ArticlePDF Available

Abstract and Figures

The Service Recovery Paradox (SRP) has emerged as an important effect in the marketing literature. However, empirical research testing the SRP has produced mixed results, with only some studies supporting this paradox. Because of these inconsistencies, a meta-analysis was conducted to integrate the studies dealing with the SRP and to test whether studies' characteristics influence the results. The analyses show that the cumulative mean effect of the SRP is significant and positive on satisfaction, supporting the SRP, but nonsignificant on repurchase intentions, word-of-mouth, and corporate image, suggesting that there is no effect of the SRP on these variables. Additional analyses of moderator variables find that design (cross-sectional versus longitudinal), subject (student versus nonstudent), and service category (hotel, restaurant, and others) influence the effect of SRP on satisfaction. Finally, implications for managers and directions for future research are presented.
Content may be subject to copyright.
http://jsr.sagepub.com
Journal of Service Research
DOI: 10.1177/1094670507303012
2007; 10; 60 Journal of Service Research
Celso Augusto de Matos, Jorge Luiz Henrique and Carlos Alberto Vargas Rossi
Service Recovery Paradox: A Meta-Analysis
http://jsr.sagepub.com/cgi/content/abstract/10/1/60
The online version of this article can be found at:
Published by:
http://www.sagepublications.com
On behalf of:
Center for Excellence in Service, University of Maryland
can be found at:Journal of Service Research Additional services and information for
http://jsr.sagepub.com/cgi/alerts Email Alerts:
http://jsr.sagepub.com/subscriptions Subscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://jsr.sagepub.com/cgi/content/refs/10/1/60 Citations
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
The Service Recovery Paradox (SRP) has emerged as an
important effect in the marketing literature. However,
empirical research testing the SRP has produced mixed
results, with only some studies supporting this paradox.
Because of these inconsistencies, a meta-analysis was
conducted to integrate the studies dealing with the SRP
and to test whether studies’ characteristics influence the
results. The analyses show that the cumulative mean
effect of the SRP is significant and positive on satisfac-
tion, supporting the SRP, but nonsignificant on repur-
chase intentions, word-of-mouth, and corporate image,
suggesting that there is no effect of the SRP on these vari-
ables. Additional analyses of moderator variables find
that design (cross-sectional versus longitudinal), subject
(student versus nonstudent), and service category (hotel,
restaurant, and others) influence the effect of SRP on sat-
isfaction. Finally, implications for managers and direc-
tions for future research are presented.
Keywords: service recovery; Service Recovery Paradox;
meta-analysis; moderation analysis
The Service Recovery Paradox (SRP) is a peculiar
effect in the services marketing literature and has been
conceptually defined as a situation in which a customer’s
postfailure satisfaction exceeds prefailure satisfaction
(McCollough and Bharadwaj 1992). When reviewing this
literature, the theoretical paper by Hart, Heskett, and
Sasser (1990) is one of those frequently cited, especially
the statement that “a good recovery can turn angry, frus-
trated customers into loyal ones. It can, in fact, create more
goodwill than if things had gone smoothly in the first
place” (p. 148). In this way, recovery encounters would
mean an opportunity for service providers to increase
customer retention (Hart, Heskett, and Sasser 1990).
The topic of SRP has been of great importance for
managers and researchers. Given that failure is one of the
main reasons that drive customer-switching behavior,
understanding recovery is relevant because a successful
recovery may lead to customer retention, which will
affect company profitability (McCollough, Berry, and
Yadav 2000). On the other hand, there has always been a
question in the service literature as to whether high
recovery efforts can really create greater satisfaction
The authors are thankful for the support provided by the Brazilian Funding Council for Research (CNPq and CAPES) and the
Graduate School of Management. The authors also would like to thank various authors who sent their recent articles on the topic of
service recovery, including Chihyung Ok, David A. Cranage, Stefan Michel, Steven H. Seggie, and Vincent P. Magnini. The authors
are also grateful to Professor Frank L. Schmidt and Professor David B. Wilson for their support in answering questions about meth-
ods of meta-analysis and to the editor and three anonymous reviewers of JSR for their insightful comments.
Journal of Service Research, Volume 10, No. 1, August 2007 60-77
DOI: 10.1177/1094670507303012
© 2007 Sage Publications
Service Recovery Paradox:
A Meta-Analysis
Celso Augusto de Matos
Jorge Luiz Henrique
Carlos Alberto Vargas Rossi
School of Management, Federal University of Rio Grande do Sul (PPGA-EA-UFRGS)
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
when compared to the situation of no failure (Etzel and
Silverman 1981).
However, empirical studies investigating the SRP have
produced results that vary considerably in terms of statis-
tical significance, direction, and magnitude. Although
some studies provide support for the SRP (Hocutt,
Bowers, and Donavan 2006; Hocutt and Stone 1998;
Magnini et al. 2007; Maxham and Netemeyer 2002;
McCollough 2000; Michel 2001; Michel and Meuter
2006; Smith and Bolton 1998), others have found no sup-
port (Andreassen 2001; Halstead and Page 1992; Hocutt,
Chakraborty, and Mowen 1997; Maxham 2001;
McCollough, Berry, and Yadav 2000; Ok, Back, and
Shanklin 2006; Zeithaml, Berry, and Parasuraman 1996).
These conflicting results might be a consequence of a
number of factors, from different methodological aspects
in the studies to certain conditions moderating the para-
dox. In this respect, some variables have been proposed
in the service recovery literature as potential moderators
of the paradox, including severity of the failure, prior
failure with the firm, stability of the cause of the failure,
and perceived control (Magnini et al. 2007).
Along with these conflicting results, “there is a con-
siderable body of conjecture and intuition pertaining to
the existence of the service recovery paradox”
(Andreassen 2001, p. 40). Thus, this inconsistency with
regard to the effect of the SRP suggests the need for a
meta-analysis to provide both a systematic review and a
quantitative integration of all the available SRP research.
A meta-analysis can provide insights into these inconsis-
tencies by accumulating effects across studies after
adjusting for the studies’ main artifacts (i.e., measure-
ment and sampling error), identifying measurement and
sample characteristics that affect the support/nonsupport
of the SRP, and also testing the generalizability of the
results (Farley, Lehmann, and Sawyer 1995).
Through meta-analysis we aim to (a) reflect on the
different methodological approaches used to test the SRP,
(b) map the dependent variables that have been consid-
ered when the SRP is tested in the literature, (c) reveal
which of these dependent variables support the SRP, (d)
investigate which methodological differences across the
studies moderate the results for the effects of the SRP,
and (e) identify research questions worthy of future
empirical investigations regarding the SRP.
First, we present a theoretical background about the
SRP to guide the meta-analysis. Second, we discuss the
procedures for building the database, computing, and
integrating the effect sizes. Third, we present a quantita-
tive summary that includes the adjusted cumulative mean
values of the effect of service recovery on dependent
variables and test whether the paradox is supported or
not. Fourth, we conduct a more detailed examination,
using subgroup meta-analysis and hierarchical moderator
analysis (Hunter and Schmidt 2004) to provide insights
regarding studies’ characteristics that might moderate the
effects of the SRP. Finally, we present a discussion with
theoretical and managerial implications, limitations, and
future research directions.
THE SERVICE RECOVERY PARADOX
The SRP is defined as the situation in which postre-
covery satisfaction is greater than that prior to the service
failure when customers receive high recovery perfor-
mance (Maxham 2001; McCollough 1995; McCollough
and Bharadwaj 1992; Smith and Bolton 1998). In this
context, effective service recovery may lead to higher sat-
isfaction compared to the service that was correctly per-
formed the first time, and recovery encounters would
mean an opportunity for service providers to increase
customer retention (Hart, Heskett, and Sasser 1990).
Based on the disconfirmation framework (McCollough,
Berry, and Yadav 2000; Oliver 1997), the paradox is related
to a secondary satisfaction following a service failure in
which customers compare their expectations for recovery to
their perceptions of the service recovery performance. If
there is a positive disconfirmation, that is, if perceptions of
service recovery performance are greater than expectations,
a paradox might emerge (secondary satisfaction becomes
greater than prefailure satisfaction). Otherwise, in the case
of a negative disconfirmation, there is a double negative
effect, as service failure is followed by a flawed recovery
(Bitner, Booms, and Tetreault 1990; McCollough, Berry,
and Yadav 2000; Smith and Bolton 1998).
The paradox can also be justified by the script theory
and the commitment–trust theory for relationship mar-
keting (Magnini et al. 2007). Script theory proposes that
there is a common sequence of acts in a service delivery,
in such a way that employees and customers share simi-
lar beliefs regarding the expected order of events and
their respective roles in the process (Bitner, Booms, and
Mohr 1994). If a service failure occurs, it works as a
deviation from the predicted script and produces an
increased sensitivity in the customer regarding the failure
and the redress process. Because of this, satisfaction with
the recovery process becomes more relevant than satis-
faction with the initial attributes in influencing the final
cumulative satisfaction (Bitner, Booms, and Tetreault
1990; Magnini et al. 2007).
Because an excellent service recovery has a direct
impact on how much consumers trust the firm (Kau and
Loh 2006; Tax, Brown, and Chandrashekaran 1998),
there is also a foundation for the Service Recovery
Paradox in Morgan and Hunt’s (1994) commitment-trust
Matos et al. / SERVICE RECOVERY PARADOX 61
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
62 JOURNAL OF SERVICE RESEARCH / August 2007
theory for relationship marketing (Magnini et al. 2007).
In this view, both service recovery efforts and relation-
ship marketing focus on customer satisfaction, trust, and
commitment. Trust exists when one party has confidence
in another’s reliability and integrity (Moorman, Zaltman,
and Deshpande 1992; Morgan and Hunt 1994). As fail-
ures contribute to create insecurity in the customers and
affect trust in the firm, an effective service recovery can
be an opportunity to make customers feel that the firm is
able and willing to correct the problem. As a result, a fair
conflict resolution may have a positive impact on con-
sumer trust (Achrol 1991).
Although most studies test the Service Recovery
Paradox for satisfaction and repurchase intentions,
1
there
are also studies considering the paradox for word-of-mouth
(Hocutt, Bowers, and Donavan 2006; Kau and Loh 2006;
Maxham 2001; Maxham and Netemeyer 2002; Ok, Back,
and Shanklin 2006), corporate image (Andreassen 2001;
Kwortnik 2006), trust (Kau and Loh 2006), quality
(McCollough 1995), complaint intentions (Hocutt,
Chakraborty, and Mowen 1997), switching intentions, pay-
more intentions, and external response (Zeithaml, Berry,
and Parasuraman 1996). Figure 1 provides an overarching
conceptual framework for our meta-analysis and also syn-
thesizes key insights from previous studies and discussions
about the SRP in the extant literature. If the SRP exists, a
service failure that is followed by a high recovery effort
should produce outcome variables that are higher when
compared to a situation in which no failure occurred. Based
on this framework, we expect the following:
Hypothesis 1a: There is a significant positive SRP
effect for satisfaction.
Hypothesis 1b: There is a significant positive SRP
effect for repurchase intentions.
Hypothesis 1c: There is a significant positive SRP
effect for word-of-mouth.
Hypothesis 1d: There is a significant positive SRP
effect for corporate image.
Conflicting results have been found in the literature,
with some studies supporting the SRP (Hocutt, Bowers,
and Donavan 2006; Hocutt and Stone 1998; Magnini
et al. 2007; Maxham and Netemeyer 2002; McCollough
2000; Michel 2001; Michel and Meuter 2006; Smith and
Bolton 1998) and others not supporting this effect
(Andreassen 2001; Halstead and Page 1992; Hocutt,
Chakraborty, and Mowen 1997; Mattila 1999; Maxham
2001; McCollough, Berry, and Yadav 2000; Ok, Back,
and Shanklin 2006; Zeithaml, Berry, and Parasuraman
1996). Also, there is support for the notion that the SRP
is more likely when service failure causes low harm, indi-
cating that recovery strategies may be more effective
when the failure is perceived by the customers as less
severe (Magnini et al. 2007; Mattila 1999; Smith and
Bolton 1998). These contingencies are discussed later in
the Moderators section.
Another possible explanation for the mixed findings
might be related to the nature of the paradox (Michel and
Meuter 2006). In this view, it is considered that the SRP is
a very rare event (Boshoff 1997), that only a minority of
dissatisfied customers complains (Singh 1990), and that
only few recoveries lead to customer satisfaction (Kelley,
Hoffman, and Davis 1993). As a result, it becomes very
difficult to achieve a large sample of customers who have
received a very satisfactory recovery,
2
and this requirement
may have an influence on nonsignificant results presented
in the literature (Michel and Meuter 2006).
