Journal of Business Research 62 (2009) 1152–1159

Contents lists available at ScienceDirect

Journal of Business Research

Understanding the psychological process underlying customer satisfaction and

retention in a relational service
Lin Guo a,1, Jing Jian Xiao b,⁎, Chuanyi Tang a,1
a
The University of Arizona, PO Box 210078, Tucson, AZ 85721-0078, USA
b
University of Rhode Island, Transition Center 202, 2 Lower College Road, Kingston, RI 02881, USA

a r t i c l e i n f o a b s t r a c t

Article history: This paper reexamines the psychological process underlying the relationship between customer satisfaction
Received 13 May 2008 and customer retention in a relational service context. It is based on a composite model of the attitude-
Accepted 20 October 2008 behavior relation, which takes both attitude toward target and attitude toward behavior into account. Data
were collected from a survey with a sample of 209 clients from a major credit counseling organization in the
Keywords: USA. Account data indicating those clients' actual retention behavior were matched with the survey data. A
Attitude toward target
two-step structural equation modeling (SEM) procedure was employed to establish the construct validity
Attitude toward behavior
Customer satisfaction
and test the hypotheses. The results revealed that the effect of customer satisfaction on retention is mediated
Customer retention by the more proximal determinants of the actual retention behavior. Specifically, customer satisfaction has
Relational service both direct and indirect effects on customers' attitudes toward remaining in the Debt Management Program
(DMP), which is the proximal predictor of customer retention. In addition, satisfaction alone cannot
guarantee customer retention. Customers' perceived behavioral control exerts direct effects on their
intention to remain and retention.
© 2008 Elsevier Inc. All rights reserved.

1. Introduction
With the increasing chances of losing current customers and the
enhanced cost of attracting new customers in today's intensely com-
petitive market, companies are devoting themselves to more strategic
efforts to retain customers (Fornell and Wernerfelt, 1987). In the
extant literature of customer retention, satisfaction has been identi-
fied as a central determinant of customer retention (Cronin et al.,
2000; Rust and Zahorik, 1993; Mittal and Kamakura, 2001; González
et al., 2007), and its effect on customer retention has been intensively
discussed within the paradigm shift from transactional marketing to
relationship marketing.
Despite the general popularity of satisfaction research, the findings
regarding the satisfaction-retention link are inconsistent. First, the
relationship between satisfaction and actual retention has been found
to be fairly weak (Hennig-Thurau and Klee, 1997; Anderson and
Sullivan, 1993). For example, Newman and Werbe (1973) found that
satisfaction can explain only 6% or 7% of the variance of retention.
Second, some recent studies (Seiders et al., 2005; Jolley et al., 2006;
Balabanis et al., 2006) uncovered the contingent nature of the
relationship, and several moderating effects of customer, relational,
and marketplace characteristics have been identified. Third, whether
⁎ Corresponding author. Tel.: +1 401 874 2547; fax: +1 401 874 4020.
E-mail addresses: lguo@email.arizona.edu (L. Guo), xiao@uri.edu (J.J. Xiao),
tang@email.arizona.edu (C. Tang).
1
Tel.: +1 520 621 1295; fax: +1 520 621 3209.
there are any mediators underlying the relationship is not clear.
Although Hennig-Thurau and Klee (1997) argued that there are
potential intervening variables such as relationship quality mediating
this relationship, only limited empirical research has dealt with this
issue. Overall, how and why customer satisfaction has an impact on
customer retention remains unclear and the underlying psychological
process by which customer satisfaction affects customer retention
deserves further research.
This paper proposes that a composite model of the attitude-
behavior relation can shed some light on understanding the psycho-
logical process underlying the customer satisfaction-retention link
and provide an alternative explanation of the mediating mechanism
behind this relationship. In this model, “attitude toward target” and
“attitude toward behavior” are both taken into consideration.
Customers' overall satisfaction with the organization, concerning
customers' evaluative beliefs of the organization, is considered as a
surrogate construct of “attitude toward organization,” and proposed
as a distal determinant of the actual retention behavior. Furthermore,
the indirect effect of a customer's satisfaction on retention is argued to
be carried through the impact of a proximal determinant, that is,
attitude toward remaining behavior (Eagly and Chaiken, 1993). In
addition, satisfaction may affect attitude toward remaining behavior
both directly and indirectly.
Compared with previous studies on satisfaction and retention, this
study includes the following unique features. First, this paper
examines the relationship between satisfaction and retention in a
relational exchange by studying the consumers who enrolled in a debt

