Journal of Business Research 67 (2014) 159–166

Contents lists available at ScienceDirect

Journal of Business Research

Does delaying service-failure resolution ever make sense?

Yuanyuan Zhou a,⁎, Alex S.L. Tsang b, Minxue Huang c, Nan Zhou c,d

a School of Management, Huazhong University of Science and Technology, Wuhan, Chinab Department of Marketing, School of Business, Hong Kong Baptist University, Hong Kongc School of Economics and Management, Wuhan University, Wuhan, Chinad Department of Marketing, College of Business, City University of Hong Kong, Hong Kong

⁎ Corresponding author. Tel.: +86 027 8754 2754.E-mail addresses: Zhouyuanyuanmkt@126.com (Y. Z

(A.S.L. Tsang), Huangminxue@whu.edu.cn (M. Huang), na

0148-2963/$ – see front matter © 2012 Elsevier Inc. Allhttp://dx.doi.org/10.1016/j.jbusres.2012.10.009

a b s t r a c t

a r t i c l e i n f o

Article history:Received 12 March 2012Accepted 26 October 2012Available online 22 November 2012

Keywords:Service marketingService failureService recoveryDelaying resolutionImmediate resolutionService separation

Conventional wisdom suggests that service providers should respond to their failures as quickly as possible. Someresearch, however, points out that delaying resolutionsmayproducehighlydesirable results. The studyhere inves-tigates these competing views by examining under which conditions an immediate or a delaying resolution pro-duces more positive consumer responses in term of re-patronage and negative word-of-mouth intentions.Based on the concept of service separation, this research identifies an interaction effect between service separation(separated service, non-separated service) and response timing (immediate response, delaying response) onconsumers' post-consumption intention. This research also finds that the relationship between serviceseparation-response timing interaction and consumer response is mediated by consumers' negative emotions.This study contributes to refining our understanding of consumer psychology in service recovery. Practically, thestudies also enable service providers to better allocate their resources to recover different types of services' failure.

© 2012 Elsevier Inc. All rights reserved.

Surely early interventions are better than later ones, and offeringpeople the services of a trained professional is better than askingthem to sit and write by themselves. But we would be wrong……Itturns out that making people undergo CISD [Critical IncidentStress Debriefing] right after a trauma impedes the natural healingprocess……Wilson (2011)

1. Introduction

When should a service provider recover a service failure? Mostconsumers, probably including you, will say immediately, of course!Yes, in some cases, such as in medical service, the service providermust recover its failure as quickly as possible as it may causehuman lives. However, is the as-quickly-as-possible service recoverybelief valid in all circumstances?

Main-stream research findings support the as-quickly-as-possiblebelief (Boshoff, 1997; Dubé-Rioux, Schmitt, & Leclerc, 1989; Taylor,1994). A delaying resolution requires a consumer to wait thus resultsin negative emotions (Nie, 2000). Negative emotions then in turnproduce adversarial reactions, such as non-repurchase and negativeword-of-mouth (WoM) (Chebat & Slusarczyk, 2005; Schoefer, 2008;Schoefer & Diamantopoulos, 2008). Therefore, some scholars arguethat timely resolution is a standard to which providers should con-form (Wirtz & Mattila, 2004).

hou), tsangSL@hkbu.edu.hkn.zhou@cityu.edu.hk (N. Zhou).

rights reserved.

However, some scholars propose an opposing view. They arguethat a delay may produce better consumer responses to the recoverystrategy (Davidow, 2003; Karatepe & Ekiz, 2004; Mattila & Mount,2003). Neuropsychology theory suggests that a delay can have aquenching effect on conflicts between consumers and service pro-viders (Brehm, 1999). Specifically, a passage of time can calm downthe initial negative emotions triggered by a service failure, enablingconsumers to respond to firms' resolution efforts relatively rationally.Thus, a delay can serve as a buffer enabling consumers to fade out ini-tial negative emotions, in turn enhancing the effectiveness of a recov-ery resolution.

These two streams of research suggest two competing effects ofdelaying resolutions. An exaggerating effect suggests that a delaytriggers negative consumer responses. In contrast, a buffering effectsuggests that a delay calms customers, thus producing favorable res-olution results. This research aims to distinguish situations underwhich delaying response leads to an exaggerating or a buffering ef-fect. A pilot study shows that a delaying resolution produces betterconsumer responses. Nevertheless, this pilot study should ignore var-iables that moderate the relationship between response timing andconsumer response.