Moderators
The mixed findings can also be caused by certain con-
ditions moderating the paradox. For example, although the
theoretical definition of the SRP seems to be convergent in
the literature, the same is not true for the operationaliza-
tions of the concept. Although some authors use a
between-subjects approach, comparing a recovery group
with a no-failure control group (Hocutt, Bowers, and
Donavan 2006; Kau and Loh 2006; McCollough 1995;
McCollough, Berry, and Yadav 2000; Michel and Meuter
2006; Ok, Back, and Shanklin 2006), others use a within-
subjects approach, comparing different measures from the
same subject before and after a failure and/or recovery
(Magnini et al. 2007; Maxham 2001; Maxham and
Netemeyer 2002; Smith and Bolton 1998). These differ-
ences also extend to the type of research design used in
terms of experiment or survey approach, cross-sectional
or longitudinal measures, student or nonstudent subjects,
FIGURE 1
Meta-Analytic Framework of the SRP
NOTE: dashed line (- - -) indicates path not tested in the meta-analysis.
H
2
H
3
H
1
Failure Recovery
OUTCOMES
Satisfaction
(H1a)
Repurchase intentions
(H1b)
Word-of-mouth
(H1c)
Corporate Image
(H1d)
THEORETICAL MODERATORS
- Failure severity
- Prior failure experience
- Stability attributions
- Company control over the failure
METHODOLOGICAL
MODERATORS
- Method (survey × experiment)
- Design (cross-sectional ×
longitudinal)
- Subject (student × non-student)
- Service category (hotel ×
restaurant × others)
H
2
H
1
H
3
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
Matos et al. / SERVICE RECOVERY PARADOX 63
single or multiple items measuring dependent variables,
scenario- or non-scenario-based research, and the different
manipulated factors (in case of experiments).
Assmus, Farley, and Lehmann (1984) suggested that
four categories of characteristics might help identify sys-
tematic patterns in a meta-analysis: research context,
model specification, measurement methods, and estima-
tion procedure. However, because our unit of analysis
was bivariate correlations, we seek systematic differ-
ences in the study characteristics. This procedure is
common in meta-analysis using correlations (e.g., Pan
and Zinkhan 2006). In our investigation, we examine four
potential moderators: method (survey versus experi-
ment), design (cross-sectional versus longitudinal),
subject (student versus nonstudent), and service category
(hotel versus restaurant versus others).
Based on the methodological differences across the
studies, we propose the following (see Figure 1):
Hypothesis 2a: The SRP effects differ in studies
using survey methods versus those using exper-
imental methods.
Hypothesis 2b: The SRP effects differ in studies
using cross-sectional designs versus those using
longitudinal designs.
Hypothesis 2c: The SRP effects differ in studies
using student subjects versus those using non-
student subjects.
Hypothesis 2d: The SRP effects differ across stud-
ies using different service categories.
Regarding the boundary conditions for the SRP effects,
some theoretical variables have also been proposed as
potential moderators in the service recovery literature,
including severity of the failure, prior failure with the firm,
stability of the cause of the failure, and perceived control
(Magnini et al. 2007). Most of these contingencies, how-
ever, have been tested only in the more recent literature
(with severity of the failure
3
being an exception), which
precluded their empirical assessment as moderators in our
meta-analysis. Nevertheless, they are included in Figure 1
as propositions to be investigated further in future
research. Their rationale for proposing the various theoret-
ical moderators is discussed next.
Studies support the notion that it is harder to recover
from high-magnitude failures (Magnini et al. 2007;
Mattila 1999; McCollough 1995; Smith and Bolton
1998) or that the perceived harm caused by the failure
interacts with the recovery effort to influence customer
satisfaction (McCollough, Berry, and Yadav 2000). It has
been found that the higher the magnitude or severity of
the failure, the lower the overall customer satisfaction
(Mattila 1999; Weun, Beatty, and Jones 2004), just as less
favorable recoveries tend to be more memorable (Kelley,
Hoffman, and Davis 1993). For instance, a recovery
action (e.g., apology or compensation) might increase
customer satisfaction after a delay in waiting in a line.
But what if this delay has caused a serious consequence
for the customer (e.g., he missed his flight after a delay at
the hotel desk)? It is unlikely that a recovery action
would be able to either bring the customer to the original
level of satisfaction or, even more improbable, increase
his satisfaction. Thus, we propose:
Hypothesis 3a: The SRP is more (less) likely to
occur when the customer perceives the failure as
less (more) severe.
Given that customers usually have a history of interac-
tions with the firm, their cumulative satisfaction, as
opposed to a transaction-specific satisfaction, is based on
their evaluations of multiple experiences with the firm over
time (Bolton and Drew 1991). In this way, satisfactory
recoveries may yield paradoxical gains only in the short
run, and customers will likely infer that multiple failures
are because of problems inherent to the firm (Maxham and
Netemeyer 2002). Hence, when a customer experiences a
second failure, he or she is more likely to attribute the
cause of that problem to the firm than when the customer
experienced failure for the first time (Magnini et al. 2007;
Maxham and Netemeyer 2002). Thus, we propose:
Hypothesis 3b: The SRP is more likely to occur
when the customer experiences the failure for
the first time when compared to the situation in
which there has already been a previous service
failure.
Another influencing factor is the stability of the cause of
the failure. Stability attributions refer to customers’ infer-
ences about whether similar failures are likely to occur in the
future, given the customers’ dissatisfaction with a product or
service (Blodgett, Granbois, and Walters 1993; Folkes
1984). When customers experience a service failure, they
ask themselves whether the failure has temporary (i.e.,
unstable) or permanent (i.e., stable) causes, and if they think
that the problem has stable causes (i.e., it is likely to occur
again), then they will try to avoid this service provider in the
future (Folkes 1984). Studies have indicated that consumers
who perceive a service failure as more stable present lower
repatronage intentions (Folkes 1984, 1988). Smith and
Bolton (1998) found similar results. In their findings, if a
customer believed that the unavailability of the requested
food item was because of a consistent omission of the
restaurant, he or she would be less satisfied and less
likely to repatronize this restaurant. Hence, customers are
more likely to forgive failures with unstable (temporary)
causes (Kelley, Hoffman, and Davis 1993; Magnini et al.
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
64 JOURNAL OF SERVICE RESEARCH / August 2007
2007) and to express a situation of recovery paradox in
this context (Magnini et al. 2007). Thus, we expect:
Hypothesis 3c: The SRP is more (less) likely when
customers perceive that the failure is less (more)
likely to repeat in the future.
Finally, attributions related to whether the firm had
much or little control over the occurrence of the failure
also influence the recovery paradox. When customers per-
ceive that the firm had little control over the service failure,
they are more likely to comprehend and forgive the prob-
lem (Maxham and Netemeyer 2002). This is in agreement
with studies showing that the perceived reason for a prod-
uct’s failure is an important predictor of how consumers
react (Folkes 1984). For instance, complainants who
believed that firms were responsible for the failure were
more likely to expect redress (e.g., apologies, refunds).
Also, customers who attribute failures to controllable fac-
tors will probably be more dissatisfied with the failure and
less forgiving in their evaluations. Indeed, it has been
found that an SRP is more (less) likely to occur when the
customer perceives that the firm had little (much) control
over the cause of the failure (Magnini et al. 2007). Based
on these findings, we propose:
Hypothesis 3d: The SRP is more (less) likely when
customers perceive that the firm had little
(much) control over the cause of the failure.
METHOD
Search Process and Sampling Frame
Studies were identified by a computerized biblio-
graphic search. Databases included Blackwell Synergy,
Elsevier Science Direct, Ebsco, Emerald Insight, Infotrac
College, Proquest, Scopus, Thompson Gale, Wilson Web,
and Google Scholar. First, we searched for the terms
“service failure” and “service recovery” in keywords and
abstracts. Then we narrowed our search by “service
recovery paradox” in abstracts, keywords, or full text.
Using this procedure, we found a total of 319 articles
ranging from 1987 to 2006. By searching Proquest and
Google, we could also access 14 dissertations on the
research topic, leading to a total of 333 studies. Of this
total, 42 (12.6%) were theoretical papers and the remain-
ing 291 investigated service failure and/or recovery
empirically. Among these studies, 21 were identified as
testing the SRP empirically and were chosen for the analy-
sis, producing a total of 24 observations (independent
samples) in our data set (see Table 1).
All identified studies were then examined in terms of
the following relevant variables: authors, year, journal,
service category (hotel, restaurant, and others), method
(survey versus experiment), subjects (students versus
nonstudents), number of compared groups, number of
factors manipulated or measured, dependent variables
(satisfaction, repurchase intentions, word-of-mouth,
trust, image, quality, intentions to complain, switching
intentions, pay-more intentions, and external response),
reliabilities for the dependent variables, and effect sizes.
Effect Size Computation
Our meta-analytic procedure followed common guide-
lines for meta-analysis of experimental studies (Lipsey and
Wilson 2001), in which standardized mean differences
(Cohen’s d) are computed first and then converted to cor-
relation coefficients (r). This procedure is the same as that
employed by other meta-analyses in the marketing litera-
ture (e.g., Brown and Stayman 1992; Eisend 2006). We
selected the correlation coefficient, r, as the effect-size
metric because it is easier to interpret and a scale-free mea-
sure. Also, the correlation coefficient is the mostly used
effect size in meta-analyses in the marketing literature
(e.g., DelVecchio, Henard, and Freling 2006; Eisend 2004,
2006; Franke and Park 2006; Janiszewski, Noel, and
Sawyer 2003; Palmatier et al. 2006; Pan and Zinkhan
2006). As we included in our data set both surveys and
experiments, we could integrate them by using r as the
common effect size. Positive (negative) values of the cor-
relation coefficient indicate the presence (absence) of the
SRP. This procedure followed recommendations by Lipsey
and Wilson (2001, pp. 14, 173) for conducting meta-analysis
with group contrasts (in experiments or surveys) and is
based on the following rationale:
1. The SRP refers to the effect that an outcome
variable (e.g., satisfaction) is greater for a cus-
tomer that has experienced a failure and a high
recovery effort when compared to a customer
that has experienced no failure.
2. The recovery group is considered as the treat-
ment group and the no failure group as the con-
trol group.
3. “The contrast between the experimental and
control group on the values of an outcome vari-
able is interpreted as the effect of treatment”
(Lipsey and Wilson 2001, p. 14). Treatment in
our case is the high recovery effort.
4. Effect size (ES) = (mean
Sat experimental group
mean
Sat control group
) / standard deviation.
5. Then, if satisfaction has a significantly higher
mean value in the experimental group (high
recovery) when compared to the control group
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
Matos et al. / SERVICE RECOVERY PARADOX 65
(no failure), the SRP is present and ES is positive.
Otherwise, if satisfaction is higher in the condi-
tion of no failure, then ES is negative and reflects
a situation of an inverse SRP. Finally, if there are
no significant differences between satisfaction in
the conditions of high recovery and no failure,
then ES is close to zero and the SRP is null.
6. In conclusion, a positive effect size reflects a
positive effect of the treatment (high recovery
effort) and support for the SRP.
However, direct calculation of effect size (as present
above in Item 4) is uncommon because rarely is there
enough information. Under those circumstances, effects
sizes are obtained through a range of statistical data (e.g.,
Student’s t, F ratios with one df in the numerator, χ
2
) by
means of the formulas given by Lipsey and Wilson
(2001, p. 198).
All together, using this approach, 45 effect sizes were
available for the purpose of our meta-analysis. As pre-
sented in Table 1, most studies reported multiple effect
sizes, particularly if we consider the first two outcome
variables (satisfaction and repurchase intentions), for
which there are more frequencies of effects (31 of the
total 45). However, there were no significant mean dif-
ferences for satisfaction when comparing studies report-
ing single or multiple effect sizes (t = .383, p < .706).
When considering repurchase intentions, only one study
presented a single effect size (ES = –.52), and for all oth-
ers, there were multiple effect sizes (n = 11; mean ES =
.025). Mean differences in this case were not significant
at the .05 significance level (t = 2.042, p < .068).
For most cases (35 of the 45 effect sizes), articles pre-
sented ANOVA results, including mean values of the out-
come variable (e.g., satisfaction) for the treatment group
TABLE 1
Studies Included in the Meta-Analysis
Effect Sizes for the Relationship of SRP and …
No. Study Sat Rep Wom Ima Tru Qua IntC Swi Pay Ext
1 Hocutt, Bowers, and Donavan (2006) X X
2 Kau and Loh (2006) X X X
3 Ok, Back, and Shanklin (2006) X X X
4 Magnini et al. (2007) X
5 Michel and Meuter (2006) X X
6 Kwortnik (2006) X X
7 Oh (2003)
a
X
8 Maxham and Netemeyer (2002) X X X
9 Maxham (2001), Study 1 X X X
10 Maxham (2001), Study 2 X X X
11 Andreassen (2001) X X
12 Michel (2001) X
13 McCollough, Berry, and Yadav
(2000), Study 1 X
14 McCollough, Berry, and Yadav
(2000), Study 2 X
15 McCollough (2000) X
16 Mattila (1999) X
17 Smith and Bolton (1998), Study 1 X X
18 Smith and Bolton (1998), Study 2 X X
19 Hocutt and Stone (1998) X
20 Boshoff (1997) X
21 Hocutt, Chakraborty, and Mowen
(1997) X X
22 Zeithaml, Berry, and Parasuraman
(1996) X X X X
23 McCollough (1995)
b
XX
24 Halstead and Page (1992) X
Total 19 12 6 2 1 1 1 1 1 1
NOTE: Total number of effect sizes: 45. Sat = satisfaction; Rep = repurchase intentions; Wom = word-of-mouth; Tru = trust; Ima = image; Qua =
quality; IntC = intentions to complain; Swi = switching intentions; Pay = pay more intentions; Ext = external response.
a. Classified as outlier.
b. Experiment 3 was used.
