 2000). We also
expect brand strength effects for rewarded referral. Consumers
of stronger brands feel more brand commitment
than consumers of weaker brands because their choices are
affected more by preferences than by budget limits (Blattberg
and Wisniewski 1989). This stronger commitment
gives consumers of a stronger brand more confidence in
making recommendations, thus increasing (unrewarded)
referral likelihood and limiting the incremental impact of a
reward. Furthermore, rewarding referral by consumers of a
stronger brand may yield a perception that the reward compromises
their commitment to and confidence in the brand
(Bem 1965, 1972).
In contrast, consumers of a weaker brand are likely to
have greater residual desire or lower choice confidence
(Heath et al. 2000; Simonson 1992) and, thus, less WOM
motivation. For these consumers, a reward may compensate
for this lower residual desire and increase their choice confidence.
Furthermore, the two groups may perceive the
value of a reward differently. If consumers of weaker brands
put more weight on price than consumers of stronger brands
(Kamakura and Russell 1989), they would perceive a higher
value from the (economic) reward and be more likely to be
attracted by the reward program. Thus:
H2: The impact of offering a reward on referral likelihood
depends on brand strength. The presence of a reward
(compared with no reward) increases referral likelihood
more for weaker brands than for stronger brands.
Method
Participants and design. Study 1 was a 3 × 2 × 2
between-subjects factorial experiment in which we varied
the reward size (no reward, smaller, larger), the brand
strength (weaker, stronger), and the tie strength (strong,
weak). Two hundred seventy-five undergraduate students
from a major university in Singapore were randomly
assigned to the experimental conditions, except for brand
strength, which we manipulated by having the respondents
choose either a weaker or a stronger brand.
Procedure. We used portable MP3 players as the product
category for the study because of the relevance of this
product to the student participants. Students were first
asked to imagine being in the market for an MP3 player and
then were asked to pick the brand they preferred from the
two alternatives (i.e., the stronger and weaker brand). To
avoid the influence of prior brand beliefs, we did not use
actual brand names, instead labeling the brands A and B.
We manipulated brand strength in two ways. First, a brief
description of reputation and quality was presented for each
brand. The stronger brand was described as “one of the
leading brands of electronic products, recognized for its
high quality and reputation,” and the weaker brand was
described as a “relatively less-well-known player in the
electronic products market, known for its reasonable quality
and reputation.” Second, expert ratings on several dimensions
of the brand were provided, in which the quality rating
and price of the stronger brand were higher (four of five
stars, S$549, or approximately US$350) than those of the
weaker brand (three stars, S$399). A picture and detailed
product specifications were also included.
After making a choice, participants were asked to imagine
that they had bought, used, and been very satisfied with
the brand they had chosen. Product experience details were
included to reinforce their satisfaction with the brand. The
referral reward manipulation stated, “You are reminded that
when you made the purchase, the salesperson told you that
if you recommend the product to someone who then purchases
the same model, the manufacturer would give you a
shopping voucher of (S$50 [40], S$100 [80]), which can be
redeemed at a leading local department store.” We did not
include this information in the no-reward condition.
Note that we used 10% and 20% of the product prices
for the smaller and larger rewards, respectively. The use of a
relative (rather than absolute) amount to determine reward
size is consistent with the principle of relativity (Heath et al.
2000). We manipulated tie strength by asking participants to
identify (using initials) either “one of your closest friends”
Referral Reward Programs and Referral Likelihood / 87
FIGURE 1
Study 1 Results
A: The Effect of Reward Size and Tie Strength on
Referral Likelihood
B: The Effect of Reward Size and Brand Strength on
Referral Likelihood
for strong ties or a “casual acquaintance⎯someone you
interact with from time to time, but someone not close
enough to count as a friend (e.g., a classmate you have
recently met)” for weak ties (Frenzen and Nakamoto 1993).
Next, participants indicated their referral likelihood on a
scale anchored by 0% (“certain not to tell this person”) and
100% (“certain to tell this person”). Manipulation check
and covariate measures included tie strength (a four-item
scale that Frenzen and Davis [1990] developed) and perceived
reward size (the average of two nine-point items: “a
very small amount/a very large amount” and “very unattractive/
very attractive”). Then, participants were asked
whether they could name the MP3 players used in the
experiment. Finally, we included a nine-item, nine-point
scale of product involvement (adapted from the work of
Lichtenstein, Bloch, and Black [1988]) and the six-item
market maven scale (Feick and Price 1987) to control for
individual differences.
Results
Manipulation checks. Participants perceived the value of
the larger reward as significantly greater than the smaller
reward (mean smaller = 5.40, mean larger = 6.17; t(164) =
3.84, p < .01). Furthermore, mean tie strength ratings differed
significantly (mean strong = .86, mean weak = .59;
t(247) = 12.90, p < .01). Finally, 6.2% of the participants
correctly identified at least one of the brands in the experiment
and thus were dropped.
Referral likelihood. We analyzed referral likelihood
with an analysis of covariance (ANCOVA). Participants’
product involvement and market maven scores were covariates,
and reward size, brand strength, and tie strength were
between-subjects factors. The analysis yielded a significant
main effect for market maven (F(1, 236) = 8.74, p < .01; β =
.03). As market maven tendency increases, referral likelihood
increases. This result is consistent with previous findings
that indicate that market mavens exhibit a higher level
of information provision than other consumers (Feick and
Price 1987).
Two main effects for experimental variables were significant:
tie strength (F(1, 236) = 63.49, p < .01) and reward
(F(2, 236) = 22.67, p < .01). Referral likelihood was greater
with strong ties (88.1%) than with weak ties (71.3%), a pattern
consistent with previous studies (e.g., Frenzen and
Nakamoto 1993). In addition, offering a reward significantly
increased referral likelihood (no reward = 69.7%,
smaller reward = 84.7%, larger reward = 84.1%). Contrasts
of means revealed a significant difference between the noreward
and the smaller-reward conditions (t(168) = 4.89,
p < .01) and between the no-reward and the larger-reward
conditions (t(161) = 4.36, p < .01). The difference between
the smaller- and larger-reward conditions was not significant
(t(165) = .27, p = .79).