A dominant assumption within service literature is that service isnon-separated so that consumers are always on the spot from serviceproduction to consumption. However, many researchers severely chal-lenge this assumption in recent years. They have provided ample exam-ples indicating that separating service production and its consumptionis increasingly common (Edvardsson, Gustafsson, & Roos, 2005; Keh &Pang, 2010; Lovelock, 2000; Lovelock & Gummesson, 2004; Vargo &Lusch, 2004). Based on this argument, this research introduces service

160 Y. Zhou et al. / Journal of Business Research 67 (2014) 159–166

separation (separated service, non-separated service) as a moderatoraffecting whether different response timing (immediate, delaying) ofa resolution produces favorable or unfavorable consumer responses.This research argues that in a non-separated service's failure, an imme-diate resolution produces more favorable consumer responses, with re-gard to re-patronage and negative WoM intention, than a resolutiondelay. However, a resolution delay produces more favorable responsesin a separated service's failure. This research further includes con-sumers' negative emotions as a mediator in explaining the interactionbetween the two types of service separation.

A second study identifies an interaction effect between service sep-aration and response timing on consumers' post-consumption inten-tion. Compared with an immediate resolution, the level of negativeemotion is higher for a non-separated service's failure and lower for aseparated service's failure with a resolution delay. The findings providevaluable insights challenging two assumptions about servicemarketing.Theoretically, viewing all services as inseparable may limit understand-ing of how consumers respond to service failures and resolutions. Prac-tically, the studies also suggest that providing an immediate resolutionto service failures does not always produce the best results.

2. Theoretical background and hypotheses development

2.1. Two competing effects of delaying resolution

Service failures are inevitable in a firm's daily operation. Serviceproviders should resolve their failures; otherwise, the failures willcreate negative consequences, such as consumers spreading negativeWoM and switching to competitors (Chang, 2006; Maxham, 2001;McColl-Kennedy & Sparks, 2003; Vázquez-Casielles, Suárez Álvarez,& Díaz Martín, 2010; Zhou, Huang, Tsang, & Zhou, forthcoming). Re-sponse timing, an important attribute in a recovery strategy, hasattracted researchers' attention. Extant literature reports two com-peting effects of delaying resolutions.

2.1.1. Time is a scarce resourceThe prevalent view is that a response delay to a service failure will

negatively influence customers' post-consumption behavior (Bitner,Booms, & Tetreault, 1990; Boshoff, 1997; Dubé-Rioux et al., 1989;Johnston, 1995; Smart & Martin, 1992; Smith, Bolton, & Wagner, 1999;Taylor, 1994). Advocates of this propositionmainly base their argumentson a premise that time is a scarce resource. Thus, (1) timely response iscommonly a desired resolution a firm should perform (Clark, Kaminski,& Rink, 1992; Conlon & Murray, 1996; Gary, Kaminski, and Rink, 1992;Gilly & Gelb, 1982). Through a quick response, a service provider signalscustomers its commitment (Smart & Martin, 1992), efficiency (Folkes,1984; Wirtz & Mattila, 2004), and fairness (Chebat & Slusarczyk, 2005;Smith et al., 1999). In contrast, delays hinder consumers from resolvingtheir problem and can create serious problems (sometimes even fatal).(2) A delaying resolution makes consumers to wait, which is commonlya negative experience and an obstacle to goal achievement (Hui, Thakor,& Gill, 1998; Nie, 2000). Due to as-quickly-as-possible belief, people pre-fer avoid waiting (Miller, Kahn, & Luce, 2008). Thus,

H1a. An immediate resolution produces more favorable consumerresponses in terms of re-patronage and negative WoM intentionthan a resolution following a delay.

2.1.2. Time heals all woundsAnother line of research, in contrast, suggests that timely responses

to complaints may not produce the desired results (Davidow, 2003;Karatepe& Ekiz, 2004;Mattila &Mount, 2003;Miller et al., 2008). Advo-cates mainly base their arguments on the proposition that time has aquenching effect on conflicts between consumers and service providers.First, a timely responsemay create a negative conjecture for consumers.They may perceive that the service provider's quick response indicates

that it does not take the service failure seriously. For example,Davidow (2003) points out that “answering too soonmay leave the im-pression that the organization did not even look into the problem, po-tentially leaving consumers more frustrated than before.” In addition,time passes in the delay acts like a buffer thus can somewhat reducethe tension between the service provider and the affected consumer(Miller et al., 2008). Thus,

H1b. A resolution following a delay produces more favorable con-sumer responses in terms of re-patronage and negative WoM inten-tions than an immediate resolution.