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
66 JOURNAL OF SERVICE RESEARCH / August 2007
(high recovery effort) and for the control group (no fail-
ure), sample size for each one, and the t or F statistic.
This was enough for computing the standardized differ-
ence and transforming it into r. In some cases (3 of 45),
nonparametric tests were presented (e.g., chi-square) and
the appropriate formula was used for conversion. In the
remaining cases (7 of 45), authors provided either com-
plete information for the direct calculation (mean, stan-
dard deviation, and sample size of each group) or F value
and group sizes (without group means).
Effect Size Integration
Effect size integration (mean, significance, and confi-
dence intervals) was performed following common guide-
lines (Lipsey and Wilson 2001). Because the true
relationship between variables is mainly influenced by
sampling and measurement error, correlations were first
weighted by the inverse variance and then by the inverse
variance corrected for measurement error (cf. Lipsey and
Wilson 2001, p. 110).
4
Thus, our effect size integration is
presented in three stages, based first on observed correla-
tions, then on correlations corrected for sampling error,
and finally on correlations corrected for both sampling and
measurement error. A confidence interval is presented for
each effect size, and it is significant when it does not
include zero. Significance for the mean effect size can also
be tested by z statistic (p < .05 if z > 1.96). When the mean
effect size is significant, a fail-safe N (also known as file
drawer N) is calculated, estimating the number of non-
significant and unavailable studies that would be necessary
to bring the cumulated effect size to a nonsignificant value
(known as the “file drawer problem”; Rosenthal 1979).
This statistic is an indication of how robust results are.
Table 2 presents a summary of this information.
Homogeneity of the effect size distribution was tested
by the Q statistic, which is distributed as a chi-square
(Hedges and Olkin 1985).
5
If the null hypothesis of
homogeneity is rejected, it indicates that the variability in
effect sizes is larger than it would be expected from sam-
pling error, or in other words, differences in effect sizes
may be attributed to factors other than sampling error
alone, maybe moderating variables related to studies
characteristics (Lipsey and Wilson 2001, p. 115).
For each relationship in which homogeneity of effect
size was rejected (i.e., heterogeneity evidence), an analy-
sis of moderating effects was performed, considering the
study characteristics that were coded based on informa-
tion provided in the articles. These variables included
service category (hotels, restaurants, and others), method
(survey versus experiment), design (cross-sectional ver-
sus longitudinal), subjects (students versus nonstudents),
scenario (used or not), sample size (total, control group,
and treatment group),
6
number of items used to measure
the dependent variable,
7
and reliability of the dependent
variable. These data were available for most studies.
When information about the number of subjects in each
experimental group was not available, we used the mean
group size by dividing the total sample size by the number
of groups in the experiment (in 6 of the 24 observations).
In some studies, authors mentioned in tables’ notes the
sizes of minimum and maximum groups. In these cases,
we used the minimum as the sample size for both the treat-
ment and the control group (in 7 of the 24 observations).
When studies measured dependent variables using sin-
gle items
8
or when reliability values were unavailable,
9
these reliabilities were estimated using the Spearman-
Brown procedure suggested by Hunter and Schmidt (2004,
p. 311), a common approach in meta-analyses in market-
ing (e.g., Grewal et al. 1997). Among the 19 studies
TABLE 2
Descriptive Statistics for Effect Size Integration
Sample-Weighted Sample-Weighted Q Statistic for File
Dependent Simple Adjusted Reliability Homogeneity Drawer
Variable k
a
O
b
N
c
Min. Max. Average r Average r Adjusted r Sig. LCI UCI Test Sig. N
Sat 15 18 7,502 –.450 .595 .031 .154 .125 .017 .032 .217 45.992 .000 27
Rep 10 12 7,788 –.520 .435 –.020 –.061 –.072 .068 –.143 –.002 32.965 .001 5
Wom 5 6 672 –.292 .152 –.120 –.078 –.080 .330 –.227 .066 5.875 .319 NC
Image 2 2 762 –.084 .435 .175 .020 –.084 .556 –.282 .113 .000 1.000 NC
NOTE: LCI = lower confidence interval; UCI = upper confidence interval; Sig. = significance; Min. = minimum; Max. = maximum; Sat = satisfaction;
Rep = repurchase intentions; Wom = word-of-mouth; Image = corporate image. Fail-safe number attenuated at .05. NC means the file drawer N was
not calculated because the 95% confidence interval contained zero (effect size with significance p > .05).
a. Number of studies.
b. Number of observations.
c. Combined N over all independent samples.
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
providing effect size for satisfaction, 5 were based on sin-
gle items and had an average reliability estimated as .701.
In the same way, among the 12 studies providing effect
size for repurchase intentions, 5 studies did not provide
information on reliability (2 were based on single items
and had reliability estimated as .66, 2 were based on two
items and had reliability estimated as .795, and 1 used four
items and had reliability estimated as .886).
ANALYSIS AND RESULTS
Table 2 presents the results for the integration of effect
sizes of the SRP on satisfaction, repurchase intentions,
word-of-mouth, and image. Note that only those depen-
dent variables with 2 or more observations are presented
(trust, quality, intentions to complain, switching inten-
tions, pay-more intentions, and external response are not
submitted to further analyses because they were based on
a single observation).
In this stage of the analysis, 1 of the 24 studies pre-
sented in Table 1 (Oh 2003) had to be excluded because
it was classified as an outlier. Its sample size was 30,905
for the experimental group and 17,278 for the control
group (the mean sample size for the remaining studies
was 90 for the experimental group and 418 for the con-
trol group). If included, this study would produce an
inverse variance weight of 11,053, whereas the mean
value for the other studies was only 23. In summary, the
study was excluded because it would bias the cumulated
effect size.
10
All of the subsequent analyses were per-
formed after the exclusion of this study from the data set.
The integrated effect size (corrected for sampling and
measurement error) shows a positive and significant
value for satisfaction (r = .125, p < .017), supporting the
existence of the SRP for satisfaction (Hypothesis 1a),
with a medium effect size (Lipsey and Wilson [2001]
consider r .10 as small effect; .10 < r < .40 as medium
effect; and r .40 as large effect). This indicates that sat-
isfaction increases after a high service-recovery effort.
The confidence interval ranged from .032 to .217, sug-
gesting a small to medium effect. These results were
based on 18 independent observations and 7,502
subjects. Fail-safe N suggests that 27 studies with non-
significant effect size would be needed to reduce the
cumulated effect size to a level of just significant (a level
of .05 was used as “just significant,” similar to Grewal
et al. 1997).
11
In other words, to bring the significant
effect of the SRP on satisfaction down to the level of just
significant at α=.05, it would be necessary to find 27
studies with null results to be included in our analysis.
This is rather a small number, but it is somewhat unlikely
that with only 18 studies identified for satisfaction, 27
studies remain in the “file drawer,” especially because
this is a recent research topic in the services marketing
literature.
Conversely, the cumulated effect of the SRP on repur-
chase intentions is negative and not significant at the .05
significant level (r = –.07, p < .068), not supporting
Hypothesis 1b and suggesting that there is no SRP effect
on repurchase intentions. Confidence intervals ranged
from –.143 to –.002, indicating a small effect. These find-
ings were based on 12 independent observations and
7,788 subjects. Because of this small number of observa-
tions and the low magnitude of the average effect size, a
relatively small number was found for the fail-safe N,
indicating that 5 unpublished studies with null results
would be needed to reduce the average effect size to the
level of .05. Compared with satisfaction, it is more likely
that this number of unpublished studies exists. Therefore,
these results do not support Hypothesis 1b and indicate
that repurchase intentions are not increased by a high
service-recovery performance.
The integrated effect sizes of word-of-mouth and image
were also negative but not significant, and they were based
on a smaller number of observations (six and two, respec-
tively). These results do not support Hypotheses 1c and 1d.
Because of these null results, the file drawer number is not
calculated for word-of-mouth and image.
A heterogeneous subset of effect sizes was obtained
both for satisfaction and for repurchase intentions (see Q
test of homogeneity in Table 2), indicating that moderating
variables might help explain the variance in the effect
sizes. Hence, we tested whether there were differences in
effect sizes across study characteristics or, in other words,
if the coded study characteristics were significant modera-
tors. These analyses are presented in the next section.
Moderating Effects
A common procedure for testing whether studies’
characteristics can explain variability in the effect sizes is
regression analysis, in which effect sizes are entered as
dependent variables and moderators as independent vari-
ables (e.g., Eisend 2006; Szymanski and Henard 2001).
However, this procedure may be limited if there are only
few observations for each level of the moderators and/or
there is small number of effect sizes. In this case, confi-
dence in the results is threatened by low statistical power
and capitalization on sampling error (Hunter and
Schmidt 2004, p. 70). This was the case in our data set,
as there were only 18 effect sizes for satisfaction and 12
for repurchase intentions.
12
Because of this, we conducted a subgroup meta-analysis,
comparing the mean effect size between the levels of
each moderator, a common procedure in meta-analyses in
Matos et al. / SERVICE RECOVERY PARADOX 67
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
68 JOURNAL OF SERVICE RESEARCH / August 2007
marketing (see, e.g., Geyskens, Steenkamp, and Kumar
1998; Grewal et al. 1997; Palmatier et al. 2006; Pan and
Zinkhan 2006). Also, an additional moderator analysis
was conducted following the hierarchical method pro-
posed by Hunter and Schmidt (2004). In this method,
moderator variables are considered in combination to
avoid confounding of correlated moderators.
In Table 3, we present the results of the moderator
analyses, based on the subgroup meta-analyses. These
results show the mean corrected effect size (both for mea-
surement and sampling error) in each level of the moder-
ators, together with the number of studies and the test of
mean differences. Our moderator variables included
service category (hotels, restaurants, and others), method
(Survey × Experiment), design (Cross-Sectional ×
Longitudinal), subjects (Students × Nonstudents), and
scenario (used or not).
13
As indicated in Table 3, the use of surveys or exper-
iments did not change the direction of effect sizes,
either in satisfaction or in repurchase intentions, not
supporting Hypothesis 2a. On the other hand, longitudi-
nal studies tended to present relatively higher means for
satisfaction when compared to cross-sectional studies
(.192 versus .042, p < .056). This factor did not have
influence on the effect sizes of repurchase intentions,
not supporting Hypothesis 2b for this variable. Also,
studies using student samples presented higher mean-
effect sizes for satisfaction when compared to those
using nonstudent samples (.184 versus .060, p < .095).
Thus, there was support for Hypothesis 2c for satisfac-
tion only at the .10 significance level. Findings sug-
gested no difference between Students × Nonstudents in
the repurchase intentions, not supporting Hypothesis 2c
for this variable.
Considering service categories, studies conducted in
hotels presented relatively higher effect sizes for satisfac-
tion (.495) when compared to those in restaurants (.044)
or in the other categories (–.057), with significant
differences (p < .000 in both cases), which support
Hypothesis 2d for satisfaction. In repurchase intentions,
the difference between hotels versus others was repli-
cated (p < .002), and the studies conducted in restaurants
presented higher effect sizes when compared to studies in
other categories (p < .056), giving partial support for
Hypothesis 2d for repurchase intentions. These findings
indicated that the context in which the SRP is investi-
gated might also influence the results.
14
Following a recommendation by Hunter and Schmidt
(2004, p. 424) that subgroup meta-analysis may yield con-
founded results if moderators are correlated, we conducted
an additional test for moderators, using hierarchical analy-
sis. In this analysis, moderators are considered together.
We provide results from this analysis in Table 4, consider-
ing the three moderators dealing with the methodological
differences. One difficulty we encountered in this analysis
was the small number of studies in each cell when consid-
ering the eight groups (a combination of 2 × 2 × 2).
Despite this limitation, we could make five comparisons
for the satisfaction effect sizes and one for the repurchase
intentions. In these analyses, two factors are fixed and the
levels of the third factor are compared.
In Table 4, we first compare cross-sectional versus
longitudinal studies between experiments conducted with
students (.086 versus .239). Results did not suggest a sig-
nificant difference (p < .129). In the sequence, the same
comparison (cross-sectional versus longitudinal) was
made between experiments conducted with nonstudents
and a relatively higher effect size of satisfaction was
found for longitudinal studies (–.149 versus .156, p <
.085). When we compared cross-sectional versus longitu-
dinal between surveys conducted with nonstudents (.147
versus .103), no significant difference was found (p <
.399). This is an indication that the difference between
cross-sectional versus longitudinal (suggested in Table 3)
TABLE 3
Effects of Moderator Variables
Satisfaction Repurchase Intentions
No. of Effect Test Statistic No. of Effect Test Statistic
Moderator Level Studies Size (Z-Value) Sig. Studies Size (Z-Value) Sig.