The main effect of reward was moderated by two significant
two-way interactions (see Figure 1, Panels A and
B). As H1 proposed, there was a significant interaction
between reward and tie strength (F(2, 236) = 22.65, p <
.01). With strong ties, the presence of a reward did not
affect referral likelihood (no reward = 87.2%, smaller
reward = 89.7%, larger reward = 87.3%; F(2, 114) = .19,
p = .83). In contrast, with weak ties, offering a reward
increased consumers’ referral likelihood (no reward =
52.6%, smaller reward = 79.8%, larger reward = 81.1%;
F(2, 120) = 35.42, p < .01).
As H2 proposed, there was a significant interaction
between reward and brand strength (F(2, 236) = 5.28, p <
.01). Offering a reward increased referral likelihood by
more than 20 percentage points for the weaker brand but by
less than 10 percentage points for the stronger brand
(weaker brand: no reward = 63.7%, smaller reward = 86%,
larger reward = 85.9%; F(2, 118) = 20.28, p < .001; stronger
brand: no reward = 75.1%, smaller reward = 83.1%, larger
reward = 82.6%; F(2, 126) = 2.52, p < .10). The effect is
driven by the no-reward condition in which referral likelihood
for the stronger brand is significantly higher than for
the weaker brand (75.1% versus 63.7%; t(81) = 2.08, p <
.05).
To gain insight into the mechanism underlying the
effects of brand strength, we compared participants who
chose the stronger (51.6%) and weaker (48.4%) brands.
Participants evaluated both brands after their choice, and we
computed the difference between the evaluation of the
brands chosen and those not chosen. This difference was
88 / Journal of Marketing, January 2007
significantly larger for participants who chose the stronger
brand than for those who chose the weaker brand (1.71 versus
.53; t(248) = 6.64, p < .01). In addition, participants’
relative ratings of the importance of price and quality in
evaluating and choosing an MP3 player differed significantly
(t(247) = 9.82, p < .01). Those who chose the
stronger brand gave larger weights to quality than did those
who chose the weaker brand (61.2/100 versus 46.5/100,
respectively).
Discussion
Compared with offering no reward, offering a reward
increased the likelihood of referral. Thus, the results suggest
that referral reward programs can be effective. However,
at least with the reward sizes we used, size does not
matter. An increase in reward size did not increase referral
likelihood. There are several possible explanations for this
result. There may simply be a ceiling effect; that is,
although the participants perceived the reward sizes as significantly
different (based on the manipulation check), the
smaller reward may have been large enough to generate the
same effect as the larger reward. In addition, and more substantively,
it may be that increases in reward size (which
increase benefits of referral) also increase psychological
and social costs by creating feelings of guilt about the
inequity of the exchange (Austin and Walster 1974; Smith,
Bolton, and Wagner 1999).
Weaker brands benefited more from offering a reward
than did stronger brands. This result may be due to consumers
of weaker brands having a lower level of commitment
to or confidence in their brand and price being more
important to them. Our results suggest the usefulness of
referral reward programs for weaker brands.
Finally, consumers behaved differently when a reward
was offered only with weak ties. With strong ties, there was
no increase in referral likelihood with a reward. This point
is critical in thinking about the marginal impact of referral
reward programs. Programs need to be carefully crafted to
achieve results greater than those that would have occurred
in due course through naturally occurring WOM.
Study 2
Study 1 results are consistent with exchange theory, but we
designed Study 2 to gain greater insight into the mechanism
underlying the effects. With strong ties, we expect consumers
to place greater weight on the sociopsychological
costs and benefits of making a referral. If they are rewarded,
they will produce thoughts related to these costs and benefits.
With weak ties, we expect consumers to give less
weight to (and produce fewer thoughts about) sociopsychological
costs and benefits but to give more consideration to
the economic benefits of the interaction.
Method
Study 2 was a 2 × 2 between-subjects design in which we
manipulated the presence of a reward (no, small) and tie
strength (strong, weak). Eighty-one Singaporean students
were randomly assigned to the experimental conditions.
Study 2 differed from Study 1 in that participants did not
make a product choice but rather were asked to imagine that
they had bought an MP3 player and were very satisfied with
it. Furthermore, we provided a few detailed experiences
with the product to simulate usage. In addition to Study 1
measures, in Study 2, we collected written thoughts about
making a referral after collecting referral likelihood. We
also measured perceptions of a referral’s benefits and costs,
using items developed from existing WOM research (Arndt
1967; Dichter 1966; Gatignon and Robertson 1986): For
social benefits (α = .89), we measured “others’ perceptions
of showing genuine concern,” “helping others make the best
choice,” and “developing (maintaining) a good relationship
with others,” and for psychological costs (r = .86), we measured
“feeling of being selfish” and “feeling of being motivated
by money.” Each item was evaluated on an 11-point
(–5 to +5) scale, indicating how good or bad the consequence
would be. More extreme ratings indicate higher
weights.
Results
Perception ratings. Referral likelihood results replicate
Study 1, and thus we do not describe them. To examine the
mechanism, we ran 2 × 2 analyses of variance on the perceived
social and psychological costs and benefits of referral.