0148-2963/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jbusres.2008.10.020
 L. Guo et al. / Journal of Business Research 62 (2009) 1152–1159
management program. Different from other service industries char-
acterized by discrete and recurring purchases such as retailing and
restaurants, credit counseling services are provided on a continuous
and contractual basis (Seiders et al., 2005). Credit counselors first help
consumers work out a comprehensive repayment plan with lower
interest rates agreed to by all their creditors. Then, credit counselors
give consumers ongoing advice to help them budget their expenses
and consistently deposit their payments each month through the
program so as to pay off their debts eventually (Hunt, 2005).
Customer retention is a critical issue in such service industries
characterized by relational exchange. Only if consumers remain in the
service can the credit counseling service organization receive “fair
share” payments from the creditor as well as collect service fees from
consumers. Second, we defined customer satisfaction as a cumulative
evaluation of a firm's performance derived from customers' prior
experiences with the firm (Anderson et al., 1994; Homburg et al.,
2005). Although in some prior studies customer satisfaction has been
conceptualized at the transaction specific level as a degree of ful-
fillment (Agustin and Singh, 2005), we believe that when the nature
of exchange is relational rather than transactional, the cumulative
overall evaluations are more likely to determine customer retention
behavior (Boulding et al., 1993). Third, unlike the majority of prior
research that uses behavioral intention as a surrogate measure of
customer retention (Gustafsson et al., 2005), this paper uses customer
account information from a service provider to measure customers'
actual retention status. This allows us to examine how satisfaction and
intention are associated with actual retention. Finally, many previous
studies used self-reported, cross-sectional data in investigating the
customer satisfaction-retention link. In contrast, this paper links
survey data including satisfaction, behavioral beliefs, attitude toward
remaining, and intention to remain to customer account data of actual
retention status, which were collected 3 months after the survey was
conducted.
2. Theoretical framework
This paper employs a composite model of the attitude-behavior
relation to reexamine the psychological process underlying the
customer satisfaction-retention link. As presented in Fig. 1, satisfaction
with the organization, viewed as attitude toward target (i.e., service
organization), influences the attitude toward remaining behavior
directly or indirectly. The indirect effect is via the mediating effect of
anticipated behavioral outcomes of remaining, such as utilitarian out-
Fig. 1. Theoretical framework (Model 1: full mediation).
1153
comes and social outcomes (i.e., normative outcomes). In turn, attitude
toward remaining, as well as normative outcomes and perceived
behavioral control, influences consumers' behavioral intention. Finally,
consumers' behavioral intention and perceived behavioral control
together have an impact on consumers' actual retention. In addition,
perceived behavioral control affects satisfaction.
This composite model of the attitude-behavior relation proposed by
Eagly and Chaiken (1993) is an extension of the theory of reasoned
action (TRA) proposed by Fishbein and Ajzen (1975) and takes both
attitude toward target and attitude toward behavior into consideration,
although at different points of a sequential process. In TRA, only
proximal causes of consumers' remaining behavior are understood in
terms of their attitude toward remaining behavior, subjective norms,
and intentions. In contrast, the current composite model also captures
the sense that remaining behavior may be affected by attitude toward
target. In this paper, customer satisfaction with the service organization
is considered as a surrogate construct to “attitude toward target,” and
the justification is provided in the following section. It is necessary and
reasonable for this study to include attitude toward target into the
attitude-behavior model because consumers' attitude toward a service
organization is very relevant to their later remaining behavior and will
exert a great influence on their remaining judgment and decision
(Snyder, 1982). However, we have to mention that this composite model
is not contradictory to the spirit of TRA since attitude toward target (i.e.,
attitude toward service organization) is proposed to be the distal cause
of remaining behavior, whose effect is mediated by attitude toward
remaining behavior, the proximal predictor in the TRA model (Eagly and
Chaiken, 1993). On the whole, this composite model follows a sequence
that attitude toward target impacts behavioral beliefs, which, in turn,
impact attitude toward behavior. In turn, attitude toward behavior im-
pacts behavioral intentions, and so on. In addition, perceived behavioral
control, a unique construct in the theory of planned behavior (Ajzen,
1991), is considered as an exogenous variable in the conceptual
framework. In the following sections, first customer satisfaction as the
attitude toward target, then the relationship between attitude toward
target and attitude toward behavior, and finally the role of perceived
behavioral control is explained.
2.1. Customer's overall satisfaction with the organization as attitude
toward target
Customers' overall satisfaction with the organization is concep-
tualized as a cumulative, global evaluation of the organization derived
1154 L. Guo et al. / Journal of Business Research 62 (2009) 1152–1159
from customers' experience with an organization over time (Homburg
et al., 2005). A person's attitude toward target is the evaluation of the
entity toward which a particular behavior is directed (Eagly and
Chaiken, 1993). In the current context, the service organization is the
target of the customer's remaining behavior. Therefore, both custo-
mers' overall satisfaction with the service organization and their
attitude toward the organization (i.e., attitude toward target) sum-
marize their evaluative beliefs about the service organization and can
be considered similar to each other to some extent. Additional evi-
dence that can support this similarity comes from the literature of
organizational behavior: an employee's satisfaction with his or her job
has already been considered as a measure of that employee's attitude
toward target (i.e., attitude toward the organization) (van Breukelen
et al., 2004). However, it is not our position in this paper to say that
satisfaction can be simply equated with attitude. Instead, when satis-
faction is conceptualized in a cumulative, global manner, the similarity
between satisfaction and attitude toward target allows us to infer that