Method Survey 4 .124 .005 .498 8 –.073 .049 .481
Experiment 14 .125 4 –.068
Design Cross-sectional 11 .042 1.590 .056 7 –.090 .962 .168
Longitudinal 7 .192 5 –.004
Subject Student 8 .184 1.311 .095 2 –.218 1.198 .115
Nonstudent 10 .060 10 –.060
Service category Hotel (a) 4 .495 3.388 (a versus b) .000 2 .324 –.746 (a versus b) .772
Restaurant (b) 5 .044 5.007 (a versus c) .000 2 .160 2.953 (a versus c) .002
Others (c) 9 –.057 .840 (b versus c) .200 8 –.112 1.593 (b versus c) .056
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
Matos et al. / SERVICE RECOVERY PARADOX 69
is likely to be higher when experiments are conducted
with nonstudents rather than with students.
Another comparison was made between students ver-
sus nonstudents, considering experiments using cross-
sectional design. In this case, a relatively higher effect
size of satisfaction was found for the studies using
students (.086 versus –.149, p < .088). This difference
was not statistically significant when we made this same
comparison in experiments using longitudinal approach.
Therefore, the difference in satisfaction effect sizes
between students versus nonstudents (suggested in Table
3) seems to be stronger for experiments using a cross-
sectional (rather than longitudinal) approach.
A similar analysis was conducted for the effect sizes
of repurchase intentions. However, the limitation of a
small sample size in each cell was more severe in this
case, as the total number of effect sizes was smaller for
this variable (n = 12). As a consequence, only one com-
parison was possible: cross-sectional versus longitudinal
for surveys using nonstudents (–.092 versus .100, p <
.137). However, no significant difference was found.
DISCUSSION
The meta-analysis presented in this study provides a
systematic review and a quantitative integration of the
effects of high recovery efforts on the dependent vari-
ables (satisfaction, repurchase intentions, word-of-
mouth, and corporate image), revealing the cumulative
effect of the SRP on these variables. Because there are
mixed results in the SRP literature, a meta-analysis can
help in understanding these inconsistencies by accumu-
lating results after adjusting for measurement and sam-
pling error and by identifying study characteristics that
may account for the variability in effect sizes. Although
the SRP is a relatively recent topic in the services mar-
keting literature (first empirical studies begin in the
1990s), a total of 21 studies (24 independent samples)
could be included in the meta-analysis and used for the
effect size integration.
Our primary results reveal support for the SRP only in
the case of satisfaction, with a mean adjusted effect size
of .125, which was significant at the 5% level. This is
interpreted as a medium effect size (Lipsey and Wilson
2001), with a confidence interval ranging from .032 to
.217. Thus, from the cumulative studies reviewed by our
meta-analysis, findings indicate that satisfaction
increases after a high service-recovery effort, suggesting
the existence of the SRP for this variable. This finding
supports the notion that a customer’s postfailure satisfac-
tion exceeds prefailure satisfaction (McCollough and
Bharadwaj 1992).
Based on this result, are recovery encounters really
good opportunities for service providers to increase cus-
tomer retention, as recommended by Hart, Heskett, and
Sasser (1990)? The empirical integration provided by our
meta-analysis suggests a negative answer, in that the SRP
effect was not evident for the repurchase intentions vari-
able at the 5% significance level. Even if we consider a
10% significance level, the small mean effect size with a
negative sign (–.072) suggests that the SRP effect on
TABLE 4
Hierarchical Moderator Analysis: Cross-Tabulation of Effect Sizes Among Moderators
Cross-Sectional Longitudinal Total
Student Nonstudent Student Nonstudent
Survey 2 (6) 2 (2) 4 (8)
.147
c
(–.092)
f
.103
c
(.10)
f
.453 (.072) .554 (.56)
Experiment 5 3 (1) 4 (2) 2 (1) 14 (4)
.086
a,d
–.149
b,d
(.013) .239
a,e
(–.218) .156
b,e
(.233)
.470 .353 (NC) .099 (.335) .537 (NC)
Total 5 (0) 5 (7) 4 (2) 4 (3) 18 (12)
NOTE: Values in each cell represent number of effect sizes, mean effect size, and significance; em-dashes indicate no data. Values of repurchase inten-
tions are in parentheses and values of satisfaction are outside parentheses. For example, there were two effect sizes of satisfaction in the category
Survey/Nonstudent/Cross-sectional, with a mean effect size of .147 and significance of .453. NC = significance not calculated when only one effect
size was available.
a. Contrast 1: .086 versus .239, p < .129.
b. Contrast 2: –.149 versus .156, p < .085.
c. Contrast 3: .147 versus .103, p < .399.
d. Contrast 4: .086 versus –.149, p < .088.
e. Contrast 5: .239 versus .156, p < .334.
f. Contrast 6: –.092 versus .10, p < .137.
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
repurchase intentions runs counter to what the SRP pre-
dicts. In other words, customers’ postfailure repurchase
intentions are likely to be lower than or equal to their pre-
failure intentions.
These results are interesting because they show that the
SRP works for satisfaction but not for repurchase inten-
tions. Customers are willing to make a positive evaluation
of a firm providing a high recovery effort, but they are not
likely to repatronize this firm. Why would this happen?
A possible explanation is that satisfied customers are
not necessarily loyal (Reichheld 1994). A meta-analysis
of customer satisfaction found, for instance, that it
explains less than 25% of the variance in repurchase
intentions (Szymanski and Henard 2001). In agreement
with this rationale, a recent study in e-retailing by Forbes,
Kelley, and Hoffman (2006) found that customers are not
likely to repurchase once a failure has been experienced,
even if they are satisfied with the recovery effort. These find-
ings might be influenced by the low switching costs that
customers experience in online shopping. Nevertheless,
they are further evidence of the importance of consider-
ing satisfaction levels and switching levels in combina-
tion (Jones and Sasser 1995).
Another possible explanation is the following line of
reasoning: When evaluating postrecovery satisfaction,
customers can be more influenced by the recovery
process itself and the positive rewards that it may provide
(e.g., a free service, a compensation for the failure), but
when evaluating their likelihood of repurchasing from
the same firm, customers might think that their original
desired result was not accomplished in the purchasing
process (the company did not provide a correct service in
the first time), and therefore, it is not worth repurchasing
from this firm. In agreement with this, a recent study by
Magnini et al. (2007) supports the notion that customers
who have experienced previous failure do not experience
the SRP effect (i.e., the SRP is more likely to occur when
it is the firm’s first failure with the customer).
The accumulated effect size was not significant for
word-of-mouth and corporate image. A limitation in this
case was the reduced number of available studies for test-
ing the SRP effect for these variables. As a consequence
of the nonsignificant result and the homogeneous effect
sizes, they were not included in the subsequent analysis
of moderators.
Further analyses of homogeneity of effect sizes for
satisfaction and repurchase intentions suggested possible
moderators, as heterogeneous effect sizes were found for
both variables. By using subgroup meta-analysis and
hierarchical moderating analysis, we first identified three
factors that moderated the SRP effect for satisfaction
(design, subject, and service category) and one factor
influencing repurchase intentions (service category).
Results suggested that effect sizes were not significantly
different across studies using experiments or surveys,
either for satisfaction or for repurchase intentions. This is
an indication that this methodological decision did not
affect the support/nonsupport of the SRP in the related
studies, in agreement with what is predicted by Michel
and Meuter (2006).
When comparing longitudinal versus cross-sectional
studies, it was found that longitudinal studies provided
stronger evidence for the SRP in satisfaction. This differ-
ence was further investigated in the hierarchical moderator
analysis. It was found not only that longitudinal studies
provided stronger support for the SRP effect for satisfac-
tion (contrary to the prediction of Michel and Meuter
2006), but also that the difference between cross-sectional
versus longitudinal was likely to be higher when experi-
ments were conducted with nonstudents, rather than with
students.
15
This finding is interesting because it suggests a
possible interaction between moderators.
Another influencing factor in the effect of SRP on sat-
isfaction was the use or nonuse of students as research
subjects (significance was found only at the 10% level).
It was found that students were more likely to support the
SRP than nonstudents were.
A possible explanation may be that nonstudents are
usually more experienced customers, and because of this,
they are less likely to experience a positive disconfirma-
tion if they have higher expectations. Furthermore,
students are usually researched out of the purchasing
environment and, therefore, their satisfaction-evaluation
process may not be influenced by their past experiences,
as in the case of “real” customers. Hence, satisfaction
evaluations of nonstudents will likely be lower than those
of students, and in the case of a service failure, this pat-
tern may be maintained because students out of their pur-
chasing context might not be as severe in their
evaluations as the nonstudents, who will also be closer to
their previous experiences and encounters with the firm.
These differences between students versus nonstu-
dents were more pronounced in cross-sectional experi-
ments when compared to longitudinal experiments,
suggesting again a possible interaction between modera-
tors. In summary, these results should make us reflect on
the following: If researchers conduct cross-sectional
experiments, then a difference in support for the SRP
might exist when using students or nonstudents; but
when conducting longitudinal experiments, no signifi-
cant difference exists between students or nonstudents.
As longitudinal studies provided stronger effect sizes for
the SRP, this is an indication that longitudinal experi-
ments are more likely to provide support for the SRP,
whether having students as respondents or not. This find-
ing contributes to the SRP literature by shedding light on
70 JOURNAL OF SERVICE RESEARCH / August 2007
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
Matos et al. / SERVICE RECOVERY PARADOX 71
the question “Are methodological aspects of the studies
influencing their support/nonsupport for the SRP?” Our
meta-analysis seems to offer an affirmative answer and
suggests that future studies should focus on these bound-
ary conditions of the SRP.
Finally, service category was also a significant moder-
ator, with influence both on satisfaction and on repur-
chase intentions. It was found that studies in hotels
provided higher support for the SRP effect on satisfaction
when compared to all other categories. This should also
be a point of future debate: “Are there differences across
service categories that facilitate the support for the
SRP?” For example, when a service failure occurs in a
context of the hospitality industry, most customers will
seek a solution to their problem, and therefore, there may
be a greater tendency toward redress-seeking behavior in
this context (McCollough 2000), and they may also have
a higher likelihood of receiving recovery service. In this
case, customers cannot easily look for another service
provider or interrupt their travel plans. As mentioned
above, switching costs may be an influencing factor in
this context. The same factor may explain why hotels and
restaurants scored higher in effect sizes for repurchase
intentions. Future studies could investigate this proposi-
tion further and test, for instance, the differences in the
SRP effects for customers with high versus low switch-
ing costs within a given industry.
An additional analysis of possible moderators indicated
that studies using more reliable scales did not provide sup-
port (or provided weaker support) for the SRP (a signifi-
cant negative correlation was found) both for satisfaction
and repurchase intentions. As expected, the same influence
was found for the number of items used to measure the
dependent variables. On the other hand, sample size was
not significantly related (at the 5% level) to the effect sizes
in these two dependent variables, although a correlation of
.43 (significant at the 10% level) was found between treat-
ment group and satisfaction, indicating that studies using
larger samples for the service recovery group were more
likely to provide support for the SRP. This is in agreement
with Michel and Meuter’s (2006) argument that, once the
SRP is a very rare event (Boshoff 1997), it becomes very
difficult to achieve a large sample of customers who have
received a very satisfactory recovery, and this limitation
may be responsible for the nonsignificant results presented
in the literature. Indeed, as we found, studies with a larger
number of respondents in the recovery group tend to pre-
sent greater support for the SRP and have higher statistical
power, as discussed in the next section.
Statistical Power
16
Statistical power is related to the probability of not
rejecting a false null hypothesis (Type II error, defined
by β). The power of an experiment is thus defined as
(1 – β) and interpreted as the probability that a statistical
test will correctly reject a false null hypothesis (Cohen
1988). Although Cohen (1988) recommends using .80 as
the threshold for statistical power, there is also a sugges-
tion for using .50 for the social sciences, in which errors
are less likely to have life-threatening consequences
(Muncer, Craigie, and Holmes 2003).
It has been argued that including the statistical power
discussion in the context of a meta-analysis can con-
tribute to enhance the reliability of the meta-analysis
(Muncer, Craigie, and Holmes 2003). Following the
guidelines presented by these authors, we (a) used the
mean effect size computed for the included studies to
estimate the average statistical power of the combined
studies and (b) estimated the statistical power of each
study, indicating the ability to detect an effect size of the
magnitude of the mean effect size obtained in the meta-
analysis (estimated as population effect size), given the
sample size and the significance level of .05. We used
G*Power 3.0
®
(Faul et al. in press) for these analyses and
the results are presented and discussed below.
The mean effect size of satisfaction (r
=
.125; d =
.251), in conjunction with the mean sample size of the
groups (n
treatment
= 57; n
control
= 401)
17
and the significance
level of .05, produced a statistical power of .42. This
value is below the recommended .80 threshold level but
relatively close to the level of .50 (Muncer, Craigie, and
Holmes 2003). For the studies of repurchase intentions
(r
=
–.072; d = –.145; n
treatment
= 134; n
control
= 558), statis-
tical power was estimated as .32. This analysis indicates
that the average statistical power was smaller in the 12
studies of repurchase intentions when compared to the 18
studies of satisfaction.