Tie strength (F(1, 77) = 20.4, p < .01) and reward
(F(1, 77) = 4.60, p < .05) had significant main effects on
social benefits. There are greater perceived social benefits
with referral to a strong (3.14) than to a weak (1.59) tie and
for no reward than for a reward (no reward = 2.74, reward =
1.95). The interaction between reward and tie strength was
significant (F(1, 77) = 3.86, p < .05). With strong ties, there
was little difference in the evaluation of social benefits,
regardless of whether a reward was offered (no reward =
3.17, reward = 3.11; t(39) = .148, p > .85). With weak ties,
less social benefit was perceived with a reward (.90) than
without (2.30; t(40) = 2.61, p < .01).
Tie strength yielded a significant main effect on psychological
costs (F(1, 77) = 11.70, p < .01). Perceived costs
were greater for referral to a strong (–3.35) than to a weak
(–1.73) tie. A reward also resulted in a significant main
effect (F(1, 77) = 19.66, p < .01). Participants evaluated perceived
costs less negatively when there was a reward than
when there was no reward (no reward = –3.56, reward =
–1.48). In addition, the interaction between reward and tie
strength was significant (F(1, 77) = 4.35, p < .05). There
was a greater difference between no reward and reward with
weak ties than with strong ties (weak tie: no reward =
–3.25, reward = –.29; strong tie: no reward = –3.86,
reward = –2.79).
Thought listings. We developed three categories of
thought types: social benefits, social costs, and psychological
costs. Two independent judges who were blind to the
study’s purpose coded the data. Agreement between the
judges was high (92.5%); disagreements were resolved by
discussion. We ran separate 2 × 2 analyses of variance on
the number of thoughts within each of the three categories.
For social benefits, there was only a significant main effect
of tie strength (F(1, 77) = 14.57, p < .01). There were more
thoughts about social benefits of a referral for strong ties
Referral Reward Programs and Referral Likelihood / 89
(1.58) than for weak ties (.68). For social costs, the main
effects of reward and tie strength were both significant
(F(1, 77) = 5.60 and 7.37, p < .05 and p < .01, respectively).
There were more thoughts about social costs with a reward
(.55) than without (.24) and when the referral was to a
strong (.58) than to a weak (.22) tie. Moreover, the interaction
between reward and tie strength was significant
(F(1, 77) = 4.36, p < .05). With weak ties, participants
showed little difference in their thoughts about social costs,
regardless of whether a reward was offered (no reward =
.20, reward = .24; t(40) = .25, p = .80). With strong ties,
however, there were more thoughts about social costs with a
reward (.89) than without (.29; t(39) = 2.65, p < .01).
For psychological costs, there was a significant main
effect of reward and tie strength (F(1, 77) = 52.04 and 4.31,
p < .01 and p < .05, respectively). Participants generated
more thoughts about the costs of making a referral when a
reward was offered (.73) than when it was not (.00) and to
strong (.45) than to weak (.29) ties. In addition, the interaction
between reward and tie strength was significant
(F(1, 77) = 4.31, p < .05). A reward yielded a greater
increase in thoughts about psychological costs with strong
ties than with weak ties (weak tie: no reward = .00,
reward = .52; strong tie: no reward = .00, reward = .95).
Discussion
Study 2 provides direct support for our exchange theory
argument. In general, the scales we used to measure costs
and benefits and the thought listings were consistent and
suggest that consumers evaluate the social and psychological
costs and benefits of a referral differently when a reward
is involved. The potential social benefits of a referral are
perceived as lower and the social and psychological costs
are perceived as higher when a reward is offered than when
one is not. Tie strength also matters. Consumers perceive
greater potential social and psychological costs and benefits
when the referral is to strong ties in the presence of rewards.
In contrast, consumers tend to discount the importance of
social and psychological costs and benefits and are more
likely to recognize economic benefits when the referral is to
weak ties.
Study 3
Studies 1 and 2 suggest that rewards matter and that their
impact depends on tie strength and brand strength. What
cannot be determined from these studies is whether it matters
to whom the reward is given.
Referral Reward Schemes
Sales promotion benefits typically are provided only to the
customer who makes use of the promotion. However, in
customer referral programs, because there is an existing
customer who makes a recommendation and a new customer
who receives the referral, there could be three reward
schemes. The first is “Reward Me,” in which the recommender
(the existing customer) receives the reward. This
scheme is the focus in Studies 1 and 2 and, seemingly, the
most typical in practice. In addition, however, there could
be a scheme called “Reward You,” in which the receiver of
the recommendation (the new customer) receives the
reward. Finally, there could be a blend of the two scenarios,
or “Reward Both.”
Costs and benefits vary depending on the scheme. The
personal economic benefit that accrues to the existing customer
is affected. He or she receives the full benefit for
Reward Me, partial benefit for Reward Both, and nothing
for Reward You, but the potential social and psychological
costs decrease in the same order. The offsetting of economic
gains by psychological/social costs makes it difficult
to predict a main effect of reward scheme.
We propose tie strength as a moderator. We expect that
consumers will weight benefits and costs differently and
will be governed by different rules of exchange depending
on the receiver of the recommendation. In recommending to
weak ties, consumers are motivated more by self-interest
and are less concerned about psychosocial dimensions.
Thus, with weak ties, we expect the lowest referral likelihood
for Reward You. Our weak ties prediction is also
based on an equity argument; specifically, people expect to
receive resources (e.g., rewards) in exchange for resources
(e.g., referral) they provide (Walster, Berscheid, and Walster
1973; Walster, Walster, and Berscheid 1978). With Reward
You, inequity exists because the consumer receives nothing
in return for a referral. Conversely, with Reward Me, the
recommender is rewarded, and balance is achieved, increasing
referral likelihood. Reward Both should be in between.