customers' overall satisfaction can influence attitude toward behavior,
as attitude toward target does in TRA model.
2.2. The relationship between attitude toward target and attitude toward
behavior
Attitude toward behavior is different from attitude toward target.
Attitude toward behavior is the evaluation of the respondent's
engaging in a single behavior or a set of behaviors (Fishbein and
Ajzen, 1975). These attitudes necessarily name an action or a range of
actions as their objects. For example, attitude toward remaining in the
DMP specifies the behavior of remaining as its object. However, most
attitudes toward behavior may indicate an entity (e.g., a person,
group, or institution) toward which the behavior is directed. The
evaluation of this entity is called attitude toward target. For example,
customers' attitude toward remaining in the DMP specifies the credit
counseling organization that provides the DMP service as the entity
toward which the remaining behavior is directed. Hence, customers'
attitude toward this organization is called attitude toward target. As
previously mentioned, we consider customers' satisfaction with the
service organization as a reasonable proxy of attitude toward target.
In TRA, Fishbein and Ajzen (1975) assert that behavior is deter-
mined by an individual's intention, and the intention, in turn, can be
predicted by one's attitude toward behavior and subjective norms.
They argued that the components of reasoned action are the proximal
predictors of behavior, and incorporating external factors such as
attitudes toward target into TRA does not enhance the model's pre-
dictability of behavior. However, Fazio (1986) argued that one's
attitude toward target is automatically activated and highly accessible
and should affect the behavior related to the target. Furthermore,
according to the relevant principle proposed by Snyder (1982), once a
person perceives his or her attitude toward a target as pertinent to a
behavior, this attitude should lead them to have a consistent attitude
toward the behavior that directs to the target. It implies that cus-
tomers' satisfaction with a service organization contributes to
customers' favorable attitude toward remaining with the service of
the provider. By integrating these thoughts on the relationship
between attitude toward target and attitude toward behavior from
different perspectives, Eagly and Chaiken (1993) proposed an
extended model of TRA. In this model, attitude toward behavior, the
major component of TRA, is a proximal determinant of the behaviors,
while attitude toward target and all other variables are distal variables
whose effect on behavior is mediated by their impact on attitude
toward behaviors.
2.3. The role of perceived behavioral control
Beyond the composite model proposed by Eagly and Chaiken
(1993), perceived behavioral control, a unique construct in the theory
of planned behavior (TPB) (Ajzen, 1991) is also incorporated into the
model as an exogenous variable. Perceived behavioral control is
defined as one's perception of how easy or difficult it is to perform a
behavior. As posited by Ajzen (1991), perceived behavioral control has
a particular behavior as target and varies among situations. This differs
significantly with Rotter's (1966) concept of perceived locus of
control, which is a generalized expectancy that remains stable across
forms of action. Ajzen (1991) further contended that the conceptua-
lization of perceived behavioral control is most compatible with
Bandura's (1982) concept of perceived self-efficacy. This is because
one's intention to perform a behavior is constrained by both resources
and abilities one has to perform the behavior; one's self-efficacy
beliefs to perform a behavior (i.e., perceived behavioral control) will,
therefore, strongly influence one's intention and actual behavior. By
incorporating perceived behavioral control, TPB goes beyond the
boundary condition of TRA and can predict more than volitional
behaviors (Madden et al., 1992). It is proposed that perceived
behavioral control has a direct effect on customer satisfaction,
intention to remain, and actual retention behavior.
3. Hypotheses development
3.1. The validity of the theory of planned behavior in predicting customer
retention
According to TPB (Ajzen, 1991), the intention to perform a behavior
can be predicted with high accuracy by attitude toward a behavior
(i.e., one's favorable or unfavorable evaluation or appraisal of
performing the behavior), subjective norms (i.e., one's perception of
important others' approval or disapproval of performing the behavior)
and perceived behavioral control (i.e., one's perception of the
accessibility of resources and opportunities necessary for performing
the behavior). These intentions, together with perceived behavioral
control can, in turn, affect actual behavior. Since this theory has been
widely applied in various settings in business literature, such as e-
commerce adoption (Nasco et al., 2008) and intentions to start a
business (Carr and Sequeira, 2007), we propose that this theory can
also be applied to predict customers' remaining behavior in credit
counseling service (i.e., DMP).
H1a. Customers' intention to remain has a positive influence on cus-
tomers' actual retention.
H1b. Perceived behavioral control has a positive influence on cus-
tomers' actual retention.
H1c. Attitude towards remaining has a positive influence on cus-
tomers' intention to remain.
H1d. Anticipated normative outcomes associated with remaining
have a positive influence on customers' intention to remain.
H1e. Perceived behavioral control has a positive influence on cus-
tomers' intention to remain.
3.2. The psychological process underlying customer satisfaction and
attitude toward remaining
As discussed in the theoretical framework, one's attitude toward
target (i.e. satisfaction with the organization) is automatically
activated and comes to mind before attitudes toward behaviors in
which one might engage. Furthermore, attitude toward target is a
distal predictor of behavior whose effect is mediated by attitude
toward behavior. It implies that attitude towards target may have a
positive influence on attitude toward behavior. As suggested by the
relevance principle proposed by Snyder (1982), as long as one's
behavior is relevant to one's attitude toward target, attitude towards
target should affect attitude toward behavior. In the current research
 L. Guo et al. / Journal of Business Research 62 (2009) 1152–1159
context, customers' continued membership in DMP is a way of ex-
pressing their satisfaction with the current counseling organization's
service. That is, customers' satisfaction with the counseling service
provider has behavioral implications and is relevant to their remaining
behavior. Therefore, customers' satisfaction with a service organiza-