A deeper investigation of the statistical power of the
individual studies showed that power varied between .11
and .98 (mean = .30; SD = .23; n = 18) for the satisfaction
variable and between .07 and .80 (mean = .22; SD = .20;
n =12) for repurchase intentions. There was no significant
difference between these two means (t = .91, sig. = .37),
indicating that statistical power was not statistically differ-
ent between the observations of satisfaction and repur-
chase intentions. Note that these means are not weighted
by any factor, justifying why they are different from the
previous mean statistical power presented for satisfaction
(.42) and repurchase intentions (.32), when average statis-
tical power was computed directly from the final average
effect size weighted by sample size and reliability.
We also recomputed statistical power after excluding
from the database those studies with power lower than .5
and checked whether there were major changes in the
estimates of the mean effect size (weighted by sample
size and reliability), significance and confidence inter-
vals. For satisfaction, four studies produced a mean effect
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
72 JOURNAL OF SERVICE RESEARCH / August 2007
size of r = .334, with sig. = .018, and confidence intervals
between .20 and .47. We can note that these studies pro-
vide stronger support for the SRP in the case of satisfac-
tion (r changed from .125 to .334). Considering only
these four studies with power greater than .50, we should
conclude that there is a stronger SRP effect for satisfac-
tion, although it is still a medium effect.
For repurchase intentions, only one study presented
power higher than .50 (power = .80). This study pre-
sented an effect size of r = –.10, lower bound = –.20,
upper bound = .00. These values are similar to the ones
obtained from the 12 integrated studies (see Table 2: r =
–.072, lower bound = –.143, upper bound = .002), indi-
cating that including the studies with lower power did not
change the interpretation of the results in the cumulated
effect size of repurchase intentions.
18
Overall, these power analyses indicate that the studies
conducted to test the SRP and included in our meta-analy-
sis have relatively low statistical power, which can be one
additional reason why “conflicting results” are found
regarding the existence or nonexistence of the SRP. Our
analyses showed, for example, that the mean effect size for
satisfaction was greater (r = .33) when only studies with
acceptable power (higher than .50) were retained in the
analysis. This suggests a stronger effect (r = .33) than
when studies with lower power are also included in the
meta-analysis (r = .125). Because of the limited number of
studies, however, we could not include the power estimates
as one of the variables in our moderation analysis.
Implications and Further Research
The main implication of this meta-analysis is to show
that the SRP effect is more likely to occur for satisfaction
than for repurchase intentions. This result challenges us
to understand why satisfied customers are not neces-
sarily loyal. In a recent investigation of this topic,
Chandrashekaran et al. (2007), by decomposing satisfac-
tion in two factors (satisfaction level and satisfaction
strength), have theorized that weakly held satisfaction
does not translate into loyalty and that only strongly held
satisfaction is able to translate into loyalty. Based on this
rationale, future SRP studies could check the influence
that satisfaction strength might exert as a possible mod-
erator (i.e., the SRP effect may occur for repurchase
intentions when it occurs for satisfaction and this satis-
faction is strongly held).
Also, the lack of support for the SRP effect on repur-
chase intentions may suggest differences of the SRP on
the diverse stages of the loyalty pyramid (cognitive
affective conative action; Oliver 1997). Because
the studies included in our meta-analysis did not take this
factor into account empirically, future studies should be
conducted to investigate whether the SRP can exist for
the different stages of loyalty (e.g., the SRP may exist
during the cognitive stage but not the action stage).
Future studies are needed to provide further investiga-
tion of the reasons behind this difference of support of
the SRP in satisfaction but not in repurchase intentions.
In this sense, switching costs may be one of the relevant
factors accounting for this difference, as suggested by the
findings of Forbes, Kelley, and Hoffman (2006). The dif-
ference between “satisfaction level” and “satisfaction
strength” (Chandrashekaran et al. 2007) can also con-
tribute to this discussion.
Another possibility is that studies investigating repur-
chase intentions may have used approaches that tend to
give no support for the SRP. For example, 6 out of 12 of
the effect sizes for repurchase intentions came from
cross-sectional surveys with nonstudents. It was sug-
gested in the moderator analysis of satisfaction that (a)
studies using cross-sectional design produce relatively
smaller effect sizes when compared to longitudinal
approach, and (b) studies with nonstudents produced
lower effect sizes when compared to students use. Given
that 50% of the effect sizes for repurchase intentions used
this design, it may be possible that a SRP was not sup-
ported for repurchase intentions because of these
methodological differences across the studies.
Our moderation analyses were intended to be only an
exploratory investigation of the possible methodological
aspects that might have an influence on the effects of
SRP, especially because of the limited number of studies
in each cell in our hierarchical moderation procedure
(Hunter and Schmidt 2004). Because of this, future
research is necessary to provide further investigation of
these moderating effects.
Also, although theoretical moderators have been sug-
gested in the SRP literature (see Magnini et al. 2007),
because only few studies have tested them empirically,
they could not be included as variables in our data set,
19
as illustrated in Figure 1. Examples of these moderators
include severity of the failure, prior failure with the firm,
stability of the cause of the failure, and perceived
company control. Hence, we suggest that these modera-
tors be further investigated in future SRP studies.
Our revision of the mean effect sizes when taking sta-
tistical power into account, as suggested by Muncer,
Craigie, and Holmes (2003), indicates the relevance of
considering statistical power in the context of the studies
testing the SRP, especially for the satisfaction variable,
whose findings suggested a stronger effect of the SRP.
Thus, future studies of the SRP should consider statisti-
cal power a priori so as to be able to achieve greater con-
fidence in the results of the significance tests.
Our statistical power analysis can also be used to sug-
gest sample sizes for future studies. Based on the inte-
grated effect size as an estimate of the effect size of the
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
considered population, we can calculate the required
sample size that future studies should use to be able to
detect the desired effect. For example, for satisfaction,
considering sig. = .05, power = .8, it would be necessary
to have a sample size of 251 in each group (experimental
and control) to be able to identify an effect size of r
=
.125 (d = .251) as significant. This value would be
reduced to 123 if we considered the statistical power in
the suggested level of .50 (Muncer, Craigie, and Holmes
2003). On the other hand, for the repurchase intentions, it
would be necessary to have a sample size of 748 in each
of the two groups to identify an effect size of r
=
–.072
(d = –.145) with a statistical power of .8. If we lowered
the power to .5, the sample size would be reduced to 367.
As expected, for a given power, it is necessary to have a
larger sample size to identify smaller effect sizes.
Another relevant area for further research includes the
cognitive/affective mechanisms behind the SRP. Because
there is support for the SRP effect on satisfaction, as sug-
gested by the integrated results of the meta-analysis, and
satisfaction involves cognitive and affective dimensions
in prepurchase, purchase, and postpurchase phases of
consumption (e.g., Westbrook 1980), future studies
should investigate further the cognitive and/or affective
mechanisms driving the SRP effect. Although previous
studies have investigated the influence of the cognitive
and/or affective factors on satisfaction with service
recovery (e.g., Andreassen 2000), there is a lack of stud-
ies examining the cognitive and affective dimensions in
the context of the SRP.
As discussed by Parasuraman (2007), future research
should also investigate whether there is an optimal mix of
reliability versus recovery investments or, in other words,
how much should be invested in delivering reliable
service rather than in superior recovery when problems
occur. What are the main variables in this context and
what are their influences? These questions are also rele-
vant for future research on the service-recovery context.
Finally, we would suggest that more studies should be
conducted to investigate not only satisfaction and repur-
chase intentions as main recovery outcomes but also impor-
tant constructs like word-of-mouth and corporate image, for
which only a limited number of studies were found in the lit-
erature. Other relevant variables include trust, quality, inten-
tions to complain, and switching behavior. Testing the SRP
effect on the customers’ actual behavior rather than on their
behavioral intentions alone would also contribute to the cur-
rent state of the knowledge about the SRP.
Managerial Implications
Our meta-analysis has a number of implications for
service managers. First, the reviewed studies of the SRP
indicated that customer satisfaction after a high recovery
effort is greater when compared to that satisfaction prior
to the service failure. However, the same is not true for
the customer repurchase intentions. Because of this,
service managers should make every effort to provide
services correctly on the first time, rather than permitting
failures and then trying to respond with superior recov-
ery. This view has already been advocated by single
studies (e.g., Andreassen 2001; McCollough, Berry, and
Yadav 2000), but our meta-analysis makes this argument
stronger, given that the meta-analysis provides an inte-
grative review and a quantitative integration of the con-
flicting results about the SRP.
Second, trust is considered a key variable when man-
aging customer relationships (see Morgan and Hunt
1994). In the context of service failure and recovery, it is
expected that satisfaction with service recovery would
lead to the building of trust. However, because results
have demonstrated that customers who were initially sat-
isfied with the service expressed greater trust when com-
pared to the satisfied complainants, not supporting the
SRP in trust (Kau and Loh 2006), a service failure seems
to be a serious threat to trust. From the manager’s point
of view, it is critical to manage customer trust in the
service provider. Trust can be built and/or enhanced with
a company providing a reliable service over time. Hence,
service failure should be avoided also because of its neg-
ative impact on customer trust. Indeed, research investi-
gating why customers stay, given a switching dilemma,
has suggested that the most important reason (out of all
the 28 revealed reasons) was “lack of a critical incident,
or in other words, customers stayed because a negative
critical event had not occurred (Colgate et al. 2007).
Thus, the service provider should perform as promised if
customers’ perceived confidence is expected to be
strengthened. This confidence can be increased by invest-
ing in customers’ feeling of comfort, trust in the service
provider, satisfaction with the service provider, familiar-
ity with the service provider, history with the current
service provider, and lack of negative critical incidents
(Colgate et al. 2007). Service managers should invest in
these factors to earn the customer’s confidence.
Nevertheless, achieving 100% service reliability can be
impossible or cost prohibitive in most settings. Thus, in
case of a failure, companies should strive to provide a
service recovery of high performance anyway because a
delight with the recovery can contribute in moving the cus-
tomer up in the loyalty hierarchy (cognitive affective
conative action), as suggested by Andreassen (2001). In
other words, the effect of the SRP on repurchase intentions
may be influenced by the loyalty stage in which the cus-
tomer is found. This prediction needs to be confirmed
by future studies, and the findings from such studies will
Matos et al. / SERVICE RECOVERY PARADOX 73
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
provide service managers a deeper understanding of this
process. In the interim, we suggest that if a service failure
occurs, the first step should be to provide a recovery of
high performance in such a way to restore customer satis-
faction and achieve customer delight with the recovery
process. The second step should be to consider whether
this customer’s manifested satisfaction would be translated
into future loyalty with the company. In this stage, man-
agers could estimate through surveys the customer’s satis-
faction strength (i.e., the strength with which the
satisfaction judgment is held; Chandrashekaran et al.
2007) in such a way as to target those customers with
weakly held satisfaction, as they are less likely to turn their
satisfaction into loyalty (Chandrashekaran et al. 2007). For
instance, these customers could receive long-term benefits
(e.g., discounts based on history with the company), which
would increase their switching costs, and the company
would use the future service encounters as opportunities to
increase the customer satisfaction and make it more
strongly held.
Third, service managers should be cognizant of the dif-
ferences of the SRP across service settings. Findings from
the meta-analysis indicated a stronger effect size of satis-
faction in hotels, compared to all other categories, sug-
gesting that SRP is more likely in this setting. This may be
because of the high-contact characteristic of the hospitality
industry. Furthermore, this is a service of relatively longer
duration because even customers who stay for a short time
in a hotel (e.g., 1 day) may engage in a series of service
encounters. Also, given a service failure, the customer of a
hotel may be more prone to engage in redress seeking, as
he or she would not like (or be able) to change his or her
schedule (e.g., cancel a business meeting or a vacation
plan) because of this failure. Interestingly, there was also a
higher likelihood of the SRP effect in repurchase inten-
tions in the hotel setting. Thus, managers dealing with fail-
ures in a hotel context should also be able to provide
recoveries of high performance so as to boost customer
satisfaction and repatronage intentions.
In addition, service managers should also monitor cus-
tomers’ word-of-mouth, which can be very negative if a
failure occurs and the company is not able to provide a sat-
isfactory recovery (i.e., a “double deviation,” as termed by
Bitner, Booms, and Tetreault 1990). Indeed, the findings
from the meta-analysis showed that there was a negative
average effect size for word-of-mouth resultant from the
six independent observations reviewed. As recommended
by Andreassen (2001), positive word-of-mouth from exist-
ing customers can make the company more attractive in
the eyes of the new customers. Negative word-of-mouth
derived from unsatisfactory recoveries, on the other hand,
could push to competitors not only potential new
customers but also existing customers.
Limitations
Meta-analyses offer several benefits, but they also
have intrinsic limitations, which are common in most
meta-analytic studies in the marketing literature. We dis-
cuss the main limitations of our study below.