With strong ties, consumers attempt to maximize
mutual benefits for the relationship as a whole, regardless
of which member receives the greater individual reward
(Kelly 1979; Kelly and Thibaut 1978). Furthermore, consumers
are likely to help strong ties without expecting any
economic return and enjoy psychological and/or social
benefits by helping a strong tie obtain a reward. This logic
implies an ordering for referral likelihood that is the opposite
of that for weak ties. Thus:
H3: The impact of reward allocation scheme on referral likelihood
depends on tie strength. With weak ties, consumers
are most likely to make a referral in the Reward Me condition,
followed by Reward Both and Reward You. With
strong ties, consumers are most likely to make a referral in
the Reward You condition, followed by Reward Both and
Reward Me.
Method
One hundred thirty-six undergraduate students from a Singaporean
university participated in a 3 × 2 between-subjects
factorial experiment in which we varied the scheme of the
reward distribution (Reward Me, Reward You, Reward
Both) and tie strength (strong, weak). The experimental
procedure of Study 3 was identical to Study 2 except for the
reward scheme. We manipulated scheme by varying who
received the referral reward (a shopping voucher): the recommender,
the receiver of the recommendation, or both.
Results
Manipulation checks. Manipulations of reward scheme
and tie strength were successful. Participants perceived
reward value similarly across reward schemes (Reward
Me = 6.29, Reward You = 5.95, Reward Both = 5.83;
90 / Journal of Marketing, January 2007
F(2, 133) = 1.84, p = .16), and tie strength was significantly
greater with strong ties (.85) than with weak ties (.55;
t(135) = 12.21, p < .01).
Referral likelihood. We ran an ANCOVA on referral
likelihood with reward scheme and tie strength as betweensubjects
factors and product involvement and market maven
as covariates. As in Study 1, the market maven covariate
was significant (F(1, 128) = 8.86, p < .01), and there was a
significant main effect of tie strength (strong ties = 86.5%,
weak ties = 68.7%; F(1, 128) = 42.53, p < .01).
As H3 predicted, there was a significant interaction
between reward scheme and tie strength (F(2, 128) = 4.54,
p < .01; see Figure 2). With strong ties, differences among
reward schemes were not significant, though they were
directionally consistent with our prediction (F(2, 62) = .98,
p = .38; Reward You = 90.1%, Reward Both = 86.7%, and
Reward Me = 82.8%). Conversely, with weak ties, differences
were significant (F(2, 64) = 4.37, p < .05) and in the
predicted order (Reward You = 58.3%, Reward Both =
70.4%, and Reward Me 75.8%). Reward You is marginally
lower than Reward Both (t(42) = 1.90, p = .06) and is lower
than Reward Me (t(44) = 3.15, p < .01), but the latter two
are not significantly different from each other (t(46) = 1.21,
p = .23).
Discussion
Study 3 illustrates the importance of considering the target
of the reward when designing referral reward programs. The
reward’s distribution between the existing and the new customer
had different effects on consumers’ likelihood of
making a referral, depending on the relationship between
the recommender and the receiver of the recommendation.
Study 3 showed that varying the recipient of the reward has
little effect on referral likelihood when the recipient is a
strong tie, though the direction favors rewarding the new
customer (or both). With weak ties, a reward for the recommender
(existing customer) seemed to work best. The
results are consistent with the framework we proposed; that
is, psychological and social benefits and costs are more
important with strong ties, and economic benefits and costs
are more important with weak ties.
Study 4
In Studies 1–3, we used student participants who reacted to
a scenario about MP3 players. We designed Study 4 to
examine the generalizability of these findings using adult
consumers’ actual experiences with mobile phone service.
Method
Design. Study 4 was a 3 × 2 × 2 between-subjects factorial
experiment in which we varied reward scheme (No
Reward, Reward Me, Reward Both), brand strength
(weaker, stronger), and tie strength (weak, strong). Participants
were randomly assigned to the experimental conditions,
except in the case of brand strength, which we
measured.
Participants. Two hundred ninety-eight adult consumers
were recruited from executive programs at a South Korean
university. The participants, 61% of whom were men, were
employed and had a median age of 36 at the time of the
study.
Stimuli and manipulations. We chose mobile phone service
because it differs in important ways from MP3 players.
It is a service (not a good), it has several intangible features,
and some of the critical attributes are based on experience
(not search) characteristics. In addition, high consumer
involvement and long market availability are likely to make
WOM important, influential, and prevalent for mobile
phones. Finally, mobile service is familiar and relevant to
our respondents and is not gender specific.
In Study 4, participants responded on the basis of their
own experience with their current mobile phone service. In
Korea, there are three major mobile service providers, and
at the time of the study, the leading brand had a 53% market
share, the longest presence in the market, and a reputation
for high quality and high price. The two follower brands
had 31.5% and 15.5% market shares, respectively, with a
reasonable or low quality and price image. Customers of the
leading brand were assigned to the stronger brand condition,
and customers of the two follower brands were
assigned to the weaker brand condition. We use manipulation
checks to confirm our expectations about brand
perceptions.
To manipulate reward scheme, we included No Reward,
Reward Me, and Reward Both conditions. We excluded the
Reward You condition to simplify the study because it is
much less common in practice. The reward was 60,000
Korean won in free calls (about $50 at the time of the study)
in the Reward Me condition or 30,000 won in free calls
each in the Reward Both condition. We manipulated tie
strength as previously.
Procedure. After answering basic questions about their
mobile phone service, participants responded to the tie
strength manipulation. Next, they were told that their current
mobile phone service provider was beginning a referral
reward program. This part was not presented in the No
Reward condition. Then, they read a mobile phone–related
WOM scenario in which they were asked to imagine discovering
that the individual described previously was considering
a new or changed mobile phone service.
FIGURE 2
Study 3: The Effect of Reward Scheme and Tie
Strength on Referral Likelihood
Referral Reward Programs and Referral Likelihood / 91
Measures. Because participants had real brand experience,
we needed a referral likelihood measure that included
valence and extremity. We used two nine-point items
(“strongly recommend that they not subscribe to the service/
strongly recommend that they subscribe to the service”
and “sure to tell the friend [acquaintance] not to join the
service/sure to tell the friend [acquaintance] to join the service”)
to measure referral likelihood (Blodgett, Hill, and
Tax 1997).
