Escolar Documentos
Profissional Documentos
Cultura Documentos
ABSTRACT
INTRODUCTION
Many studies have tried to identify factors that predict such postcon-
sumption behaviors as switching brands or loyalty, complaint behaviors,
and negative and positive word-of-mouth communications (e.g.,
Andreasen, 1985; Athnassopoulos, Gounaris, & Stathakopoulos, 2001;
Caruana, 2002; Day & Ash, 1978; Day, 1983; Gronhaug & Zaltman, 1977;
Gustafsson, Johnson, & Ross, 2005; Krishnan & Valle, 1979; Mittal &
Kamakura, 2001; Sigh, 1990; Stepens & Gwinner, 1998; Yi & La, 2004).
These behaviors are principally viewed as consequences of customer
satisfaction and dissatisfaction (Anderson & Fornell, 1994; Yi, 1989).
Nevertheless, customer dissatisfaction may not be the only determi-
nant of negative postconsumption behaviors (Day, 1983), and may lead
to different negative postpurchase responses (Bougie, Pieters, &
Zeelenberg, 2003). In particular, with respect to the latter point, it is
not clear under what conditions consumer dissatisfaction produces
complaints rather than switching brands or negative word of mouth
(Nyer, 1997).
In addition, satisfaction may not be sufficient for determining positive
postconsumption behaviors. For example, satisfaction may fail to prevent
switching behavior if, as sometimes happens, one is satisfied with brand
A but brand B is perceived to be even better. Most studies have been pre-
occupied primarily with predicting the different consequences of customer
dissatisfaction. Relatively little work has been done, however, to under-
stand under what conditions satisfaction/dissatisfaction determines
different behaviors (Bougie, Pieters, & Zeelenberg, 2003; Nyer, 1997).
Knowing why customers select a specific postconsumption behavior
is important for marketers. In fact, rather than seeking redress, some dis-
satisfied consumers choose to not repurchase or to engage in negative
word of mouth. Exit and negative word of mouth are particularly dan-
gerous responses because they are “invisible” (Richins, 1987), and there-
fore the firm loses the opportunity to remedy the problem and retain
customers (Hirschmann, 1970; Lapidus & Pinkerton, 1995): in this way,
the firm loses sales and profits. It is important for manufacturers and
service providers to understand why some dissatisfied customers “exit”
or engage in negative word of mouth behavior, whereas others give the
firm a chance to remedy the problem. Moreover, negative word of mouth
damages the firm’s reputation and, in this way a firm loses potential new
customers. By contrast, Fornell and Wernerfelt (1987) maintain that cus-
tomer loyalty can be increased by encouraging dissatisfied consumers
to complain. Marketers need a theory that explains why customers are
likely to select one postpurchase behavior or another, and the purpose of
this study is thus to develop and test a theory to predict some of the
aforementioned postconsumption behaviors.
Actually, it might be expected that satisfaction by itself will not dif-
ferentiate between different postconsumption behaviors. It seems that
there are variables that interfere or interact with the relationship between
customer satisfaction/dissatisfaction (CS/D) and postpurchase behav-
iors and are capable of producing different positive and negative actions.
According to the literature, the variables that influence the relationship
between CS/D and postconsumption behaviors are CS/D attributions
(Krishnan & Valle, 1979), industry characteristics (Andreasen, 1985;
Hirschmann, 1970; Singh, 1990; Stephens & Gwinner, 1998), the per-
ceived probability of successful complaining (Andreasen, 1985; Blodgett,
872 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 873
Wakefield, & Barnes 1995; Gelb & Johnson, 1995; Hirschman, 1970; Singh,
1990; Singh & Wilkes, 1996; Stephens & Guinner, 1998), attitude toward
complaining (Blodgett, Wakefield, & Barnes, 1995; Keng, Richmond, &
Han, 1995; Stephens & Guinner, 1998), product importance (Blodgett,
Wakefield, & Barnes, 1995), and consumer sophistication (Gelb & Johnson,
1995; Gronhaug & Zaltman, 1977; Hirschman, 1970; Singh, 1990).