First, our analyses are based on secondary data, and
therefore, we cannot use information other than those
presented in the articles. For example, we could test the
methodological moderators presented in Figure 1 but not
the theoretical moderators because very few studies
tested the SRP considering these categories. Also, we had
to consider a mean group size when studies did not
inform the exact size of the groups being compared.
Moreover, only satisfaction and repurchase intentions
presented relatively high frequency of studies and could
be entered in further analysis of moderators. For
example, even though there was reference for other
recovery outcomes (e.g., trust, quality, intentions to com-
plain, and switching intentions), they could not be
included in our effect-size integration.
Second, although there were a large number of studies
investigating service failure and recovery (about 300 were
identified), a relatively small number of them tested the
SRP empirically and could be included in the analysis.
Even with the recommendation that 10 or more studies
should be an acceptable minimum number (see note 12),
we should be careful in interpreting the results based on
a small number of studies, especially regarding the mod-
erator analysis. We recognize that the presented modera-
tor analysis has a more exploratory perspective.
Nevertheless, results from the moderator analysis can
help researchers to design new studies that address the
boundary conditions for the SRP effect.
Finally, studies included in the meta-analysis pre-
sented relatively limited statistical power in general.
Because of this, the number of studies limited the con-
servative procedure of calculating more robust effect
sizes only from studies with acceptable statistical power.
When we implemented this procedure for satisfaction,
only 4 studies remained, with the remaining 14 studies
having low power. This is a clear indication that future
studies in this context should consider statistical power a
priori and determine the minimum sample size required
to detect the effect size.
CONCLUSION
Notwithstanding the presented limitations, the find-
ings from this meta-analysis contribute to a greater
understanding of the SRP by (a) estimating its cumulated
mean effect for the key dependent variables (satisfaction
74 JOURNAL OF SERVICE RESEARCH / August 2007
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
Matos et al. / SERVICE RECOVERY PARADOX 75
and repurchase intentions), (b) testing how studies char-
acteristics might influence these results, and (c) suggest-
ing further research directions.
Meta-analyses should not be viewed as conclusive or
as a substitute of new primary research but only as a
methodological tool that makes a temporary “balance
sheet” of the current state of the knowledge in a given
area. Its main contribution to science is to help
researchers to direct their next wave of research toward
still-unexplored questions and boundary conditions.
In this spirit, we hope that the results reported by our
meta-analysis provide managers and researchers with
inspirations for designing new in-depth and extensive
investigations that will keep advancing the services mar-
keting literature.
NOTES
1. Among our 24 studies, 10 used the term “repurchase intentions”
as dependent variables and only 2 considered “loyalty” (Kau and Loh
2006; Zeithaml, Berry, and Parasuraman 1996). Because these two
studies worked with loyalty intentions, we considered these variables
together, named as “repurchase intentions.
2. Indeed, we noticed in our data analysis process that the treatment
group was relatively smaller (mean = 90) when compared to the control
group (mean = 418).
3. Even severity of the failure could not be empirically evaluated in
our moderation analysis because only 4 studies tested SRP considering
Low × High severity. With missing values in the other 20 studies, this
variable could not be integrated in our moderation analyses.
4. The adjustments for unreliability can be applied directly to the
inverse variance weights using w = w × (r
yy
). In this approach, the effect
size is corrected both for sampling and measurement error (cf. Lipsey
and Wilson 2001, p. 110).
5. The formula used is: Q =
6. For most of the studies the total sample was greater than the sum
of the treatment and the control group. This was because each study
tested relationships other than the paradox (e.g., a given study used a
2 × 2 × 2 experiment with a control group, and the paradox was tested
comparing one of the eight experimental groups with the control
group). In this case, we coded three variables: (a) total sample size from
the nine groups, (b) size of treatment group, and (c) size of the control
group.
7. It is expected that measures with more items produce greater reli-
ability (Nunnally and Bernstein 1994) and then stronger effect sizes.
8. Seven out of the 24 studies used single items for any of the mea-
sured variables; a total of nine observations of the 45 values in our data set.
9. There were only four missing values of reliability for studies
using multiple items in dependent variables.
10. The cumulated effect size for satisfaction would become –.135
if Oh (2003) were included in the data set. The observed effect size for
this study was –.149.
11. The formula for the fail-safe number is k × (r r
c
) / r
c
, where k
is the accumulated number of studies, r is the mean effect size, and r
c
is
the critical effect size, or the “just significant” level (Hunter and
Schmidt 2004, p. 501; Lipsey and Wilson 2001, p.166).
12. We asked experts in meta-analysis what would be the minimum
number of studies needed to conduct a meta-analysis. Professor Frank
L. Schmidt (author of the book Methods of Meta-Analysis; see Hunter
and Schmidt 2004) told us, “The really bare minimum is 2 studies, but
most journals will not publish a meta-analysis unless it contains at least
5 to 7 studies. So with 10 or more, you are OK” (personal communica-
tion, October 16, 2006). Professor David B. Wilson (author of the book
Practical Meta-Analysis; see Lipsey and Wilson 2001) also said,
“Minimum number of studies: 2. Of course, this limits the analyses that
you can do” (personal communication, October 14, 2006).
13. Method and scenario are analyzed together because all studies
using experiments are based on scenarios, and none of the surveys used
scenarios.
14. To achieve a better understanding of the variability of the effect
sizes, we also checked if effect sizes were influenced by scale reliabil-
ity, number of items measuring dependent variables, and sample size.
We found a significant correlation between effect size and scale relia-
bility (–.62 in satisfaction and –.57 in repurchase intentions), with neg-
ative values indicating that studies using more reliable scales did not
provide support (or provided weaker support) for the SRP. As expected,
significant positive correlations were found between number of items
measuring the dependent variable and scale reliability (.59 in satisfac-
tion and .81 in repurchase intentions). As a result, the number of items
measuring satisfaction was also negatively related to the satisfaction
effect size (r = –.47) and the same was true for repurchase intentions
(r = –.57). Sample size (total, treatment, and control) was not signifi-
cantly correlated with effect sizes either in satisfaction or in repurchase
intentions (the highest correlation was found in the pair treatment
group–satisfaction, r = .43, p < .075). A regression analysis of these
variables on the effect sizes produced no significant results for either
satisfaction or repurchase intentions. However, these regression results
might not be reliable because they may be influenced by low statistical
power and capitalization on chance because of the small number of
observations (see Hunter and Schmidt 2004, p. 70).
15. We could not test if surveys with nonstudents (rather than with
students) produce greater difference between cross-sectional versus
longitudinal because there were no observation for the cells “survey
with students.
16. We are very thankful to Reviewer C for suggesting the inclusion
of this topic in our discussion.
17. These are mean values based on 18 observations with total sam-
ple size of 7,218 in the control group (7,218 / 18 = 401) and 1,015 in the
experimental group (1,015 / 18 = 57). For repurchase intentions, 12 obser-
vations had a total sample size of 6,686 in the control group (6,686 / 12 =
558) and 1,601 in the experimental group (1,601 / 12 = 134).
18. We also checked if there was a correlation between power and
effect sizes of the individual studies. This correlation was not signifi-
cant for either satisfaction or repurchase intentions.
19. Most of the studies did not test these moderators or provide
information that could allow the authors to classify the study in one cat-
egory or another. Because of this, we chose to tabulate only information
available in the studies.
REFERENCES
*Asterisks denote studies used in meta-analysis.
Achrol, Ravi S. (1991), “Evolution of the Marketing Organization: New
Forms for Turbulent Environments,Journal of Marketing, 55 (4),
77-93.
Andreassen, Tor Wallin (2000), “Antecedents to Satisfaction with Service
Recovery, European Journal of Marketing, 34 (1/2), 156-75.
*——— (2001), “From Disgust to Delight—Do Customers Hold a
Grudge?” Journal of Service Research, 4 (1), 39-49.
Assmus, Gert, John U. Farley, and Donald R. Lehmann (1984), “How
Advertising Affects Sales: Meta-Analysis of Econometric Results,
Journal of Marketing Research, 21 (1), 65-74.
Bitner, Mary Jo, Bernard H. Booms, and Mary S. Tetreault (1990), “The
Service Encounter: Diagnosing Favorable and Unfavorable
Incidents,” Journal of Marketing, 54 (1), 71-84.
———, ———, and L. A. Mohr (1994), “Critical Service Encounters:
The Employee’s Viewpoint,Journal of Marketing, 58 (3), 95-106.
(
Σ
w
i
ES
i
2
(
Σ
w
i
ES
i
)
2
)
Σ
w
i
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
76 JOURNAL OF SERVICE RESEARCH / August 2007
Blodgett, Jeffrey G., Donald H. Granbois, and Rockney G. Walters
(1993), “The Effects of Perceived Justice on Complainants’
Negative Word-of-Mouth Behavior and Repatronage Intentions,
Journal of Retailing, 69 (4), 399-428.
Bolton, Ruth N. and James H. Drew (1991), “A Multistage Model of
Customers’ Assessments of Service Quality and Value,Journal of
Consumer Research, 17 (March), 375-84.
*Boshoff, Christo (1997), “An Experimental Study of Service Recovery
Options,International Journal of Service Industry Management,8
(2), 110-30.
Brown, Steven and Douglas M. Stayman (1992), “Antecedents and
Consequences of Attitude toward the Ad: A Meta-Analysis,
Journal of Consumer Research, 19 (1), 34-51.
Chandrashekaran, Murali, Kristin Rotte, Stephen S. Tax, and Rajdeep
Grewal (2007), “Satisfaction Strength and Customer Loyalty,
Journal of Marketing Research, 44 (1), 153-63.
Cohen, Jacob (1988), Statistical Power Analysis for the Behavioral
Sciences, 2nd ed. Hillsdale, NJ: Lawrence Erlbaum.
Colgate, Mark, Vicky T. U. Tong, Christina K. C. Lee, and John U.
Farley (2007), “Back from the Brink: Why Customers Stay,
Journal of Service Research, 9 (3), 211-28.
DelVecchio, Devon, David H. Henard, and Traci H. Freling (2006),
“The Effect of Sales Promotion on Post-Promotion Brand
Preference: A Meta-Analysis,Journal of Retailing, 82 (3), 203-13.
Eisend, Martin (2004), “Is It Still Worth to be Credible? A Meta-
Analysis of Temporal Patterns of Source Credibility Effects in
Marketing, in Advances in Consumer Research, Vol. 31, Barbara
Kahn and Mary Frances Luce, eds. Urbana, IL: Association of
Consumer Research, 352-57.
——— (2006), “Two-Sided Advertising: A Meta-Analysis,
International Journal of Research in Marketing, 23 (2), 187-98.
Etzel, Michael J. and Bernard I. Silverman (1981), “A Managerial
Perspective on Directions for Retail Customer Dissatisfaction
Research,Journal of Retailing, 57 (3), 124-36.
Farley, John U., Donald R. Lehmann, and Alan Sawyer (1995),
“Empirical Marketing Generalization Using Meta-Analysis,
Marketing Science, 14 (3), 36-46.
Faul, Franz, Edgar Erdfelder, Albert-Georg Lang, and Axel Buchner (in
press), “G*Power 3: A Flexible Statistical Power Analysis Program
for the Social, Behavioral, and Biomedical Sciences,Behavior
Research Methods, http://www.psycho.uni-duesseldorf.de/abteil
ungen/aap/gpower3/.
Folkes, Valerie (1984), “Consumer Reactions to Product Failure: An
Attributional Approach, Journal of Consumer Research,10
(March), 398-409.
——— (1988), “Recent Attribution Research in Consumer Behavior: A
Review and New Directions,Journal of Consumer Research,14
(March), 548-60.
Forbes, Lukas P., Scott W. Kelley, and K. Douglas Hoffman (2006),
“Typologies of E-Commerce Retail Failures and Recovery
Strategies, Journal of Services Marketing, 19 (5), 280-92.
Franke, George R. and Jeong-Eun Park (2006), “Salesperson Adaptive
Selling Behavior and Customer Orientation: A Meta-Analysis,
Journal of Marketing Research, 43 (4), 693-702.
Geyskens, Inge, Jan-Benedict E. M. Steenkamp, and Nirmalya Kumar
(1998), “Generalizations about Trust in Marketing Channel
Relationships Using Meta-Analysis,International Journal of
Research in Marketing, 15 (3), 223-49.
Grewal, Dhruv, Sukumar Kavanoor, Edward F. Fern, Carolyn Costley,
and James Barnes (1997), “Comparative versus Noncomparative
Advertising: A Meta-Analysis,Journal of Marketing, 61 (4), 1-15.
*Halstead, Diane and Thomas J. Page (1992), “The Effects of
Satisfaction and Complaining Behavior on Consumer Repurchase
Intentions, Journal of Consumer Satisfaction, Dissatisfaction and
Complaining Behavior, 5, 1-11.
Hart, Christopher W., James L. Heskett, and W. Earl Sasser, Jr. (1990),
“The Profitable Art of Service Recovery,Harvard Business
Review, 68 (August), 148-56.