We used the same manipulation checks for tie strength
and perceived value of rewards as in Studies 1 and 3. We
also measured (on seven-point scales) perceived reputation,
quality, and price of the participants’ brands. Again, we
included product involvement and market maven scales as
covariates. In addition, we included three seven-point items
that measured brand satisfaction (“dissatisfied/satisfied,”
“displeased/pleased,” and “unfavorable/favorable”; Crosby
and Stephens 1987).
Results
Manipulation checks. Of the participants, 56% were
customers of the leading brand, and 44% were customers of
one or the other of the followers. Perceived brand reputation,
quality, and price of the stronger brand were significantly
higher than those of the weaker brands (reputation:
5.18 versus 4.79; t(292) = 2.25, p < .05; quality: 5.09 versus
4.35; t(292) = 5.47, p < .01; price: 3.92 versus 3.43;
t(292) = 3.40, p < .01). Mean tie strength ratings also differed
significantly (.88 for strong ties and .66 for weak ties;
t(292) = 11.45, p < .01). Participants perceived the reward
as being of the same value in the two reward conditions
(Reward Me = 5.01, Reward Both = 5.18; t(197) = .62, p >
.50).
Referral likelihood. The two items measuring referral
likelihood were highly correlated (r = .93), and we averaged
them. We analyzed referral likelihood with an ANCOVA.
Participants’ product involvement, market maven, and satisfaction
scores were covariates, and reward scheme, brand
strength, and tie strength were between-subject factors. The
analysis yielded a significant main effect for all three
covariates (F(1, 275) = 10.01 to 14.85, all ps < .01).
Increases in market maven (β = .27), involvement (β = .27),
and satisfaction (β = .18) yielded increases in referral
likelihood.
Main effects for tie strength (F(1, 275) = 16.77, p < .01)
and reward scheme (F(2, 275) = 10.66, p < .01) were significant.
Again, there was greater referral likelihood with
strong ties (6.37) than with weak ties (5.67). In addition,
referral likelihood was greatest when both customers were
rewarded and lowest when neither was (Reward Both =
6.51, Reward Me = 6.12, and No Reward = 5.43). The differences
between No Reward and Reward Me (t(195) =
3.08, p < .01) and between No Reward and Reward Both
(t(195) = 4.37, p < .01) were significant, but the difference
between Reward Me and Reward Both was not.
There was a significant interaction between reward
scheme and tie strength (F(2, 275) = 5.19, p < .01; see Figure
3). In the strong tie condition, referral likelihood for
Reward Both (6.79) was marginally greater than Reward
Me (6.17; t(103) = 1.89, p = .06) and greater than No
Reward (6.13; t(101) = 2.49, p < .05). Reward Me and No
Reward were not different from each other (t(100) = .82,
p = .42). Conversely, with weak ties, referral likelihood was
greater for both Reward Me (6.08) and Reward Both (6.19)
than for No Reward (4.72; t(93) = 3.66, p < .01; t(92) =
3.80, p < .01; respectively). The difference between Reward
Me and Reward Both was not significant.
In addition, the interaction between reward scheme and
brand strength was significant (F(2, 275) = 3.35, p < .05).
For the stronger brand, referral likelihood was greater for
Reward Both (6.90) than for No Reward (5.80; t(107) =
3.90, p < .01). Reward Me (5.90) was not different from No
Reward (t(109) = .83, p = .41), but it was different from
Reward Both (t(114) = 3.32, p < .01). For the weaker
brands, referral likelihood was greater for both Reward Me
(6.41) and Reward Both (6.02) than for No Reward (5.01;
t(84) = 3.55, p < .01; t(86) = 2.37, p < .05; respectively).
The difference between Reward Me and Reward Both was
not significant (t(84) = 1.08, p = .28).
Discussion
Study 4 replicated the moderating effects of both tie
strength and brand strength on referral likelihood that we
FIGURE 3
Study 4 Results
A: The Effect of Reward Scheme and Tie Strength on
Referral Likelihood
B: The Effect of Reward Scheme and Brand Strength on
Referral Likelihood
92 / Journal of Marketing, January 2007
found in Studies 1 and 2. A reward offered to an existing
customer who makes a referral (i.e., Reward Me) increased
referral likelihood to weak ties but not to strong ties (H1).
We also found that referral likelihood was greater when the
reward was offered by a weaker brand (H2). Furthermore,
Study 4 results were similar to those of Study 3 for the
interaction of tie strength with reward scheme. There was
little difference between Reward Both and Reward Me with
weak ties. With strong ties, referral likelihood was greater
with Reward Both than with Reward Me (H3).
In Study 4, we also found an interaction between brand
strength and scheme. For the stronger brand, Reward Both
performed significantly better than Reward Me, whereas for
the weaker brand, Reward Me fared slightly (albeit not significantly)
better than Reward Both. This result has direct
implications for the design of reward programs. Furthermore,
this interaction is similar to the one between tie
strength and scheme, perhaps for similar reasons. It is possible
that for consumers of a stronger brand, Reward Me
induces negative self-perceptions because of their high level
of commitment to the brand and strong intrinsic motivation
for referral. Conversely, Reward Both may reduce such psychological
costs. In addition, the incremental economic gain
of Reward Me over Reward Both is likely to be perceived as
less important by consumers of a stronger brand, because
they tend to be less price sensitive.