2.2. Pilot study: testing the two competing effects

2.2.1. Design, participants, and procedureTo assess the validity of the two competing effects, this study first

conducts a pilot experiment, which manipulates response timing(immediate, delaying) to a service failure. One hundred undergradu-ate students participate for extra course credits. Of the participants,58% are female, and the average age is 20. This pilot experiment ran-domly assigns participants to either the immediate response or thedelaying response condition.

Each participant first read a scenario and projects himself or herselfinto a service failure. The scenario describes that the participant pur-chases a back cushion from an online shop. The participant later findsthat the cushion's quality is significantly worse than the shop describes.Thus, the participant complains to the shop. Then, this pilot study ma-nipulates the response timing. Specifically, in the immediate responsecondition, participants immediately receive a problem resolution,which is a free return plus a 10 RMB (about $1.57 US) gift certificate,after they complain. In the delaying response condition, participantsfirst complain of a service failure and approximately 30 min later re-ceive the same problem resolution. During the 30 min, this pilot exper-iment conduct a learning task for the participants, the content of whichis unrelated to the research. The participants then answer questionsabout dependent variables, including re-patronage and negative WoMintentions. Four items measure re-patronage intention (α=0.84): Iwill spend money at this online shop/I won't use this online shopagain (R)/I would like to use this online shop if needed/If needed, Iwill use another online shop (R). Three items measure negative WoMintention (α=0.68): I will spread negative word-of-mouth hereabout the online shop/I will bad-mouth this online shop to myfriends/When my friends are looking for a similar service, I will tellthem not to consume in this online shop. We revise all the aforemen-tioned scales from Grégoire and Fisher (2006) and Grégoire, Tripp,and Legoux (2009). They are on an 11-point Likert scale (1=stronglydisagree; 11=strongly agree).

2.2.2. Results and discussionResults of a MANOVA (Fig. 1) show that participants score signif-

icantly higher on re-patronage intention measures (Mimmediate=5.23, SD=2.05; Mdelaying=6.97, SD=1. 82; F(1, 98)=20.15,pb0.001) and lower on negative WoM measures (Mimmediate=5.94,SD=2.07; Mdelaying=4.18, SD=1. 70; F(1, 98)=21.56, pb0.001) inthe delaying response condition than in the immediate responsecondition.

The pilot study supports the notion that a delaying resolution to aservice failure produces better consumer responses than an immedi-ate resolution. However, it is too early to conclude that immediateresolutions produce relatively inferior consumer responses. Instead,a subsequent question needs consideration about any variables thatdefine the condition in which an immediate resolution produce betterconsumer responses and vice versa. A comprehensive literature re-view helps in revealing a dominant assumption within the servicemarketing literature that service is non-separated in that consumersare always on the spot of service, from its production through

0

2

4

6

8

Re-patronage intention Negative WoM intention

Immediate-response condition Delaying-response condition

Fig. 1. The effect of response timing on re-patronage intention and negative WoMintention.

161Y. Zhou et al. / Journal of Business Research 67 (2014) 159–166

consumption (Keh & Pang, 2010). The pilot study is in a separatedservice context that confirms the postulation about more favorableresponses to a delaying resolution. Will service separation (separatedservice, non-separated service) influence the effect of a delayingresolution? If there is an interaction effect between response timing(immediate response, delaying response) and service separation(separated service, non-separated service), what factor drives this ef-fect? This research shed light on these issues.

2.3. Service separation

As one of service's distinguishing characteristics, inseparability re-fers to the simultaneity of production and consumption, during the pro-cess of which the presence of consumers is an important factor (Berry,Seiders, & Grewal, 2002; Grove, Fish, & John, 2003; Lovelock &Gummesson, 2004; Sierra & McQuitty, 2005; Zeithaml, Parasuraman,& Berry, 1993). A large number of services fall into this characteristic,such as haircut and public transportation.