Hedges, Larry V. and Ingram Olkin (1985), Statistical Methods for
Meta-Analysis. Orlando, FL: Academic Press.
*Hocutt, Mary Ann, Goutam Chakraborty, and John Mowen (1997),
“The Impact of Perceived Justice on Customer Satisfaction and
Intention to Complain in a Service Recovery,Advances in
Consumer Research, 24, 457-63.
*——— and Thomas H. Stone (1998), “The Impact of Employee
Empowerment on the Quality of a Service Recovery Effort,
Journal of Quality Management, 3 (1), 117-32.
*———, Michael R. Bowers, and D. Todd Donavan (2006), “The Art
of Service Recovery: Fact or Fiction?” Journal of Services
Marketing, 20 (3), 199-207.
Hunter, John E. and Frank L. Schmidt (2004), Methods of Meta-
Analysis: Correcting Error and Bias in Research Findings, 2nd ed.
Beverly Hills, CA: Sage.
Janiszewski, Chris, Hayden Noel, and Alan G. Sawyer (2003), “A
Meta-Analysis of the Spacing Effect in Verbal Learning:
Implications for Research on Advertising Repetition and Consumer
Memory,Journal of Consumer Researcher, 30 (1), 138-49.
Jones, Thomas O. and W. Earl Sasser, Jr. (1995), “Why Satisfied
Customers Defect,Harvard Business Review, 73 (November-
December), 88-99.
*Kau, Ah-Keng and Elizabeth Wan-Yium Loh (2006), “The Effects of
Service Recovery on Consumer Satisfaction: A Comparison
between Complainants and Non-Complainants,Journal of
Services Marketing, 20 (2), 101-11.
Kelley, Scott W., K. Douglas Hoffman, and Mark A. Davis (1993), “A
Typology of Retail Failures and Recoveries,Journal of Retailing,
69 (4), 429-52.
*Kwortnik, Robert J. (2006), “Shining Examples of Service when the
Lights Went Out: Hotel Employees and Service Recovery during
the Blackout of 2003,Journal of Hospitality and Leisure
Marketing, 14 (2), 23-45.
Lipsey, Mark W. and David B. Wilson (2001), Practical Meta-Analysis.
Thousand Oaks, CA: Sage.
*Magnini, Vincent P., John B. Ford, Edward P. Markowski, and Earl D.
Honeycutt (2007), “The Service Recovery Paradox: Justifiable
Theory or Smoldering Myth?” Journal of Services Marketing,
21 (3), 213-225.
*Mattila, Anna S. (1999), “An Examination of Factors Affecting
Service Recovery in a Restaurant Setting,Journal of Hospitality &
Tourism Research, 23 (3), 284-98.
*Maxham, James G., III (2001), “Service Recovery’s Influence on
Consumer Satisfaction, Positive Word-of-Mouth, and Purchase
Intentions,Journal of Business Research, 54 (1), 11-24.
*——— and Richard G. Netemeyer (2002), “A Longitudinal Study
of Complaining Customer’s Evaluations of Multiple Service
Failures and Recovery Efforts,Journal of Marketing, 66 (4),
57-71.
*McCollough, Michael A. (1995), “The Recovery Paradox: A
Conceptual Model and Empirical Investigation of Customer
Satisfaction and Service Quality Attitudes after Service Failure and
Recovery,” doctoral dissertation, Texas A&M University, College
Station, United States.
*——— (2000), “The Effect of Perceived Justice and Attributions
regarding Service Failure and Recovery on Post-Recovery
Customer Satisfaction and Service Quality Attitudes,Journal of
Hospitality & Tourism Research, 24 (4), 423-47.
——— and Sundar G. Bharadwaj (1992), “The Recovery Paradox: An
Examination of Consumer Satisfaction in Relation to
Disconfirmation, Service Quality, and Attribution Based Theories,
in Marketing Theory and Applications, Chris T. Allen et al., eds.
Chicago: American Marketing Association, 119.
*———, Leonard L. Berry, and Manjit S. Yadav (2000), “An Empirical
Investigation of Customer Satisfaction after Service Failure and
Recovery, Journal of Service Research, 3 (2), 121-37.
*Michel, Stefan (2001), “Analyzing Service Failures and Recoveries: A
Process Approach, International Journal of Service Industry
Management, 12 (1), 20-33.
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
Matos et al. / SERVICE RECOVERY PARADOX 77
*——— and Matthew L. Meuter (2006), “The Service Recovery
Paradox: True but Overrated?” working paper, www.dienstleis
tungsmarketing.ch/documents/IJSIMParadoxMichelMeuter.pdf.
Moorman, Christine, Gerald Zaltman, and Rohit Deshpande (1992),
“Relationships between Providers and Users of Market Research:
The Dynamics of Trust within and between Organizations,Journal
of Marketing Research, 29 (3), 314-39.
Morgan, Robert and Shelby Hunt (1994), “The Commitment-Trust
Theory of Marketing Relationships,Journal of Marketing,
58 (3), 20-38.
Muncer, Steven, Mark Craigie, and Joni Holmes (2003), “Meta-
Analysis and Power: Some Suggestions for the Use of Power in
Research Synthesis,Understanding Statistics, 21 (1), 1-12.
Nunnally, Jum and Ira Bernstein (1994), Psychometric Theory. New
York: McGraw-Hill.
*Oh, Haemoon (2003), “Reexamining Recovery Paradox Effects and
Impact Ranges of Service Failure and Recovery,Journal of
Hospitality & Tourism Research, 27 (4), 402-18.
*Ok, Chihyung, Ki-Joon Back, and Carol W. Shanklin (2006), “Service
Recovery Paradox: Implications from an Experimental Study in a
Restaurant Setting,Journal of Hospitality & Leisure Marketing,14
(3), 17-33.
Oliver, Richard L. (1997), Satisfaction: A Behavioral Perspective on the
Consumer. New York: McGraw-Hill.
Palmatier, Robert W., Rajiv P. Dant, Dhruv Grewal, and Kenneth R.
Evans (2006), “Factors Influencing the Effectiveness of
Relationship Marketing: A Meta-Analysis,Journal of Marketing,
70 (4), 136-53.
Pan, Yue and George M. Zinkhan (2006), “Determinants of Retail
Patronage: A Meta-Analytical Perspective,Journal of Retailing,82
(3), 229-43.
Parasuraman, A. (2007), “Modeling Opportunities in Service Recovery
and Customer-Managed Interactions,Marketing Science, 25 (6),
590-93.
Reichheld, Frederick F. (1994), “Loyalty and the Renaissance of
Marketing,Marketing Management, 2 (4), 10-21.
Rosenthal, Robert (1979), “The ‘File Drawer Problem’ and Tolerance
for Null Results,Psychological Bulletin, 86, 638-41.
Singh, J. (1990), “Voice, Exit and Negative Word-of-Mouth Behaviors:
An Investigation across Three Service Categories,Journal of the
Academy of Marketing Science, 18 (1), 1-15.
*Smith, Amy K. and Ruth N. Bolton (1998), “An Experimental
Investigation of Service Failure and Recovery: Paradox or Peril?”
Journal of Service Research, 1 (1), 65-81.
Szymanski, David M. and David H. Henard (2001), “Customer
Satisfaction: A Meta-Analysis of the Empirical Evidence,Journal
of the Academy of Marketing Science, 29 (1), 16-35.
Tax, Stephen S., Stephen W. Brown, and Murali Chandrashekaran
(1998), “Customer Evaluations of Experiences: Implications for
Relationship Marketing,Journal of Marketing, 62 (2), 60-76.
Weun, Seungoog, Sharon E. Beatty, and Michael A. Jones (2004), “The
Impact of Service Failure Severity on Service Recovery Evaluations
and Post-Recovery Relationships,Journal of Services Marketing,
18 (2/3), 133-46.
Westbrook, Robert A. (1980), “Intrapersonal Affective Influences upon
Consumer Satisfaction with Products,Journal of Consumer
Research, 7 (1), 49-54.
*Zeithaml, Valerie A., Leonard L. Berry, and A. Parasuraman (1996),
“The Behavioral Consequences of Service Quality,Journal of
Marketing, 60 (2), 31-47.
Celso Augusto de Matos is a marketing doctoral candidate at the
School of Management, Federal University of Rio Grande do Sul
(PPGA-EA-UFRGS), Brazil. His main research interests lie in the
areas of consumer behavior in services, attitude formation and
change, and marketing research. His research has been published
in the Journal of Consumer Marketing, International Journal of
Consumer Studies, ACR Conference, and in a number of Brazilian
journals and proceedings. He can be contacted at:
celsomatos@yahoo.com.br.
Jorge Luiz Henrique is a marketing doctoral candidate at the
School of Management, Federal University of Rio Grande do Sul
(PPGA-EA-UFRGS), Brazil. His research focuses on consumer
behavior in services and relationship marketing. He is a marketing
manager at Banco do Brasil (Bank of Brazil). His research has been
published in the Journal of Internet Banking and Commerce,
Global Information Technology Management (GITM), The
Business Association of Latin American Studies (BALAS),
International Association for Management of Technology (IAMOT)
conferences, and in a number of Brazilian journals and proceedings.
Carlos Alberto Vargas Rossi is a professor of marketing at the
School of Management, Federal University of Rio Grande do Sul
(PPGA-EA-UFRGS), Brazil. His research interests are consumer
behavior and marketing theory. His research has been published
in the Journal of Consumer Marketing, International Journal of
Consumer Studies, ACR, AMA and EMAC conferences, and in a
number of Brazilian journals and proceedings.
at CAPES on October 30, 2009 http://jsr.sagepub.comDownloaded from
... When there is a problem with the services offered by an organization, consumers' trust in the firm will be compromised, requiring the organization to retain its consumers through efforts such as service recovery or trust repair [43]. A prior meta-analysis [44] identified the outcomes of service recovery as satisfaction, re-purchase intention, word-of-mouth, and corporate image. Trust repair, broadly speaking, shares a similar purpose with service recovery, in such a manner that their potential outcomes should thus be similar. ...
... This may be performed in-line with restoring the customers' trust. The second phase involves the intention to continue cooperation, which indicates that the customers are satisfied with the organization's services, and they may even engage in word-of-mouth communication [44]. With regards to the context of this study, the main purpose of the trust repair strategies proposed by healthcare institutions in response to medical disputes (i.e., affective, functional, and informational) are to affect trust repair, which is a prerequisite of patient satisfaction and willingness-to-partake in word-of-mouth communication. ...
Article
Full-text available
Orthodontic treatment has popularized in Taiwan. Healthcare institutions can be responsive in their coping strategies and determine whether third-party intervention should take place involving medical disputes related to orthodontics in order to repair patient trust. This study draws on orthodontic treatment to explore the effect of various trust repair strategies employed by healthcare institutions and third-party involvement positively affecting outcomes related to trust repair. Patients were recruited among those who have undergone orthodontic treatments, and 353 valid scenario-based questionnaires were collected through an online survey. Results revealed that: (1) the affective and informational repair strategies positively impacted trust repair while the functional repair strategy did not; (2) trust repair positively impacted patient satisfaction/word-of-mouth and mediated between repair strategies and satisfaction/word-of-mouth; and (3) third-party involvement moderated the relationship between trust repair and word-of-mouth. The findings suggest that rather than receiving monetary compensation, patients usually prefer that healthcare institutions acknowledge their fault, offer apologies, and engage in active communications to clarify the causes of medical dispute. Further, an objective third party should be involved to mediate the medical disputes to afford satisfaction all around.
... Converging evidence shows that devoting appropriate recovery efforts can mitigate the negative effect of service failures (Jeong and Lee, 2017;Muhammad and Gul-E-Rana, 2019;Riaz and Khan, 2016). Service recovery efforts refer to the perceived energy and resources dedicated by service employees (Mostafa et al., 2014) and organizations (De Matos et al., 2007). Since the service recovery efforts are aimed to achieve customers' positive evaluation of service recovery, previous studies provide mixed findings on the effectiveness of service recovery efforts (Harun et al., 2018). ...
... Following previous studies (e.g. Cai and Qu, 2018;De Matos et al., 2007;Mostafa et al., 2014) this study aims to extend the knowledge by examining the underlying mediator in firms' service recovery efforts and consumer forgiveness and how varying levels of recovery efforts lead to consumer forgiveness. ...
Article
Purpose This paper investigated the impact of firms' service recovery efforts on consumers' desire to reciprocate and forgiveness in the hospitality industry of Pakistan. Additionally, this study examined the mediating role of perceived justice between service recovery efforts and their outcomes. Design/methodology/approach Using snowball sampling technique, an online survey was administered and 259 responses were collected from casual-dining restaurant customers. A partial least squares structural equation modeling (PLS-SEM) and multivariate analysis of covariance (MANCOVA) were used to examine the hypotheses. Findings The results indicate that perceived justice significantly mediates the effect of service recovery efforts on the consumers' desire to reciprocate and forgiveness. Moreover, high (vs. low) service recovery efforts lead to high consumer forgiveness. Practical implications The study provides insights for managers on how optimal recovery efforts predict consumers' positive responses and minimize the effect of service failure in South Asian consumers. Originality/value This research is among the early endeavors to examine consumers' desire to reciprocate in service recovery context. Also, this is the first study to validate the impact of service recovery efforts on consumers' desire to reciprocate and consumer forgiveness in a South Asian country.