tion has a positive effect on customers' attitude toward remaining
with the service organization.
Eagly and Chaiken (1993) proposed in their extended model of
TRA that attitude toward target may influence the attitude toward
behavior either directly or indirectly. On the one hand, attitudes
toward targets could directly impact attitudes toward behaviors via a
spontaneous or less deliberative route. A behavior would seem
attractive merely because the attitude toward the target comes to
mind, and there would be little or no consideration of the potential
behavioral outcomes (Fazio, 1986). On the other hand, the effect of
attitude toward target on attitude toward behavior can be carried
through its impact on the anticipated behavioral outcomes. Eagly and
Chaiken (1993) suggested that there are different kinds of behavior
outcomes, including utilitarian and normative outcomes. Since
consumers' attitude toward target biases their perception of any
behaviors toward that target, if a customer is satisfied with a credit
counselor (attitude toward target), the customer's perception of
future utilitarian outcomes through remaining in the current credit
counseling program will increase. For example, remaining in the
current credit counseling organization may help consumers alleviate
financial pressure and regain control of their personal finances
through the consumption of the satisfactory service. The same applies
to normative outcomes associated with the approval and disapproval
of the remaining behavior by significant others.
H2a. Customers' satisfaction with the organization has a positive
influence on customers' attitude toward remaining.
H2b. Customers' satisfaction with the organization has a positive
influence on customers' anticipated utilitarian outcomes of remaining.
H2c. Customers' satisfaction with the organization has a positive
influence on customers' anticipated normative outcomes of remaining.
Customers' anticipated utilitarian outcomes of remaining in the
program are the perceived future positive utilitarian consequences of
the behavior. In Fishbein and Ajzen's (1975) TRA model, utilitarian
outcomes associated with the behavior are expressed in terms of
behavioral beliefs. From the perspective of the information-processing
approach, beliefs are formed by evaluating certain attributes of
performing the behavior (Ajzen, 1991). People will favor the behaviors
that they believe have positive attributes and vice versa.
H3a. Customers' anticipated utilitarian outcomes associated with
remaining have a positive influence on customers' attitude toward
remaining.
On the other hand, a customer's anticipated normative outcomes
of remaining refer to an individual's perception of significant others'
opinions on whether or not he or she should perform a particular
behavior. In Fishbein and Ajzen's (1975) TRA model, normative
outcomes associated with the behavior are expressed in terms of
subjective norms. Obviously, subjective norms that pertain to the
approval or disapproval that important others are expected to convey
regarding the behavior are one kind of normative outcome associated
with the intended behavior. Eagly and Chaiken (1993) suggest that a
customer's anticipated normative outcomes of behavior will have a
positive impact on attitude toward behavior.
H3b. Customers' anticipated normative outcomes associated with
remaining will have a positive influence on customers' attitude to-
ward remaining.
1155
3.3. The influence of perceived behavioral control on satisfaction with the
organization
Perceived behavioral control refers to peoples' perceived ease or
difficulty of performing the behavior of interest (Ajzen and Madden,
1986). Westbrook (1980) has found that individuals who have higher
competence and efficacy in dealing with the environment tend to
evaluate objects in a more favorable way, thus having a significantly
higher level of satisfaction than those with lower competence and
efficacy. When a person perceives that it is easy to remain in the
service, his or her distress from financial trouble is lessened to some
extent. Thus the person's attitude toward this organization would be
more favorable compared with the attitudes of those who have more
difficulty following the guidance of the counselors.
H4. Customers' perceived behavioral control regarding remaining has
a positive influence on customers' satisfaction with the organization.
3.4. The full mediating effect of TRA components
Although in the existing literature on customer retention, some
studies (e.g., Beatty and Smith, 1987; Bolton et al., 2000) have found a
direct effect of satisfaction on customers' actual retention behavior or
intention to remain, Fishbein and Ajzen (1975) have argued that
attitudes toward behaviors (attitude toward remaining) from the theory
of reasoned action is a more proximal predictor of behavior than
attitudes toward target towards which the behavior is directed
(satisfaction with the organization). They argued that if attitude toward
target does have an impact on behavior, the effect should be indirect
through the mediating effects of the major components of TRA. This
argument indicates that the components of TRA fully mediate the effects
of attitude toward target (satisfaction with the organization) on actual
behavior (retention). In other words, satisfaction can explain only
limited variance of retention when the effects of the components of the
theory of planned behavior are counted.
H5. The effect of customer satisfaction on actual retention behavior is
fully mediated by the variables proposed in the model.
4. Methodology
4.1. Sample
The clients of a national credit counseling organization who were
enrolling in a debt management program were invited to participate
in this research. Due to privacy concerns, the firm did not provide the
client contact information to the researchers; rather, the firm helped
to issue a recruiting announcement of this study in their monthly
statement, and interested clients were requested to contact the
investigators directly via email, telephone, or fax. Ten dollars was
provided as an incentive for each respondent. Three hundred and
twenty-six people volunteered for this study. Of all the 220 returned
surveys, 11 were found incomplete, leaving us a total of 209 usable
surveys. The overall response rate after adjusting for incomplete
surveys was 64%.
4.2. Measurement
Except for the construct of actual retention, all of the other
variables employed multi-item measures adapted to the context of
this study from existing literature. Each of them was measured with a
7-point semantic differential or Likert scale. Table 1 presents the
constructs, measures, and Cronbach's alpha. All the alpha levels were
above.70 as recommended by Nunnally (1978), and therefore the
reliability was adequate.
1156 L. Guo et al. / Journal of Business Research 62 (2009) 1152–1159

