Professional Documents
Culture Documents
Employees' Decision Making in The Face of Customers' Fuzzy Return Requests
Employees' Decision Making in The Face of Customers' Fuzzy Return Requests
T
Keywords: fuzzy requests, compliance, script theory, role theory, motivated reasoning, retail returns, legitimacy
he marketing literature has long recognized the impor- a table with a view. While some requests may be innocent
tant roles of frontline employees in service delivery problems (e.g., returning a failed product bought from the
and recovery (e.g., Bitner, Booms, and Tetreault store), others may involve intentionally taking advantage of
1990; Schneider 1991). According to Langeard et al.’s a service provider (e.g., returning a dress after wearing it for
(1981) Servuction Service Model, the quality of service a special occasion [e.g., Schmidt et al. 1999]).
performance depends on the interaction and performance of In contrast to both the common organization assumption
three key players in the delivery process: the service organi- that customers are always right (e.g., Ringberg, Odekerken-
zation system, the contact employee, and the customer. In Schroder, and Christensen 2007) and the emphasis in cur-
this relationship, customers are motivated to compete with rent literature on problem or fraudulent customers, in which
the firm and its employees to seize control of the service customers or their actions are completely wrong (e.g., Har-
process to increase their utility and personal satisfaction ris 2010), many customers’ requests lie between these
(Bateson 1985). One way customers do this is by asking for extremes. These requests often are somewhat or slightly
special requests that require the employee or the firm to outside company policy, not addressed by company policy,
adapt the service to these requests (Bitner, Booms, and or outside the employee’s job description or job expecta-
Tetreault 1990). tions, but they are not unacceptable or detrimental to the
Some requests greatly deviate from the normal service company. We label these requests as “fuzzy requests,”
scope and employee expectations, while others are routine, implying that how and what employees should do regarding
such as a customer asking the host to seat his or her party at these requests are not always clear.
Although the marketing literature has not identified or
examined employees’ reactions to customers’ fuzzy requests,
their occurrence is unquestionable; many excellent service
Sijun Wang is Associate Professor of Marketing, Department of Marketing
Emp
Employee
loyee F
Factors
actors Compliance
Compliance Be
Behaviors
haviors
• Customer Orientation
• Conflict Avoidance
H1
Compliance
Process
C
Customer
ustomer IInteraction
nteraction S
Styles
tyles
H2
Affiliation
R equest Le
Request gitimacy
Legitimacy
Dominance
H3
• Regulative
O rganizational F
Organizational actors
Factors H4a-c •H3 Normative H5 Compliance
• Cognitive Outcome
H6
• Norm of Flexibility
• !
Punishment Expectations
employees with higher CO will work harder to please the such as attentiveness, relaxed, open, and friendly styles. In
customer, regardless of the compliance outcome. the context of an employee–customer interaction, “affilia-
tive style” refers to the employee’s perception that a cus-
H1a: There is a positive relationship between an employee’s
CO and his or her friendliness and effort in handling the tomer engages with him or her in a warm, friendly manner.
fuzzy request (i.e., compliance process behaviors). In contrast, “dominant style” refers to the employee’s per-
ception that a customer is trying to control or dominate the
Conflict avoidance (CA). Researchers in psychology interaction. Emotional contagion theory (Hatfield, Cacioppo,
have extensively studied CA, which is an attempt to protect and Rapson 1994) proposes a direct link between employee
the self from conflict, disapproval, and negative attention and customer by affective transfer during interpersonal con-
(Rahim 1983). People who have a CA personality prefer not tact. In the service encounter literature, Dallimore, Sparks,
to assert themselves to preserve rapport and smooth rela- and Butcher (2007) demonstrate that angry customers often
tionships with others (Schroeder 1965). Oddly enough, produce similar facial displays and affective states in ser-
although marketers have studied group CA in the context of vice workers through an emotional contagion process. In an
marketing strategy decision making (Atuahene-Gima and even broader sense, each party of the client–service
Murray 2004), researchers have not linked individual provider dyad influences the other party’s behavior (Ma and
employees’ CA to their handling of customer complaints or Dubé 2011). For example, Webster (2005) finds that a cus-
requests (including return requests). Yet in our qualitative tomer evaluates an employee with a more affiliative style
interviews, informants frequently talked about trying to more favorably and is more satisfied, whereas an employee
avoid arguments with customers if possible. Thus, a posi- displaying more dominance receives less favorable cus-
tomer evaluations (Korsch and Negrete 1972). While cus-
tive relationship between an employee’s tendency to avoid
tomer interaction styles have not been associated with
conflicts and his or her compliance process is likely:
employee reactions to customers, given the dyadic, interac-
H1b: There is a positive relationship between an employee’s tive nature of customer–employee dyads, the findings
CA and his or her friendliness and effort in handling the between employee interaction styles and customer evalua-
fuzzy request (i.e., compliance process behaviors). tions should work in reverse. That is, the employee is likely
to more positively handle the request (i.e., compliance
Customer Interaction Styles process) when the customer approaches him or her with a
The social interaction model (Ben-Sira 1980) identifies two friendly, warm manner.