However, technological development has begun to question this as-sumption. More and more service researchers argue to reassess the va-lidity of inseparability as a distinctive characteristic of services(Edvardsson et al., 2005; Keh & Pang, 2010; Lovelock, 2000; Lovelock& Gummesson, 2004; Vargo & Lusch, 2004). Lovelock (2000) calls in-separability a dangerous oversimplification. Lovelock and Gummesson(2004) argue that “advances in information technology and telecom-munications … have made it possible to separate customers in bothtime and space from the production of numerous information-basedservices” (p. 32). Edvardsson et al. (2005) also points out that a largegroup of services does not involve the consumer directly, meaningthat production and consumption need not be simultaneous. Keh andPang (2010) provide abundant examples, including financial, entertain-ment, freight transport, and information services, to argue against theinseparability assumption. For instance, a consumer can either go to atraining service company to attend a course or watch uploadedwebcastvideos of the training program from the company's website. For thelatter service, consumers need not be present during the service pro-duction. This means that technology has enabled marketers to choosebetween offering their services in a separated or non-separated mode.

RESPONSE TIMING

Immediate response

Delaying response

NEGATIVE

SERVICE SEPARATION

Non-separated service

Separated service

Fig. 2. Conceptua

According to this concept of service separation, this research positsthat the difference between non-separated service and separated ser-vice will influence consumers' post-consumption intention across dif-ferent response timing (Fig. 2).

2.4. The role of negative emotion in response timing-service separationinteraction

Ample evidence in the service literature highlights the role of neg-ative emotion in the context of service failure and recovery encoun-ters. For example, Oliver (1997) shows that negative emotion iscentral in understanding consumers' reaction to service failure andrecovery. Consumers experience emotional reactions in response toservice failures and waiting for a resolution (Berry & Parasuraman,1991; Gustafsson, 2009; Kalamas, Laroche, & Makdessian, 2008;Oliver, 1997; Smith & Bolton, 2002; Zeithaml, Berry, & Parasuraman,1985). Consumers' emotional responses will subsequently influencetheir responses to firms' recovery efforts. Individuals who experiencea strong negative emotion are likely to respond less favorably thanothers who are in a less negative emotional state. Therefore,

H2. Negative emotion mediates the relationship of response timingand consumer responses in terms of re-patronage and negativeWoM intention.

According to above discussion, the major difference betweennon-separated and separated services is whether or not consumersbecome physically involved in the process of service production.This difference determines whether consumers are free to allocatetime when waiting for a resolution (i.e., redirect their attention toother things while waiting). This research argues that the effects ofnegative emotion vary across different service separation and subse-quently influence consumer responses.

For a non-separated service's failure, consumers are always on thespot of service, from service production through consumption. Thiskind of service will heighten consumers' attention on their time andeffort in gaining the service (Keh & Pang, 2010). Consumers cannotreallocate time across activities to achieve greater efficiency(Carman & Langeard, 1980). They will be more sensitive about thewaste of time in waiting for a resolution and demand the serviceprovider's immediate response.

Thus, a non-separated service failure will create a more aversive ex-perience for consumers waiting for a resolution and subsequently mag-nify their negative emotional state. Extant work has shown thatnegative emotions evoked during the recovery experience have a nega-tive impact on service recovery judgments (Chebat & Slusarczyk, 2005;Schoefer, 2008; Schoefer & Diamantopoulos, 2008). Therefore, waitingfor a resolution as an aversive experience suggests that a delay nega-tively affects consumers' post-consumption behavior. Thus:

H3. In a non-separated service's failure, an immediate resolution pro-duces a lower level of negative emotion than a resolution following adelay.

EMOTION CONSUMER

RESPONSES

l framework.

162 Y. Zhou et al. / Journal of Business Research 67 (2014) 159–166

H4. In a non-separated service's failure, an immediate resolution pro-duces more favorable consumer responses than a resolution follow-ing a delay.

In contrast, in a separated service's failure, consumers do not needto be on the spot and, thus, can allocate time and effort more effi-ciently and effectively, leading to perceived time and effort savings.In addition, a reasonable response delay will result in consumersreacting more favorably to the service provider's efforts (Miller etal., 2008). When encountering a service failure, consumer negativeemotions, such as anger, discontent, and anxiety, can occur essential-ly instantly. These initial negative emotions triggered by a service fail-ure also have a negative impact on customers' response to firms'recovery efforts (Andreassen, 1999; Smith & Bolton, 2002). Con-sumers with high negative emotions usually prevent things frommoving in a positive direction (Bless, Bohner, Schwarz, & Strack,1990; Reisenzein, 1994; Schwarz, 2002; Smith & Bolton, 2002).