Table 1
Measurement.

Construct Items Measures Scale Source

Satisfaction α = .901
Anticipated utilitarian
Outcomes toward remaining α = .826
Attitude toward remaining α = .748
Anticipated normative
Outcomes toward remaining α = .842
Perceived behavioral control α = .763
Intention to Remain α = .828
SATIS1 I am satisfied with DMP's service. 1 = extremely disagree, Hennig-Thurau and Klee (1997)
7 = extremely agree
SATIS2 I rate my relationship with DMP as: 1 = extremely bad,
7 = extremely good
BENE1 My being in DMP would help me stop using credit cards. 1 = extremely unlikely, Hilgert et al. (2003); Xiao et al.
7 = extremely likely (2006)
BENE2 My being in DMP would help me reduce unnecessary spending. 1 = extremely unlikely,
7 = extremely likely
BENE3 My being in DMP would help me develop a plan for my 1 = extremely unlikely,
financial future. 7 = extremely likely
BENE4 My being in DMP would help me set up a plan for my family's 1 = extremely unlikely,
future financial needs. 7 = extremely likely
ATT1 My being in the debt management program (DMP) to reduce 1 = extremely foolish, Ajzen (1991)
debt is: foolish-wise 7 = extremely wise
ATT2 My being in DMP to reduce debt is: bad-good 1 = extremely bad,
7 = extremely good
ATT3 My being in DMP to reduce debt is: harmful-beneficial 1 = extremely harmful,
7 = extremely beneficial
NORM1 Most members of my family think I should be in DMP 1 = extremely unlikely, Hilgert et al. (2003); Xiao et al.
to reduce debt. 7 = extremely likely (2006)
NORM2 My close friends think I should be in DMP to reduce debt. 1 = extremely unlikely,
7 = extremely likely
NORM3 My colleagues think I should be in DMP to reduce debt. 1 = extremely unlikely,
7 = extremely likely
CONTR1 Please rate the difficulty of staying in DMP. 1 = extremely easy, Ajzen (1991)
7 = extremely difficult
CONTR2 If I wanted to, I could easily stay in DMP. 1 = extremely false,
7 = extremely true
INTENT1 My intention to staying DMP to reduce debt is: weak-strong 1 = extremely weak, Ajzen (1991)
7 = extremely strong
INTENT2 I will stay in DMP to reduce my debt till it ends. 1 = extremely unlikely,
7 = extremely likely

We used a client's current status in the program as a proxy measure of
actual retention. The information was obtained from the credit
counseling organization's account data 3 months after the survey data
were collected. The status of a client recorded the extent to which a client
fulfilled their payment on time and in full and, to some extent,
represented the actual remaining behavior of a client. This is possible
because if a person intends to remain in the program, he or she has to
behave in a certain way (i.e., make the payments in full and on time) in
order to avoid being excluded from the program after several late or non-
payments. In other words, one's appropriate payment behavior, a logical
consequence following one's intention to remain in the program, is
equivalent to one's remaining behavior in the current context. Specifi-
cally, if a client remains in the program until completion, the status is
coded as 6. If a client never misses any payment and remains active in the
DMP account, the status is coded as 5. When a client misses more than
one payment or is late more than twice, the status is coded as 4. If a client
misses more than two payments or is late more than four times, the
status is coded as 3. If a client's account is closed because of continuous
non-payments, the status is coded as 2. If a client's account is closed at the
client's request due to unwillingness to pay, the status is coded as 1.

Fig. 2. Partial mediation (Model 2).

 L. Guo et al. / Journal of Business Research 62 (2009) 1152–1159 1157

5. Analysis and results Table 3

Measurement model results.

In the data analysis, two models were tested and compared. The Construct Items Estimates of Standardized p level
first is the basic model used to test the relationship between customer regression estimates of
satisfaction, customer retention, and the mediating effects of TRA weights regression weights

components (See Fig. 1). The second model presents an alternative Satisfaction α = .901 SATIS1 1.021 .855 b .001
SATIS2 1.004 .972 b.001
within which the direct paths from satisfaction to intention to remain
Anticipated utilitarian BENE1 .784 .608 b.001
and actual retention are added (See Fig. 2). A nested model com- Outcomes α = .826 BENE2 .722 .605 b.001
parison was conducted to examine whether the effect of satisfaction is BENE3 1.156 .861 b.001
fully mediated by TRA components proposed in the first model. BENE4 1.079 .755 b.001
The two-step procedure of structural equation modeling proposed Attitude toward ATT1 1.247 .831 b.001
remaining α = .748 ATT2 1.074 .733 b.001
by Anderson and Gerbing (1988) was used to test the model and ATT3 .986 .681 b.001
hypotheses. The measurement model was established by conducting a Anticipated normative SUBNORM1 .947 .795 b.001
confirmatory factor analysis, and the structural model analysis was Outcomes α = .842 SUBNORM2 .667 .777 b.001
conducted to test the relationship among the constructs. AMOS 6.0 SUBNORM3 .848 .608 b.001
Perceived behavioral CONTR1 1.594 .885 b.001
was employed to conduct the analysis, and raw data were used as
control α = .763 CONTR2 1.123 .703 b.001
inputs. Intention to remain INTENT1 .860 .815 b.001
α = .828 INTENT2 .966 .872 b.001
5.1. Measurement model testing and results χ2(86) = 101.872, p = .116; NFI = .938; CFI = .990; IFI = .990; RMSEA = .030.