General Discussion
Contributions to Theory
This article makes several contributions to the literature on
WOM and exchange theory. First and at the most basic level,
our work broadens the scope of exchange theory from a
focus on exchange between two parties (for a discussion of
types of exchange, see Ekeh 1974) to the examination of a
more complex relationship among an existing customer, a
new customer, and a brand. Furthermore, although exchange
theory has been used to examine a wide range of behaviors,
scholars have rarely approached WOM from this perspective
(an exception is Gatignon and Robertson’s [1986] conceptual
study). We believe that our findings show the utility of
applying exchange theory to unde, rstanding WOM. In the
equity view of exchange, people prefer an equitable or balanced
exchange, and in the presence of an inequitable or
unbalanced exchange, they feel distress and try to remedy
the situation by either reducing their input or increasing their
output to achieve equilibrium (Walster, Berscheid, and Walster
1973; Walster, Walster, and Berscheid 1978). Our finding
that rewards increase people’s referral likelihood supports
the exchange theory explanation. That is, making a
referral without any extrinsic reward may create feelings of
inequity for a customer; the referral is an unreciprocated
favor done for the other consumer and the company. In this
framework, rewards for referral work because they reduce
the level of inequity and create movement toward equilibrium.
The lack of difference in referral likelihood between
smaller and larger rewards can be interpreted in the same
light; that is, the larger reward may have been perceived as
too much compensation for a referral (Austin and Walster
1974; Smith, Bolton, and Wagner 1999).
We expected that equity concerns would dominate in
casual relationships but not in close relationships. In close
relationships, people respond to the other person’s needs,
and we linked our prediction for these relationships to the
concept of strong ties (Clark 1984; Clark, Mills, and Powell
1986; Granovetter 1973). The prediction for close relationships
was supported; the presence or amount of reward did
not affect referral likelihood.
Our research also increases the understanding of the
operation of norms or rules of exchange by examining how
reward allocation influences referral likelihood. Because
the recommender receives the full economic benefit with
Reward Me, partial benefit for Reward Both, and nothing
for Reward You, the level of equity in exchange is in the
same order (i.e., the greatest inequity is with Reward You).
For weak ties, referral likelihood is directly linked to the
amount of equity. For strong ties, in which equity should
not matter, we found that the link between equity and referral
likelihood was broken.
Implications for Managers
In general, our results show that offering a reward increases
referral likelihood but that there was no difference in effect
between smaller and larger rewards. These results suggest
that firms need to calibrate reward size carefully by computing
the revenue impact of the size of a reward on marginal
referral likelihood and then comparing that with the
cost of alternative reward programs.
Studies 1 and 2 found that with strong ties, rewarding
the existing customer (Reward Me) did not increase referral
likelihood. For marketing managers, the results create a
challenge. Referral reward programs tend to target strong
ties explicitly (e.g., cell phone family-and-friends programs)
or end up with strong ties (because people interact
most frequently with strong ties, these ties are more likely
to receive recommendations). A solution may be the allocation
scheme. The results of Studies 3 and 4 suggest that
rewarding either the new customer or both the new and the
existing customer can increase referral likelihood with
strong ties (though not by much). In addition, our study
examined only referral likelihood. Rewards may have other
effects on strong ties; perhaps they serve as a reminder to
talk about the brand, even to friends.
Conversely, rewards are important for increasing referral
likelihood to weak ties. Frenzen and Nakamoto (1993) find
that weak ties are not as effective as strong ties in the flow of
certain types of information, though they play a critical role
in bridging the gap between different groups (Granovetter
1973). Although our results show that information transmission
through weak ties can be improved with incentives, the
managerial issue is how to devise programs that target weak
ties. The first referrals from a customer will probably be
family or close friends for whom the recommendation is
likely to have occurred anyway. It is probably subsequent
referrals, presumably to weaker ties, that need encouragement.
An option would be to increase the reward as the number
of referrals increases. This segmentation approach would
pay the least for referrals that are most likely to occur naturally
(i.e., strong ties) and the most for referrals that are least
likely to occur naturally (i.e., weak ties).
Aaker, Jennifer L. and Angela Y. Lee (2001), “‘I’ Seek Pleasures
and ‘We’ Avoid Pains: The Role of Self-Regulatory Goals in
Information Processing and Persuasion,” Journal of Consumer
Research, 28 (June), 33–49.
Arndt, Johan (1967), “Word of Mouth Advertising and Informal
Communication” in Risk Taking and Information Handling in
Consumer Behavior, Donald F. Cox, ed. Boston: Division of
Research, Graduate School of Business Administration, Harvard
University, 188–239.
Austin, William and Elaine Walster (1974), “Reactions to Confirmations
and Disconfirmations of Expectancies of Equality and
Inequality,” Journal of Personality and Social Psychology, 30
(2), 208–216.
Referral Reward Programs and Referral Likelihood / 93
Furthermore, the results in Studies 3 and 4 show that
with weak ties, the recommender needs to receive something
in return for the recommendation. The reward scheme
could be calibrated to reflect this result by tilting the payoff
toward the recommender for the subsequent referrals. Our
results in Study 4 support this notion; the referral levels
with strong ties drop off as the reward switches from
Reward Both to Reward Me. Thus, tilting the rewards
toward the recommender should naturally switch the
emphasis to weak ties.
In Study 1, we also found that the impact of reward programs
on referral likelihood varied by brand strength. The
increase in referral likelihood with a reward compared with
no reward was much greater for the consumers of the
weaker brand than for the consumers of the stronger brand.
These results imply that referral reward programs may be
particularly important for brands that are perceived as
weaker. Even if the focus is on increasing brand strength in
the long run, in the short run, weaker brands can use a
reward program to increase referrals and, perhaps, product
trial. As we noted in Study 4, our results provide managers
with guidance in the choice of scheme by brand strength;
specifically, they should use Reward Me for weaker brands
and Reward Both for stronger brands.