Neuropsychological research has proposed that time has aquenching effect on conflicts between consumers and service pro-viders. Emotions which result from biological feedback intend to pro-tect a person in problematic or emergency situations (Brehm, 1999).Emotions are visceral factors, producing an immediate urge with highintensity and limited duration that subsides over time (Nowlis,Mandel, & McCabe, 2004).

For a separated service's failure, a delay creates a passage of time.The impact of negative emotions toward the service failure on con-sumers' response to service providers' resolutions will be on the de-crease. Negative emotions become too costly to maintain (Bonifield& Cole, 2007; Miller et al., 2008). Work by Miller et al. (2008)shows that the wait itself can facilitate consumers' coping with nega-tive events. The research of Grégoire et al. (2009) and McCullough,Fincham, and Tsang (2003) also point out that time is necessary forpeople's anger to die down and for them to forgive others. As the in-tensity of the negative emotions is shrinking, consumers will reactmore favorably to the service provider's efforts. Thus,

H5. In a separated service's failure, a resolution following a delay pro-duces a lower level of negative emotion than an immediate resolution.

H6. In a separated service's failure, a resolution following a delay pro-ducesmore favorable consumer responses than an immediate resolution.

3. Experiment

3.1. Design and participants

This scenario-based experiment consists of a 2 (service separa-tion: separated service, non-separated service)×2 (response timing:immediate, delaying) between-subjects design in the context of anEnglish training program. Two hundred undergraduate students par-ticipate in the experiment for extra course credit. Of the participants,52.5% are female, and the average age is 20. This experiment random-ly assigns participants to one of the four experimental conditions.

3.2. Materials

The scenarios involve two sections. The first section describes aconsumer choosing an English training program in a fictional organi-zation and then encountering a service failure. The manipulation ofservice separation follows the work of Keh and Pang (2010) by havingtwo modes of delivery for the English training program service(see Appendix A). Specifically, this experiment operationalizes the ma-nipulation as in-class instruction (non-separated service) or watching awebcast video (separated service). In the non-separated service'sfailure condition, the student cannot attend the class because anotherperson is occupying the classroom; in the separated service's failure

condition, the student cannot download the webcast video because ofan unknown login failure. After encountering the problem, the studentcomplains to the service organization.

The second section describes the service provider's resolution. Thestudent receives a free makeup course for both separated andnon-separated service conditions. The difference reflects the serviceprovider's response timing (immediate, delaying), which this experi-ment manipulates in the same way as the pilot study.

3.3. Procedures and variables

This experiment first randomly assigns participants to either thenon-separated service's failure condition or the separated service'sfailure condition. After reading the scenario related to the service fail-ure, participants complete a questionnaire expressing their initialemotions (At this moment, I feel angry/discontented/disappointed/self-pity/anxious; α=0.68). We revise the scales from Smith andBolton (2002) by using a 7-point Likert scale (1=strongly disagree;7=strongly agree). In the second step, this experiment employs arandomly assigned response timing (immediate or delaying), as inthe pilot study. Third, participants read the firm's resolution fortheir complaint (see Appendix B).

Participants then answer questions about the main variables, in-cluding re-patronage intention, negative WoM intention, and nega-tive emotion. The measurements about re-patronage intention (α=0.83) and negative WoM (α=0.76) are the same as those in thepilot study. The negative emotion measurements are the same asthe scales measuring participants' initial emotions (α=0.84).

Finally, participants answer manipulation check questions abouttheir perception of (1) service separation (no special time/no specialplace/according to one's own schedule; α=0.90) from Keh and Pang(2010) and (2) response modes (response timely/immediately; α=0.90). The scales are a 7-point Likert scale (1=strongly disagree;7=strongly agree).

3.4. Results

3.4.1. Manipulation checksThe two-way ANOVA shows that there is no significant difference

across the two independent variables on the score of initial emotions(F(3, 196)=1.49, p=0.22). Second, consistent with expectations,one-way ANOVA shows that participants perceive separated andnon-separated services differently (Mnon-separated=3.25, SD=1.25;Mseparated=5.38, SD=1.21; F(1, 199)=151.11, pb0.001). Also,participants perceive that the training company responds morequickly in the immediate condition than in the delaying condition(Mimmediate=4.48, SD=1.45; Mdelaying=5.00, SD=1.21; F(1, 199)=7.59, pb0.01). Therefore, the manipulations are successful.