The descriptive statistics and correlation is summarized in Table 2.

The CFA results for overall model fit were χ2(86) = 101.872, p = 0.116;
CFI = .990; NFI = .938, IFI = .990; RMSEA = .030. These indices were
acceptable (Bollen, 1989). The convergent validity of constructs was
assessed by examining the indicator loadings: factor loadings of the
indicators for each construct were statistically significant and
sufficiently high enough (See Table 3). In addition, reliabilities were
adequate. Thus convergent validity was achieved. To examine
discriminant validity, a pairwise chi-square difference test was
conducted for each pair of constructs in the overall model (Bagozzi
et al., 1991). Discriminant validity is supported if an unconstrained
model (in which the correlation between the two constructs is freely
estimated) demonstrates significantly better fit than a constrained
model (in which the correlation between the two constructs is
constrained to 1). For 15 out of all 15 cases, the chi-square of un-
constrained model, compared with that of constrained model,
significantly decreased and indicated a better model fit. Therefore,
discriminant validity for the constructs was achieved. Since model 1
and model 2 are differentiated from each other only by the direct
paths from satisfaction to intention to remain and retention, they
share the same measurement model.
organization has a significant influence on customers' attitude toward
remaining, anticipated utilitarian and normative outcomes of remain-
ing in the DMP, which are consistent with H2a, H2b, and H2c. Both
anticipated utilitarian and normative outcomes have a positive
influence on consumers' attitude toward remaining, thus supporting
H3a and H3b. Customers' perceived behavioral control has a signif-
icant impact on customers' satisfaction, supporting H4. Consistent
with TPB, consumers' attitude toward remaining and consumers'
perceived behavioral control of remaining positively affect consumers'
intention to remain, which lends support to H1c and H1e. Further-
more, intention to remain and perceived behavioral control signifi-
cantly influence actual retention behavior and H1a and H1b are
supported. Anticipated normative outcomes do not have a significant
impact on consumer's intention to remain, which is contradictory to
H1d. The failure to support this hypothesis may imply that the
relationship between anticipated normative outcomes and intention
to remain is fully mediated by attitude toward remaining, since
attitude is measured at a global level in the current study.
To test whether the effect of satisfaction on retention is fully
mediated by the variables proposed in the model, a nested model