Limitations and Further Research
We concentrated on the recommender’s referral likelihood,
but for a referral program to be effective, there must be
referrals combined with receiver receptivity. Receivers may
form different perceptions of the recommender and evaluate
the brand differently if a referral arises from a reward program
rather than from naturally occurring WOM (for a
related idea, see Wiener and Mowen 1986). Thus, further
research should more explicitly focus on the dyad, examining
reward program conditions under which both referral
likelihood and receiver receptivity are high. In addition,
because it is possible that referral likelihood does not
always lead to actual referrals (Mittal and Kamakura 2001),
further research should collect behavioral data to complement
and validate our findings.
Kivetz and Simonson (2002) show that as consumers
invest more effort to obtain a reward in a frequency program,
their preference moves more toward luxury rewards.
This result may have implications for ours. If referral
requires relatively high levels of recommender effort, our
use of more utilitarian/functional rewards might have
underestimated the potential effects of referral programs. It
seems worthwhile to consider different types of rewards in
further research.
In general, communicating an opinion makes the communicator’s
attitude more extreme, even when the opinion
is not the communicator’s own (Higgins and Rholes 1978).
In addition, self-perception theory argues that a behavior
induced by or associated with a large extrinsic incentive is
likely to lead to overjustification and thus affect the person’s
attitudes negatively (Bem 1972; Festinger 1957). The
results of this work indicate that though engaging in natural
WOM may reinforce recommenders’ satisfaction with the
brand, making rewarded referrals may undermine it. In
examining rewarded referral, further research should examine
other downstream variables (e.g., brand attitude, satisfaction,
brand loyalty) in the recommender.
Our research designs assumed that consumers learn
about the referral reward program after product purchase.
However, consumers can also learn about such programs
before purchase. If so, the program could affect brand
choice positively if it is perceived as an added value of the
brand. Offering a reward might also activate persuasion
knowledge for some consumers, leading them to make
negative inferences about the firm’s motives and perhaps
leading them to less favorable responses toward the brand
(Freistad and Wright 1994). Further research should examine
the impact of pre- versus postpurchase knowledge of the
reward program.
Finally, the role of cultural differences should be considered.
All our participants were from Asia. Asians are
more likely to have an interdependent self-construal,
whereas Westerners are more likely to be independent
(Aaker and Lee 2001). Because self-construal affects a person’s
assessment of the importance of the self versus the
group, the (presumed) interdependent self of our participants
may have exerted systematic effects on their referral
likelihood in interacting with strong versus weak ties. There
are two possible effects. On average, if Asians show more
strong-tie-oriented (e.g., altruistic) behaviors regardless of
tie strength, they should make little distinction in their
referral behavior between strong and weak ties. As a result,
the significant interactions between tie strength and other
variables that we found would seem to be unlikely. Conversely,
if Asians show more altruistic behavior with strong
ties and more selfish behavior with weak ties than Westerners,
this extremity will inflate the moderating effects of tie
strength for Asians. Only a cross-cultural comparison or a
design incorporating a manipulation of self-construal can
examine the impact of self-construal. Because our tie
strength findings are consistent with previous studies that
use U.S. participants (e.g., Frenzen and Nakamoto 1993),
we anticipate that our effects would persist with a Western
sample.
REFERENCES
Beach, Lee Roy and William B. Carter (1976), “Appropriate and
Equitable Repayment of Social Debts,” Organizational Behavior
and Human Performance, 16 (2), 280–93.
Bem, Daryl (1965), “An Experimental Analysis of Self-
Persuasion,” Journal of Experimental Social Psychology, 1
(August), 199–218.
⎯⎯⎯ (1972), “Self-Perception Theory,” in Advances in Experimental
Social Psychology, Vol. 6, Leonard Berkowitz, ed. New
York: Academic Press, 1–62.
Biyalogorsky, Eyal, Eitan Gerstner, and Barak Libai (2000), “Customer
Referral Management: Optimal Reward Programs,”
Marketing Science, 20 (Winter), 82–95.
Blattberg, Robert C. and Kenneth Wisniewski (1989), “Price-
Induced Patterns of Competition,” Marketing Science, 8 (Fall),
291–309.
Blodgett, Jeffrey G., Donna J. Hill, and Stephen S. Tax (1997),
“The Effects of Distributive, Procedural, and Interactional Justice
on Postcomplaint Behavior,” Journal of Retailing, 73
(Summer), 185–210.
Bolton, Ruth N., P.K. Kannan, and Mathew D. Bramlett (2000),
“Implication of Loyalty Program Membership and Service
Experiences for Customer Retention and Value,” Journal of the
Academy of Marketing Science, 28 (Winter), 95–108.
Brown, Jacqueline Johnson and Peter H. Reingen (1987), “Social
Ties and Word-of-Mouth Referral Behavior,” Journal of Consumer
Research, 14 (December), 350–62.
Chen, Yuxin and Mengze Shi (2001), “The Design and Implications
of Customer Recommendation Programs,” working
paper, Stern School of Business, New York University.
Clark, Margaret S. (1984), “Record Keeping in Two Types of
Relationships,” Journal of Personality and Social Psychology,
47 (September), 549–57.
⎯⎯⎯, Judson Mills, and Martha C. Powell (1986), “Keeping
Track of Needs in Communal and Exchange Relationships,”
Journal of Personality and Social Psychology, 51 (August),
333–38.
Crosby, Lawrence A. and Nancy Stephens (1987), “Effects of
Relationship Marketing on Satisfaction, Retention, and Prices
in the Life Insurance Industry,” Journal of Marketing Research,
24 (November), 401–411.
Dichter, Ernest (1966), “How Word-of-Mouth Advertising Works,”
Harvard Business Review, 44 (November–December), 147–66.
Ekeh, Peter P. (1974), Social Exchange Theory: The Two Traditions.
Cambridge, MA: Harvard University Press.