major interaction styles: affiliation and dominance. An affil- H2: The employee’s perceptions of the customer’s affiliative
iative style parallels Norton’s (1978) communication styles, style are related positively to his or her friendliness and
Beatty 2008). To aid in more vivid recall, respondents same pattern of fixed and free parameters across the two subsam-
ples. This model fit the data well (2 = 1122.66, d.f. = 760; com-
described the reason for the return, price of the returned parative fit index [CFI] = .97, nonnormed fit index [NNFI] = .96,
item, and time the incident occurred. The top three cate- and root mean square error of approximation [RMSEA] = .048).
gories of inappropriate customer return requests were “no Next, we estimated a model that constrained the factor loadings to
receipt” (22%), “beyond the return period” (20%), and “box/ be invariant across groups and found the fit indexes for this model
packaging opened or missing” (10%). In terms of the item to be acceptable (2 = 1149.71, d.f. = 780; CFI = .97; NNFI = .96;
value involved, 35% were valued at less than $25, 26% and RMSEA = .048), with the difference in fit between this model
and the baseline measurement model insignificant (2 = 27.05,
were $25–$50, 20% were $51–$100, and the rest were d.f. = 20, p = .13). Thus, the loadings for all measurement items
higher than $100. With regard to recency, 53% reported that were invariant across samples. The third model estimated specified
the incident had occurred in the past month, 25% of the invariant factor variance and covariances in addition to invariant
incidents occurred in the past one to three months, and the factor loadings. It has acceptable fit indexes (2 = 1188.91, d.f. =
rest more than three months previously. Respondents had 845; CFI = .96; NNFI = .96; and RMSEA = .051), and the differ-
no difficulty recalling such a return request incident. ence in fit between this model and the invariant factor model is not
significant (2 = 39.20, d.f. = 65, p = .99). Thus, all correlational
Measures (covariance) relationships among model constructs, as well as
construct (factor) variances, are equal across samples. Finally, we
Appendix B presents the scales measuring the constructs estimated a fully invariant measurement model (i.e., invariant fac-
central to the study. We achieved content validity through tor loadings, invariant factor variance and covariances, and invari-
an extensive literature search and a thorough analysis of ant item error loadings). Although the indexes indicate adequate fit
(2 = 1447.68, d.f. = 875; CFI = .957; NNFI = .95; and RMSEA =
interview scripts, which allowed us to define the domains of
.057), this model’s fit differed from the fit of the covariances’
constructs of interest and to generate the items. In addition, a measurement model (2 = 258.77, d.f. = 30, p < .001). Still, the
pretest with a convenience sample of retail employees (N = invariance of the factor loadings, factor variances, and covariances
430) provided a thorough assessment of the scales and suggest that pooling the two subsamples is appropriate.