Nyer (1997) suggests a framework that predicts nearly all the responses
mentioned and that can potentially integrate all the antecedents of post-
consumption behaviors (particularly intervening variables). He demon-
strates that different emotions can directly determine different types of
behaviors. Moreover, he shows that these emotions are elicited by differ-
ent appraisals. These appraisals refer to some of the variables mentioned
earlier. In particular, Nyer demonstrates that consumers’ expectations of
attribute performance, product importance, and potential to cope (per-
ceived probability of successful complaining) influence such distinct emo-
tions as joy, anger, sadness, and satisfaction. He further showed that the
effects of these three appraisals on positive and negative word of mouth
are totally mediated by the emotions. Bougie, Pieters, and Zeelenberg
(2003) adopt a similar framework and demonstrate that consumers’ expec-
tations of attribute performance and CS/D attribution elicit the emotions
anger and dissatisfaction and that these two emotions determine differ-
ent postconsumption behaviors. In particular, anger fully mediated the
effect of dissatisfaction on negative word of mouth, and complaint partially
mediated the effect of dissatisfaction on switching.
The most important contribution of these two studies for the purposes
of this research is that, in their model, attribution, product importance,
and coping potential have been added to the consumers’ expectations of
attribute performance. A number of specific appraisals thus interact to
form particular emotions that in turn determine distinct postconsump-
tion behaviors. The rich body of research conducted on the antecedents
of postconsumption behaviors presented earlier can be incorporated into
a general framework. Moreover, the presented model is able to predict
other behaviors (not only negative and positive word of mouth, and com-
plaint, as developed later in this paper).
Thus, this study begins with Nyer’s and Bougie, Pieters, and
Zeelenberg’s models and tries to extend them in two directions:
HYPOTHESES: EMOTIONS
874 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 875
876 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 877
878 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 879
Goal congruence/incongruence
Incongruence Congruence
Anger Gratitude
Table 1 summarizes the first two hypotheses. Here the specific two-way
interactions between agency and goal congruence/incongruence can be
seen, with happiness, sadness, guilt, anger, pride and gratitude predicted,
depending on the specific condition at hand. Scenarios are used to pro-
duce the interactions, as described later in this paper under Research
Methodology.
H3a: When goal incongruence occurs, the likelihood that anger will
lead to complaining will be greater than the likelihood that sad-
ness will lead to complaining.
H3b: When goal incongruence occurs, the greater the anger, the greater
will be the complaining.
880 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 881
H5a: Gratitude, but not happiness and pride, will influence repurchase
intention and positive word of mouth in goal-congruent contexts.
H5b: The greater the gratitude, the greater the repurchase intention
and positive word of mouth under goal-congruence.
RESEARCH METHODOLOGY
RESULTS
882 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 883
Question 1
How intensely was Helen feeling each of the following emotions at the end of the story? (circle your answer in each row)
Not at all Mode rately Very
intensely
Remorse 0 1 2 3 4 5 6 7 8 9 10
Dissatisfaction 0 1 2 3 4 5 6 7 8 9 10
Gladness 0 1 2 3 4 5 6 7 8 9 10
Dislike 0 1 2 3 4 5 6 7 8 9 10
Satisfaction 0 1 2 3 4 5 6 7 8 9 10
Sorrow 0 1 2 3 4 5 6 7 8 9 10
Guilt 0 1 2 3 4 5 6 7 8 9 10
Annoyance 0 1 2 3 4 5 6 7 8 9 10
Happiness 0 1 2 3 4 5 6 7 8 9 10
Pride 0 1 2 3 4 5 6 7 8 9 10
Gratitude 0 1 2 3 4 5 6 7 8 9 10
Question 2
Now please indicate how likely Helen will perform each of the following (circle your answer in each row).
belong to the same factor and to show a negative correlation with the
other items. The output was a three-factor solution. The three-factor
solution accounted for 87% of the total variance and each item loaded
highly on its hypothesized factor and relatively low on the nonhypothe-
sized factors.