Feick, Lawrence and Robin A. Higie (1992), “The Effects of Preference
Heterogeneity and Source Characteristics on Ad Processing
and Judgments About Endorsers,” Journal of Advertising,
21 (June), 9–24.
⎯⎯⎯ and Linda Price (1987), “The Market Maven: A Diffusion
of Market Place Information,” Journal of Marketing, 51 (January),
83–97.
Festinger, Leon (1957), A Theory of Cognitive Dissonance. Stanford,
CA: Stanford University Press.
Folkes, Valerie S. (1984), “Consumer Reactions to Product Failure:
An Attributional Approach,” Journal of Consumer
Research, 10 (March), 398–409.
Freistad, Marian and Peter Wright (1994), “The Persuasion
Knowledge Model: How People Cope with Persuasion
Attempts,” Journal of Consumer Research, 21 (June), 1–31.
Frenzen, K. Jonathan and Harry L. Davis (1990), “Purchasing
Behavior in Embedded Markets,” Journal of Consumer
Research, 17 (June), 1–12.
⎯⎯⎯ and Kent Nakamoto (1993), “Structure, Cooperation, and
the Flow of Market Information,” Journal of Consumer
Research, 20 (December), 360–75.
Gatignon, Hubert and Thomas S. Robertson (1986), “An Exchange
Theory Model of Interpersonal Communication,” in Advances
in Consumer Research, Vol. 13, Richard J. Lutz, ed. Provo, UT:
Association for Consumer Research, 534–38.
94 / Journal of Marketing, January 2007
Granovetter, Mark S. (1973), “The Strength of Weak Ties,” American
Journal of Sociology, 78 (May), 1360–80.
Heath, Timothy B., Gangseog Ryu, Subimal Chatterjee, Michael
S. McCarthy, David L. Mothersbaugh, Sandra Milberg, and
Gary J. Gaeth (2000), “Asymmetric Competition in Choice and
the Leveraging of Competitive Disadvantages,” Journal of
Consumer Research, 27 (December), 291–308.
Higgins, E. Tory and William S. Rholes (1978), “Saying Is Believing:
Effects of Message Modification on Memory and Liking
for the Person Described,” Journal of Experimental Social Psychology,
14 (July), 363–78.
Hogan, John, Katherine N. Lemon, and Barak Libai (2003), “What
Is the True Value of a Lost Customer?” Journal of Service
Research, 5 (February), 196–208.
Kamakura, Wagner A. and Gary J. Russell (1989), “A Probabilistic
Choice Model for Market Segmentation and Elasticity Structures,”
Journal of Marketing Research, 35 (February), 16–29.
Keller, Kevin Lane (1993), “Conceptualizing, Measuring, and
Managing Customer-Based Brand Equity,” Journal of Marketing,
57 (January), 1–22.
Kelly, Harold H. (1979), Close Relationships: Their Structures
and Processes. Hillsdale, NJ: Lawrence Erlbaum Associates.
⎯⎯⎯ and John W. Thibaut (1978), Interpersonal Relations: A
Theory of Interdependence. New York: John Wiley & Sons.
Kivetz, Ran and Itamar Simonson (2002), “Earning the Right to
Indulge: Effort as a Determinant of Customer Preferences
Toward Frequency Program Reward,” Journal of Marketing
Research, 39 (May), 155–70.
Lichtenstein, Donald R., Peter H. Bloch, and William C. Black
(1988), “Correlates of Price Acceptability,” Journal of Consumer
Research, 15 (September), 59–74.
Mittal, Vikas and Wagner Kamakura (2001), “Satisfaction, Repurchase
Intent, and Repurchase Behavior: Investigating the Moderating
Effect of Customer Characteristics,” Journal of Marketing
Research, 38 (February), 131–42.
Mummert, Hallie (2000), “The Year’s Best Bells & Whistles,” Target
Marketing, 23 (11), 3–5.
Reingen, Peter H. and Jerome B. Kernan (1986), “Analysis of
Referral Networks in Marketing: Methods and Illustration,”
Journal of Marketing Research, 13 (November), 449–54.
Rosen, Emanuel (2000), The Anatomy of Buzz: How to Create
Word of Mouth Marketing. New York: Doubleday.
Rust, Roland T., Katherine N. Lemon, and Valarie A. Zeithaml
(2004), “Return on Marketing: Using Customer Equity to
Focus Marketing Strategy,” Journal of Marketing, 68 (January),
109–127.
Simonson, Itamar (1992), “The Influence of Anticipating Regret
and Responsibility on Purchase Decisions,” Journal of Consumer
Research, 19 (June), 105–118.
Smith, Amy K., Ruth N. Bolton, and Janet Wagner (1999), “A
Model of Customer Satisfaction with Service Encounters
Involving Failure and Recovery,” Journal of Marketing
Research, 36 (August), 356–72.
Törnblom, Kjell Y. and Björn O. Nilsson (1993), “The Effects of
Matching Resources to Source on Their Perceived Importance
and Sufficiency,” in Resource Theory: Explorations and Applications,
Uriel G. Foa, John Converse Jr., Kjell Y. Törnblom,
and Edna B. Foa, eds. San Diego: Academic Press, 81–96.
Walster, Elaine, Ellen Berscheid, and G. William Walster (1973),
“New Directions in Equity Research,” Journal of Personality
and Social Psychology, 25 (2), 151–76.
⎯⎯⎯, G. William Walster, and Ellen Berscheid (1978), Equity:
Theory and Research. Boston: Allyn and Bacon.
Wiener, Joshua L. and John C. Mowen (1986), “Source Credibility:
On the Independent Effects of Trust and Expertise,” in
Advances in Consumer Research, Vol. 13, Richard J. Lutz, ed.
Provo, UT: Association for Consumer Research, 306–310.