TABlE 1
Correlations and Descriptive Statistics
Variables 1 2 3 4 5 6 7 8 9 10 11
1. Compliance process .76 .12** .04 .02 –.05 –.01 –.02 .09 .04 .44 .32**
2. Compliance outcome .13** — .36** –.40** .33** .46** .40** .24** –.18** .06 –.02
3. Affiliation .05 .37** .96 –.65** .16** .33** .45** .06 –.09 .00 –.07
4. Dominance .03 –.39** –.63** .89 –.23** –.42** –.48** –.05 .05 .07 .07
5. Regulative legitimacy –.04 .34** .17** –.22** .93 .32 .17** .00 .08 –.18** –.10
6. Normative legitimacy .01 .47** .34** –.41** .33** .89 .61** .00 .00 –.05 –.11*
7. Cognitive legitimacy –.01 .41** .46** –.47** .18** .61** .88 –.04 .02 .00 –.06
8. Norm of flexibility .10 .25** .07 –.04 .01 .01 –.03 .88 –.40** .16** .17**
9. Expectations of punishment .05 –.17** –.08 .06 .09 .01 .03 –.39** .96 –.03 .08
10. Customer orientation .45** .07 .01 .08 –.17** –.04 .01 .17** –.02 .90 .42**
11. Conflict avoidance .33** –.01 –.06 .08 –.09 –.10* –.05 .18** .09 .43** .87
12. MV (fairness) –.11* –.10* –.03 –.01 .05 .09 .01 –.34** .23** –.23** –.18**
M 5.71 3.62 4.47 4.26 2.49 2.68 2.98 4.97 3.94 6.34 5.65
SD .84 1.41 1.72 1.62 1.52 1.24 1.19 1.35 1.78 .81 1.07
*p < .05.
**p < .01.
Notes: Zero-order correlations are below the diagonal; adjusted correlations for potential common method variances (Lindell and Whitney
2001) are above the diagonal. Bold numbers in the diagonal are average variance extracted for constructs.
6The first model estimated was a baseline measurement model RMSEA = .047), and the difference in fit between this model and
of the same pattern of fixed and free parameters. This model fit the the invariant factor model is not significant (2 = 47.64, d.f. =
data well (2 = 1163.35, d.f. = 760; CFI = .97; NNFI = .96; and 65, p = .95). Thus, all correlational (covariance) relationships
RMSEA = .051). Next, we estimated a model that constrained the among model constructs, as well as construct (factor) variances,
factor loadings to be invariant across groups and found the fit are equal across samples. Finally, we estimated a fully invariant
indexes for this model to be acceptable (2 = 1182.75, d.f. = 780; measurement model (i.e., invariant factor loadings, invariant fac-
CFI = .97; NNFI = .96; and RMSEA =.050) and the difference in tor variance and covariances, and invariant item error loadings).
fit between this model and the baseline measurement model to be Although the indexes indicate adequate fit (2 = 1313.39, d.f. =
insignificant (2 = 19.4, d.f. = 20, p = .50). Thus, the loadings for 875; CFI = .967, NNFI = .96, and RMSEA = .049), this model’s fit
all measurement items are invariant across samples. The third differs from the fit of the baseline measurement model (2 =
model estimated specified invariant factor variance and covari- 83.0, d.f. = 30, p < .01). Still, the invariance of the factor loadings,
ances in addition to invariant factor loadings. It has acceptable fit factor variances, and covariances suggest that comparing model
indexes (2 = 1230.39, d.f. = 845; CFI = .96; NNFI = .96; and paths across samples is appropriate.
REFERENCES
Ahearne, Michael, C.B. Bhattacharya, and Thomas Gruen (2005), Auh, Seigyoung, Bulent Menguc, Michelle Fisher, and Abeer
“Antecedents and Consequences of Customer–Company Iden- Haddad (2011), “The Contingency Effect of Service Employee
tification: Expanding the Role of Relationship Marketing,” Personalities on Service Climate: Getting Employee Percep-
tions Aligned Can Reduce Personality Effects,” Journal of Ser-
Journal of Applied Psychology, 90 (3), 574–85. vice Research, 14 (4), 426–41.
Atuahene-Gima, Kwaku and Janet Y. Murray (2004), Autry, Chad W., Donna J. Hill, and Matthew O’Brien (2007),
“Antecedents and Outcomes of Marketing Strategy Compre- “Attitudes Toward the Customer: A Study of Product Returns
hensiveness,” Journal of Marketing, 68 (October), 33–46. Episodes,” Journal of Managerial Issues, 19 (1), 315–39.