5.2. Overall structural model and hypotheses testing

Table 4
After confirmatory factor analysis, the structural models were Tests of the hypotheses.
established and the hypotheses were tested. For model 1, the results
Path Path Standardized path p level
for the structural model were χ2(107) = 137.571, p = .025, NFI = .919, coefficients coefficients
CFI = .980, IFI = .981, RMSEA = .037. The overall model fit was good. H1a: intention to remain →retention .456 .337 b.001
All hypotheses were tested by examining path significance in model 1 H1b: perceived behavioral control → .238 .184 = .018
(see Table 4). Except for H1d, all the other hypotheses were supported retention
at the .05 level (two-tailed). Customers' overall satisfaction with the H1c: attitude toward remaining→ .521 .486 b.001
intention to remain
H1d: anticipated normative outcomes→ .079 .099 = .178
intention to remain
Table 2 H1e: perceived behavioral control → .264 .277 b.001
Descriptive statistics and intercorrelations for the constructs. intention to remain
H2a: satisfaction with the organization → .332 .370 b.001
1 2 3 4 5 6 M SD attitude toward remaining
Satisfaction 1 6.28 1.06 H2b: satisfaction with the organization→ .690 .651 b.001
Anticipated utilitarian .53⁎⁎ 1 5.80 1.07 anticipated utilitarian outcomes
outcomes H2c: satisfaction with the organization→ .309 .259 b.001
Attitude toward remaining .52⁎⁎ .51⁎⁎ 1 6.23 .96 anticipated normative outcomes
Perceived normative .25⁎⁎ .32⁎⁎ .34⁎⁎ 1 4.63 1.30 H3a: anticipated utilitarian outcomes → .353 .417 b.001
outcomes attitude toward remaining
Perceived behavioral .31⁎⁎ .17⁎ .25⁎⁎ − .018 1 5.19 1.54 H3b: anticipated normative outcomes → .102 .135 = .048
control attitude toward remaining
Intention to remain .42⁎⁎ .34⁎⁎ .40⁎⁎ .22⁎⁎ .33⁎⁎ 1 6.53 .99 H4: perceived behavioral control→ .348 .351 b.001
satisfaction with the organization
⁎Correlation is significant at the 0.05 level (2-tailed).
⁎⁎Correlation is significant at the 0.01 level (2-tailed). χ2(107) = 137.571, p = .025, NFI = .919, CFI = .980, IFI = .981, RMSEA = .037.
1158 L. Guo et al. / Journal of Business Research 62 (2009) 1152–1159
comparison was conducted between model 1 (full mediation model)
and model 2 (partial mediation model). For model 2, the results for the
structural model were χ2(105) = 134.939, p = .001, NFI = .921, CFI = .981,
IFI= .981, RMSEA = .037. Comparing model 2 with model 1, there is no
significant difference in their overall model fit (Δχ2 = 2.623, Δdf = 2,
p N .001) and only NFI and CFI increased slightly in model 2. However,
in model 2, the added direct paths from satisfaction to intention to
remain (p = .102) and from satisfaction to retention (p = .953) are
both insignificant. Thus, we can conclude that the effects of satisfaction
on customer retention are fully mediated and H5 is supported.
6. Conclusion and discussion
The findings of this paper provide important theoretical and
managerial implications. First, the results of this study indicate that
the extended model of TRA can be applied to understand the
mechanism underlying the relationship between customer satisfac-
tion and retention, which sheds light on the inconsistent empirical
results of this relationship in the literature. The results of this study
show that the effect of customer satisfaction is mediated by the more
proximal determinants of the actual retention behavior, that is,
anticipated outcomes of remaining behavior and attitude toward
remaining behavior.
Secondly, this paper provides empirical support for both direct and
indirect effects of attitude toward target (i.e., satisfaction with the
organization) on attitude toward behavior (i.e., attitude toward
remaining in DMP) proposed by Eagly and Chaiken (1993) in their
extended model of TRA. As has been discussed above, on the one hand, a
favorable attitude toward target can directly lead to a favorable attitude
toward the focused behavior that directs to the target without
considering the potential behavioral outcomes. On the other hand, if
customers are satisfied with the service provider, their anticipated utility
from this particular service provider, such as utilitarian outcome and
social outcome, will increase, thereby indirectly inducing their favorable
attitude toward remaining. The underlying mechanism of the process
indicates that besides increasing customers' satisfaction, managers have
to educate customers to recognize the possible benefits of remaining in
the DMP so that customers will develop a positive attitude toward
remaining in the DMP, thereby inducing their remaining behavior.
Third, customers' perception of ease or difficulty of staying in the
DMP plays a critical role in customer retention. It has direct effects on
customers' intention to remain and actual retention. Although satis-
faction is the necessary premise of customer retention, it alone is not
sufficient to successfully retain customers. Even if the customers are
very satisfied, their perceived behavioral control can be very limited so
that they may not actually stay in the program. According to these
findings, credit counselors should help their clients enhance their
feeling of self-efficacy and controllability regarding remaining in the
DMP in order to increase their intention to stay as well as the likelihood
of actual retention. In addition, customers' perceived behavioral con-
trol has a positive effect on customers' overall satisfaction with the
organization. It means that once credit counselors help their customers
enhance their perception of self-efficacy in remaining in the DMP,
customers will evaluate the organization in a more favorable way. This
finding is very important, not only for credit counseling industries but
also for other service contexts that require customer compliance and
cooperation, such as long-term financial planning, health care, and
preventive auto maintenance (Dellande et al., 2004). For example,
DMP is not so much different from the service provided by a weight-
loss program in the health care industry because the basic nature of
DMP is to provide counseling services to consumers and consumers
have to adopt counselors' advices, apply them in their daily life and
participate in the process of service delivery.
Overall, the findings of this paper contribute to the understanding
of the satisfaction-retention link by overturning the inappropriate
belief of an immediate and strong impact of customer satisfaction on
retention, and broaden the simplistic view of the relationship by
proposing the complexity of the underlying mediating mechanism. In
addition, this paper contributes to the attitude-behavior literature
by providing empirical support for the relationship between attitude
toward target and attitude toward behavior proposed by Eagly and
Chaiken (1993).
7. Limitations and future research
By employing a composite model of the attitude-behavior relation,
this paper provided a theoretical explanation for the psychological
process underlying the satisfaction-retention link. However, the incon-
sistent results of the satisfaction-retention link in existing literature