Sorrow .81
Dissatisfaction .82
Gladness .97
Happiness .96
Satisfaction .98
Remorse .88
Guilt .83
Annoyance .74
Dislike .79
Hypothesis 1
In order to test the first hypothesis, a one-way ANOVA was run with
goal congruence/incongruence as a single factor and happiness as the
dependent variable. The results revealed a significant effect of goal con-
gruence/incongruence (F(1,179) 1413.46, p .001) which shows higher
happiness for goal congruence (M 8.89) than goal incongruence
(M 0.70). Also sadness exhibited the presence of a significant goal con-
gruence/incongruence (F(1,178) 389.34, p .001) effect: the degree of
sadness is higher under condition of incongruence (M 6.72) than under
condition of congruence (M .79).
Hypothesis 2
An ANOVA conducted on gratitude confirmed a significant goal congru-
ence/incongruence X agency interaction: F(2, 176) 20.84, p .001. A
planned contrast showed that gratitude experienced in the case of the goal
congruent seller-outcome (M 6.67) was higher than the average
gratitude experienced in the other cases (M 1.21): t(180) 10.73,
p .001.1
1
Furthermore, the two planned contrasts revealed that, under condition of goal congruence, grat-
itude experienced in the case of the seller caused outcome (M 6.67) was higher than gratitude
experienced in the case of the self-caused outcome (M 3.27) or circumstance caused
outcome (M 1.83): t(77) 4.15 and t(77) 5.64, p .001.
884 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 885
Hypotheses 3 and 4
To investigate hypotheses 3 and 4, the correlations between the six emo-
tions and the two negative postconsumption behaviors were analyzed:
complaining and negative word of mouth. These correlations are shown
in Table 3.
2
Under Goal Incongruence, the experienced guilt in the situation of self-caused outcome (M
8.10) was significantly higher than the guilt felt in the situation of seller-caused outcome
(M 3.63) and circumstance-caused outcome (M 1.85), t(99) 7.18 and t(99) 8.76, p .001,
respectively.
3
However, under goal congruence, there was no significant difference in the level of pride expe-
rienced for the cases of self-caused outcome (M 9.48) and seller-caused outcome (M 9.19),
t(77) .54, p .59.
4
However, under goal incongruence, there was no significant difference in the level of anger expe-
rienced for the cases of seller-caused outcome (M 3.88) and self-caused outcome (M 2.92),
t(97) 1.60, p .11. Moreover, anger did not differ significantly for seller-caused and circum-
stance-caused outcomes (M 3.87 vs. M 3.29), t(97) .90, p .37.
Hypothesis 5
To study the relations between emotions and the two positive postcon-
sumption behaviors, the correlations between these variables were ana-
lyzed (see Table 5). As predicted, the correlation between gratitude and
886 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 887
Step 1 Step 2
Dependent variables: Dependent variables:
gratitude, repurchase repurchase intention and
intention and positive positive word of
word of mouth mouth Covariates:
Effect Covariates: none gratitude
Step 1 Step 2
Dependent variables: Dependent variables:
guilt, complaining and complaining and negative
negative word of mouth word of mouth
Effect Covariates: none Covariates: guilt
DISCUSSION
888 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 889
By contrast, under H2, the findings show that the interaction between
goal congruence/incongruence and agency appraisals elicits two key emo-
tions: gratitude and guilt, which have important roles in predicting dif-
ferent behaviors. In particular, gratitude leads to positive word of mouth
and repurchase intentions. Guilt, when consumers accept their respon-
sibility for prior actions and the negative outcome, inhibits negative word
of mouth and complaining behaviors.
The predicted attributions of other-directed anger did not influence
behavior outcomes. This result was probably because of the choice that
was made in the type of service for the experiment. In fact, as sports
enthusiasts know very well, it is very difficult to get physically fit when
one is less than diligent in exercising: thus, a sense of self-anger some-
times follows bad fitness results (for research on self-anger, see Ben-Ze’ev,
2000). In future research, the two forms of anger (self vs. other directed)
should be distinguished more clearly, especially with reference to their
different effects on postconsumption behaviors.
The findings show that particular combinations of antecedents deter-
mine specific emotions, that these emotions predict different postcon-
sumption behaviors, and that these emotions partially mediate the effects
of the manipulated appraisals on postconsumptions behaviors. More pre-
cisely, the results of this study suggest that attributions influence how
consumers feel in different cases of positive and negative outcomes, and
these emotions imply different consumer behavioral responses to service
success or failure.