... Marketing praxis emphasize the relevance of customer retention to overall wellness of a business. Firms and customers alike increasingly look for mutually satisfying longterm relationships because it provides increased competitiveness for firms and reduce purchase related risks for customers (Augusto de Matos et al., 2007). To be successful, a firm's effort to retain customers must start with the initial contact they make with a customer and continue throughout the duration of the relationship. ...
Article
Full-text available
The study examined the relationship between price fairness and customer loyalty in food and beverage firms in Rivers State. The study adopted correlational research design. Data were collected from a sample of 132 customers of food and beverage firms using a structured questionnaire. The study utilized the Spearman's Rank Order Correlation (r) to test its hypotheses at 0.05 level of significance, relying on the Statistical Package for the Social Sciences (SPSS) version 22.0. The study found that price fairness has significant positive relationship with customer loyalty through repeat purchase and customer retention. The study concludes price fairness drives customer loyalty in food and beverage firms in Rivers State; and recommends that managers of food and beverage firms should avoid over-pricing of their products in order not to lose their customers to competitors.
... First, the use of an undergraduate student sample with the university as the focal brand provided interesting and relevant results. However, given that the type of industry is shown to affect the strategies proposed and the response to such strategies on a crisis (De Matos et al., 2007), a replication of the study in different sectors is advised. Particularly in consideration of the contextual factors surrounding the university-student relationship (such as the contractual element, the potential "parental" ...
Article
Full-text available
This paper investigates the relationship between firm crisis behavior and the resulting consumer–brand relationship (CBR) response. Drawing from theoretical traditions in brand transgressions, service failure, and crisis communications, we use longitudinal survey data combined with archival social media data to empirically test the effect of crisis response speed and crisis information strategy on the short‐term consumer crisis response evaluations (1 month after crisis response), and the long‐term CBR (1 year after crisis response). Results show that, contrary to intuitive expectations, a faster firm response is not always better, as a slower response was found to result in higher crisis response evaluations. We also show that this effect depends on the consistency of the communication strategy with the first active response. Specifically, when a firm prioritizes safety information (instructing strategy), a faster response is better. Whereas, when the firm prioritizes well‐being information (adjusting strategy), a slower response is better. We argue the counterintuitive finding that a slower response is better implies that reacting too quickly may signal rashness and unpreparedness to the customer, leading to more negative evaluations. We term this distinction the difference between being responsive (fast but considered) and reactive (faster but rash).
Article
While monetary compensation is considered the most effective service recovery strategy, relief theory claims that humor may also be useful in service recovery situations. This study investigated the effects of humor in service recovery using dynamic causal modeling and parametric empirical Bayes analysis to identify effective connectivity (EC) patterns in the dopaminergic reward system across four conditions representing different service recovery strategies: monetary compensation and humor (MH), monetary compensation and an apology (MA), non-monetary compensation using humor (H), and non-monetary compensation using an apology (CON, the control condition). The findings support the importance of the nucleus accumbens (NAc) in the monetary compensation (MH and MA) conditions and the amygdala in the non-monetary compensation (H and CON) conditions. Monetary compensation (MH and MA) resulted in right substantia nigra (rSN) to NAc EC, suggesting the processing of recovery satisfaction associated with perceived outcome fairness. Conversely, non-monetary compensation strategies (H and CON) resulted in left substantia nigra (lSN) to amygdala EC, suggesting the processing of satisfaction related to perceived interactional fairness. The use of humor for service recovery resulted in VTA-to-lSN-to-amygdala EC during humor appreciation, while the use of apologies (CON and MA) resulted in lSN-to-amygdala and lSN-to-VTA connectivity. Surprisingly, processing satisfaction in the MH condition did not activate the amygdala during humor appreciation. Coping humor could be norm-violating for service recovery, and its effectiveness depends on multiple factors. The results suggest that monetary compensation, humorous responses, and apologies play key roles in neurological responses to service recovery strategies.
Article
Although service customization is widely adopted in the tourism industry to cater to the heterogeneous needs of customers, the high interaction in customization often results in a value decrease for the service provider and tourists (i.e., value codestruction). However, the literature has not fully explored the restoration process when value codestruction occurs during the consumption stage of customized tourism. Based on conservation of resources theory, this study identified a systematic value restoration framework by investigating the factors of codestruction situation, restoration approach, and customer characteristics. Our experiments indicate that value can only be restored effectively when the restoration approach is congruent with the value reducer type. Moreover, customers with low self-efficacy rely more on such a congruency effect. The findings provide managers and tourists with insights into how to optimize tourism experiences by displaying flexibility in dealing with value codestruction in customized tourism.
Article
Customer–company identification (CCI) has been highlighted as an important mechanism that explains the relationship between corporate social responsibility (CSR) and customer outcomes such as customer loyalty and word-of-mouth (WOM). However, findings on when and how this mechanism works are mixed. To uncover the viability and strength of CSR’s indirect effect on customer outcomes through CCI, we conduct a meta-analysis testing a moderated mediated framework. This analysis incorporates 237 independent effect sizes from 58,766 individuals and 86 papers to examine the indirect effect of CSR on customer loyalty and WOM through CCI, while simultaneously testing a range of substantive and control moderators. The results reveal that 1) CSR has a main effect on CCI, 2) CCI mediates the effect of CSR on customer loyalty and WOM, and 3) there are significant theoretical moderators that amplify and reduce CSR’s relationship with CCI. The paper’s year of publication and industry controversy (versus non-controversy) mitigate the relationship between CSR and CCI, while collectivism and a holistic focus augment it.
Article
Buy‐online‐pickup‐instore (BOPIS) services have become an increasingly important part of a retailer's omnichannel strategy. When service failures (e.g., stock‐out) occur, consumers may resort to negative word‐ofmouth (NWOM) to share their evaluation of the retailer's BOPIS service. While a retailer's service recovery policies (e.g., cross‐channel substitution) may help to fulfill its service intent, the extent to which these two signals can improve consumer satisfaction and diminish their NWOM intent remains unknown. Drawing from both service recovery literature and signaling theory, we conducted a series of five experiments and find that the intradimensional congruity of the signal set communicated by the retailer during its BOPIS service process depends on both its operational capability and the consumer's own predilection regarding the product category. These insights collectively indicate that while a retailer's operations need to support service policies to provide a congruous BOPIS service process, substitution policies offered to consumers during the transaction need to consider the extent to which a consumer's purchase decision is hedonic or utilitarian. In turn, this finding suggests that a retailer's category management needs to consider BOPIS substitution in terms of both product assortment and inventory policies.
Article
While consumers frequently attempt to resolve their own consumption problems (i.e., do-it-yourself (DIY)), they are often unsuccessful and subsequently turn to a professional. In the present research, we consider DIY failure as a form of service failure (SF) and demonstrate that experiencing DIY service failure (DIY SF) influences consumer evaluations of subsequent firm recovery. This occurs because consumers who experience DIY SF gain greater understanding of the task (i.e., learning) through their failed attempt. This learning promotes increased appreciation of the recovering service provider’s ability, ultimately resulting in greater satisfaction with the recovery offering. We further identify mindset as a moderator of this effect, wherein those with a growth mindset are more likely to learn from failure and appreciate the abilities of the recovering service provider. By highlighting DIY SF as a novel form of SF, we demonstrate the importance of understanding customers’ prior experiences with the focal consumption problem and its solution, and of training front-line employees to better manage these customers. We test our theory across four studies using lab and field data, and close by discussing theoretical and managerial implications.
Article
Full-text available
If service quality relates to retention of customers at the aggregate level, as other research has indicated, then evidence of its impact on customers’ behavioral responses should be detectable. The authors offer a conceptual model of the impact of service quality on particular behaviors that signal whether customers remain with or defect from a company. Results from a multicompany empirical study examining relationships from the model concerning customers’ behavioral intentions show strong evidence of their being influenced by service quality. The findings also reveal differences in the nature of the quality-intentions link across different dimensions of behavioral intentions. The authors’ discussion centers on ways the results and research approach of their study can be helpful to researchers and managers.
Article
Full-text available
Previous research and reviews on comparative advertising report mixed results. The authors report the results from a meta-analysis that examines the efficacy of comparative advertising. The analysis shows that comparative ads are more effective than noncomparative ads in generating attention, message and brand awareness, levels of message processing, favorable sponsored brand attitudes, and increased purchase intentions and purchase behaviors. However, comparative ads evoke lower source believability and a less favorable attitude toward the ad. Additional analyses of moderator variables find that market position (sponsor, comparison, and relative), enhanced credibility, message content, and type of dependent measure (relative versus nonrelative) affect some of the relationships between advertising format and cognition, brand attitudes, and purchase intentions. New brands comparing themselves to established brands appear to benefit most from comparative advertising.
Article
Full-text available
This meta-analysis examines the role of trust in marketing channels. First, the analysis of pairwise relationships involving trust indicates that trust, on average, exhibits a robust and strong relationship with other channel relationship constructs under a wide range of different conditions. Next, we explored systematic patterns of variation in the correlations. The results demonstrate that the use of experiments, samples drawn from multiple industries, and US data tend to produce larger effects than the use of field studies, samples drawn from a single industry, and European data respectively do. Various other methodological characteristics of studies did not have significant effects. Finally, we examined the role of trust in a nomological net, involving some of the most frequently studied antecedents and consequences of trust. We find that trust contributes to satisfaction and long-term orientation over and beyond the effects of economic outcomes of the relationship. Both trust and economic outcomes—not just one or the other—are conducive to relationship marketing success. q 1998 Published by Elsevier Science B.V. All rights reserved.
Article
The authors attempt to assess what has been learned from econometric models about the effect of advertising on sales. Short-term and long-term advertising response as well as model fit are analyzed for 128 econometric models involving the impact of advertising on sales. The approach, a form of meta-analysis called “replication analysis,” treats the studies as imperfect experimental replications and uses ANOVA to identify sources of systematic variation. For short-term advertising elasticities, systematic variability is found related to model specification, estimation, measurement, product type, and setting of study. For advertising carryover and model goodness of fit, the “quasi-experimental design” is so imperfect that a high degree of sharing of explained variance among explanatory factors makes it difficult to identify the impact of a particular factor. Because the studies mostly address mature products in the U.S., suggestions are made for research needs crucial to better understanding of how advertising affects sales.
Article
The service encounter frequently is the service from the customer's point of view. Using the critical incident method, the authors collected 700 incidents from customers of airlines, hotels, and restaurants. The incidents were categorized to isolate the particular events and related behaviors of contact employees that cause customers to distinguish very satisfactory service encounters from very dissatisfactory ones. Key implications for managers and researchers are highlighted.
Article
Relationship marketing—establishing, developing, and maintaining successful relational exchanges—constitutes a major shift in marketing theory and practice. After conceptualizing relationship marketing and discussing its ten forms, the authors (1) theorize that successful relationship marketing requires relationship commitment and trust, (2) model relationship commitment and trust as key mediating variables, (3) test this key mediating variable model using data from automobile tire retailers, and (4) compare their model with a rival that does not allow relationship commitment and trust to function as mediating variables. Given the favorable test results for the key mediating variable model, suggestions for further explicating and testing it are offered.
Article
Many companies consider investments in complaint handling as means of increasing customer commitment and building customer loyalty. Firms are not well informed, however, on how to deal successfully with service failures or the impact of complaint handling strategies. In this study, the authors find that a majority of complaining customers were dissatisfied with recent complaint handling experiences. Using justice theory, the authors also demonstrate that customers evaluate complaint incidents in terms of the outcomes they receive, the procedures used to arrive at the outcomes, and the nature of the interpersonal treatment during the process. In turn, the authors develop and test competing hypotheses regarding the interplay between satisfaction with complaint handling and prior experience in shaping customer trust and commitment. The results support a quasi “brand equity” perspective—whereas satisfaction with complaint handling has a direct impact on trust and commitment, prior positive experiences mitigate, to a limited extent, the effects of poor complaint handling. Implications for managers and scholars are discussed.
Article
Relationship marketing (RM) has emerged as one of the dominant mantras in business strategy circles, though RM investigations often yield mixed results. To help managers and researchers improve the effectiveness of their efforts, the authors synthesize RM empirical research in a meta-analytic framework. Although the fundamental premise that RM positively affects performance is well supported, many of the authors’ findings have significant implications for research and practice. Relationship investment has a large, direct effect on seller objective performance, which implies that additional meditated pathways may explain the impact of RM on performance. Objective performance is influenced most by relationship quality (a composite measure of relationship strength) and least by commitment. The results also suggest that RM is more effective when relationships are more critical to customers (e.g., service offerings, channel exchanges, business markets) and when relationships are built with an individual person rather than a selling firm (which partially explains the mixed effects between RM and performance reported in previous studies).