indicate that the underlying process may be more complex than what is
proposed here. Further research is needed to confirm the explanation
proposed by this study. Next, several limitations of this study may affect
the generalizability of its results. Data were collected from the customers
of one single credit counseling service organization. To provide a general
solution, future research should try to collect data across various credit
counseling organizations. Moreover, although the current service
context shares commonalities with other service contexts that require
customer compliance and cooperation, future research is needed to
confirm the generalizability of our findings to other service contexts. In
addition, although the outcome variable of actual remaining behavior
was measured 3 months later after the survey was conducted, the data
on satisfaction and other major psychological process variables were
collected through a cross-sectional survey. A three-wave longitudinal
study design is needed to assess the causal process underlying cus-
tomer's overall satisfaction and retention. Finally, to understand the
boundary conditions of the findings in this paper, future research could
investigate potential moderators for the relationships proposed in this
study. For example, how some consumers' individual characteristics,
such as consumers' locus of control, moderate the relationships in the
model could be explored.
References
Agustin C, Singh J. Curvilinear effects of consumer loyalty determinants in relational
exchanges. J Mark Res 2005;42:96–108. (February).
Ajzen I. The theory of planned behavior. Org Behav Human Deci Process 1991;50:
179–211.
Ajzen I, Madden TJ. Prediction of goal-directed behavior: Attitudes, intentions, and
perceived behavioral control. J Exp Soc Psychol 1986;22:453–74.
Anderson JC, Gerbing DW. Structural equation modeling in practice: a review and
recommended two-step approach. Psychol Bull 1988;103(3):411–23.
Anderson EW, Sullivan MW. The antecedents and consequences of customer satis-
faction for firms. Mark Sci 1993;12(2):125–43.
Anderson EW, Fornell C, Lehmann DR. Customer satisfaction, market share, and
profitability: finding from Sweden. J Mark 1994;58:53–66.
Bagozzi RP, Yi Y, Phillips LW. Assessing construct validity in organizational research.
Adm Sci Q 1991;36(3):421–58.
Balabanis G, Reynolds N, Simintiras A. Bases of e-store loyalty: perceived switching
barriers and satisfaction. J Bus Res 2006;59(2):214–24.
Bandura A. Self-efficacy mechanism in human agency. Am Psychol 1982;37:122–47.
Beatty SE, Smith SM. External search effort: an investigation across several product
categories. J Consum Res 1987;14:183–95. (June).
Bollen KA. Structural equations with latent variables. New York: John Wiley & Son;
1989.
Bolton RN, Kannan PK, Bramlett MD. Implications of loyalty program membership and
service experience for customer retention and value. J Acad Mark Sci 2000;28: 95–
108. (January).
Boulding W, Kalra A, Staelin R, Zeithaml VA. A dynamic process model of service quality:
from expectations to behavioral intentions. J Mark Res 1993;30(1):7–27.
Carr JC, Sequeira JM. Prior family business exposure as intergenerational influence and
entrepreneurial intent: a theory of planned behavior approach. J Bus Res, 2007;60
(10):1090–8.
Cronin JJ, Brady MK, Hult GTM. Assessing the effects of quality, value, and customer
satisfaction on consumer behavioral intentions in service environments. J Retail
2000;76(2):193–215.
Dellande S, Gilly MC, Graham JL. Gaining compliance and losing weight: the role of the
service provider in health care services. J Mark 2004;68:78–91. (July).
Eagly A, Chaiken S. The psychology of attitudes. Fort Worth, TX: Harcourt Brace
Jovanovich; 1993.
Fazio RH. How do attitudes guide behavior? In: Sorrentino RM, Higgins ET, editors. The
handbook of motivation and cognition. New York: Guilford Press; 1986. p. 204–43.
 L. Guo et al. / Journal of Business Research 62 (2009) 1152–1159
Fishbein M, Ajzen I. Belief, attitude, intention and behavior: An introduction to theory
and research. Reading, Mass: Addison-Wesley; 1975.
Fornell C, Wernerfelt B. Defensive marketing strategy by customer complaint manage-
ment: a theoretical analysis. J Mark Res 1987;24(4):337–46.
González MEA, Comesaña LR, Brea JAF. Assessing tourist behavioral intentions through
perceived service quality and customer satisfaction. J Bus Res 2007;60(2):153–60.
Gustafsson A, Johnson MD, Roos I. The effects of customer satisfaction, relationship com-
mitment dimensions, and triggers on customer retention. J Mark 2005;69(October):
210–8.
Hennig-Thurau T, Klee A. The impact of customer satisfaction and relationship quality
on customer retention: a critical reassessment and model development. Psychol
Mark 1997;14(8):737–64.
Hilgert MA, Hogarth JM, Beverly SG. Household financial management: the connection
between knowledge and behavior. Fed Reserve Bull 2003;89(7):309–22.
Homburg C, Koschate N, Hoyer WD. Do satisfied customers really pay more? A study of
the relationship between customer satisfaction and willingness to pay. J Mark
2005;69:84–96 (April).
Hunt RM. Whither consumer credit counseling? Bus Rev 2005;Q4:9–20.
Jolley B, Mizerski R, Olaru D. How habit and satisfaction affects player retention for
online gambling. J Bus Res 2006;59(6):770–7.
Madden TJ, Ellen PS, Ajzen I. A comparison of the theory of planned behavior and the
theory of reasoned action. Pers Soc Psychol Bull 1992;18(1):3–9.
Mittal V, Kamakura W. Satisfaction, repurchase intent, and repurchase behavior: inves-
tigating the moderating effects of customer characteristics. J Mark Res 2001;38:
131–42. (February).
1159
Nasco SA, Toledo EG, Mykytyn Jr PP. Predicting electronic commerce adoption in Chilean
SMEs. J Bus Res 2008;61(6):697–705.
Newman JW, Werbe RA. Multivariate analysis of brand loyalty for major household
appliances. J Mark 1973;10:404–9.
Nunnally JC. Psychometric theory. New York: McGraw-Hill Inc; 1978.
Rotter JB. Generalized expectancies for internal versus external control of reinforce-
ment. Psychological Monographs 1966;80 (1, Whole No. 699).
Rust RT, Zahorik AJ. Customer satisfaction, customer retention, and market share. J Retail
1993;69(2):193–215.
Seiders K, Voss GB, Grewal D, Godfrey AL. Do satisfied customers buy more? Examining
moderating influences in a retailing context. J Mark 2005;69:26–43. (October).
Snyder M. When believing means doing: creating links between attitudes and behavior.
In: Zanna M, Higgins E, Herman C, editors. Consistency in social behavior: The
Ontario symposium, vol. 2. Hillsdale, NJ: Lawrence Erlbaum; 1982. p. 105–30.
van Breukelen W, van der List R, Steensma H. Voluntary employee turnover: combining
variables from the 'traditional' turnover literature with the theory of planned
behavior. J Org Behav 2004;25(7):893–915.
Westbrook RA. Consumer satisfaction as a function of personal competence/efficacy.
J Acad Mark Sci 1980;8:427–37.
Xiao JJ, Sorhaindo B, Garman ET. Financial behaviors of consumers in credit counseling.
Inter J Consum Stud 2006;30(2):108–21.