With respect to service failures, this research suggests that, in this
case, the consumer asks who, if anyone, is to be blamed. When the con-
sumer feels responsible for the negative outcome, a sense of guilt lessens
the chance that such actions as negative word of mouth will be directed
at and damage the service provider. Unfortunately, from the point of view
of firms, consumers sometimes do not perform a careful attributional
appraisal: although consumers may feel confident about their inferences,
perceived reasons may differ from the “true” reasons of service/product
failure. Perhaps such possibilities led British Airways in 2001 to pub-
lish new rules for passengers asserting that the company can bar them
from boarding its aircraft if they appear abusive, insulting, threatening,
or disorderly in any way (O’Shaughnessy & O’Shaughnessy, 2001).
Managers of British Airways desire to prevent damaging or incorrect
attributions: they emphasize that there are particular situations in which
the service can be denied to consumers who act inappropriately. Managers
can facilitate consumers’ causal attributions in number of ways. For
example, as in the case of British Airways, they can use the communi-
cation mix to clarify the roles that consumers should play in the con-
sumption delivery and administration of a particular product or service.
If they want to be effective, marketers should communicate the expected
role that consumers should play in a very clear, precise, respectful,
and honest way, without relieving the company of its responsibilities
and prerogatives. But this apparently was not the case for British
Airways. According to O’Shaughnessy and O’Shaughnessy (2001), the
rules that British Airways promulgated were vague and ambiguous, with
too much left to consumer discretion: “there is the need to explicate the
relevant behavior through thick description and videos illustrating
the banned bahavior” (O’Shaughnessy & O’Shaughnessy, 2001, p. 19).
In any case, it appears that the managers of British Airways are correct
when they conclude that consumers do not always form correct attribu-
tions of negative outcomes. Sometimes uncertainty occurs, such as when
a product is complex (Folkes, 1984). Therefore, companies should do their
best to facilitate consumer attributions, and they have different oppor-
tunities to get consumers to recognize and accept their responsibilities, if
it happens that they have real responsibilities in producing negative out-
comes. Companies can do this, for example, by communicating to con-
sumers to take specific precautions in using a particular product. For
example in Italy, in a coadvertising campaign, a famous fashion company
recommends the use of a specific detergent to protect clothes from fading.
In general, advertising and other communication should convey frank
messages, messages that do not minimize the commitment that consumers
are expected to make to obtain benefits from a service or a product. Other
opportunities that marketing managers should take is to discuss with
their consumers the reasons for negative outcomes. This can be done in
response to letters from consumers requesting product information, as
well as in situations where consumers return products for refunds (Folkes,
1984): in these cases, manufacturers or service providers have the chance
to inform their customers about possible product misuse.
Also with respect to successes of products/services, the results of this
study suggest that consumers should engage in attributional informa-
tional processing search and that they will feel gratitude when they
attribute positive outcomes to service providers. The findings show that
this sense of gratitude elicits positive word of mouth and repurchase
intentions. Thus, consumers seem to reward high-effort firms. Marketers
should try to provide situations wherein customers feel that they are
the target of extra effort by sellers: this occurs, for example, in the case
in which a hotel sends birthday presents to its customers, following their
stay, or when a health club member finds a personal trainer particularly
helpful and inspirational. The findings suggest that happiness may lead
to success for a business when this emotion is connected explicitly to a
sense of gratitude.
Several avenues for future research emerge from this study. First, a lim-
itation of this work is that the experiment was run with reference to
only one kind of service: other services should be tested to obtain more
robust results and their generalizability. Second, the interactions among
only two appraisals were studied in order to determine postconsump-
tion emotions: a more comprehensive study should analyze the role of
other possible determinants of consumption emotions, such as “industry
890 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 891
REFERENCES
892 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar
mar2410_907_20188.qxp 8/14/07 12:45 PM Page 893
The author thanks Richard P. Bagozzi for his insightful comments on a previ-
ous version of this article.
894 SOSCIA
Psychology & Marketing DOI: 10.1002 /mar