3.4.2. Reactions to problem resolutionResults of a MANOVA identify significant interaction effects be-

tween service separation and response timing (re-patronageintention: F(1, 196)=37.62, pb0.001; negative WoM intention:F(1, 196)=26.27, pb0.001) (Fig. 3). Planned contrasts show that ina separated service's failure, participants score significantly higheron repurchase intention (Mimmediate=5.90, SD=2.27; Mdelaying=7.25, SD=1. 27; t=3.99, pb0.01) measures and lower on negativeWoM measures (Mimmediate=7.13, SD=2.33; Mdelaying=5.37, SD=2.18; t=4.06, pb0.001) in the delaying response condition. For anon-separated service's failure, participants score significantly higheron repurchase intention (Mimmediate=7.83, SD=1.56; Mdelaying=6.24,SD=1. 51; t=4.68, pb0.001) measures and significantly higher onnegative WoM measures (Mimmediate=5.33, SD=2.17; Mdelaying=6.71, SD=1. 97; t=3.186, pb0.01) in the immediate response condi-tion. Thus, the results support for H4 and H6.

(A) Dependent variable: Re-patronage intention

(B) Dependent variable: Negative WoM

7.83

6.245.9

7.25

5

5.5

6

6.5

7

7.5

8

Nonseparated service

Separated service

5.33

6.71

7.13

5.375

5.5

6

6.5

7

7.5

8

Immediateresponse

Delayingresponse

Immediateresponse

Delayingresponse

Nonseparated service

Separated service

Fig. 3. Interaction between response timing and service separation. (A) Dependent var-iable: Re-patronage intention. (B) Dependent variable: Negative WoM.

163Y. Zhou et al. / Journal of Business Research 67 (2014) 159–166

3.4.3. Negative emotion as mediatorThe analysis indicates a significant interaction effect between

service separation and response timing on negative emotion (F(1,196)=17.16, pb0.001) (Fig. 4). Planned contrasts show that for aseparated service's failure, participants score significantly lower onnegative emotion (Mimmediate=4.32, SD=1.40; Mdelaying=3.77,SD=1. 06; t=3.99, pb0.05) measures in the delaying response con-dition, while for a non-separated service's failure, participants scoresignificant lower on negative emotion (Mimmediate=3.69, SD=1.35;Mdelaying=4.56, SD=0. 99; t=3.604, pb0.001) measures in the im-mediate response condition. Thus, the results support for H3 and H5.

A mediated moderation analysis follows the procedure suggested inMuller, Judd, and Yzerbyt (2005) to verify the mediating role of nega-tive emotion. This analysis is first for re-patronage intention measures(Table 1). Following the first criterion, the interaction betweenresponse timing and service separation significantly predictsre-patronage intention (B=0.40, t=6.13, pb0.001). At the secondstep, results shows that the interaction between response timing andservice separation significantly predicts negative emotion (B=−0.28,

3.69

4.56

4.32

3.77

3

3.2

3.4

3.6

3.8

4

4.2

4.4

4.6

4.8

5

Immediateresponse

Delayingresponse

Nonseparated service

Separated service

Fig. 4. Interaction between response timing and service separation on negative emotion.

t=−4.14, pb0.001). Next examination shows that negative emotionsignificantly predicts re-patronage intention while controlling forthe interaction between response timing and service separation andthe interaction between service separation and negative emotion(B=−0.37, t=−5.91, pb0.001). Finally, the association between theresponse timing–service separation interaction and re-patronage inten-tion decreases from 0.40 to 0.30. A Sobel test shows that this decrease isstatistically significant, Z=3.39, pb0.001. Thus, negative emotion me-diates the relationship between the response timing–service separationinteraction and re-patronage intention.

A similar test provides strong evidence for the effect of negativeemotion on negativeWoM intention (Table 2). Following the first crite-rion, the interaction between response timing and service separationsignificantly predicts negative WoM intention (B=−0.34, t=−5.13,pb0.001). At the second step, results show that the interaction betweenresponse timing and service separation significantly predicts negativeemotion (B=−0.28, t=−4.14, pb0.001). Next examination showsthat negative emotion significantly predicts negative WoM intentionwhile controlling for the interaction between response timing and ser-vice separation and the interaction between service separation and neg-ative emotion (B=0.49, t=7.97, pb0.001). Finally, the associationbetween the response timing–service separation interaction and nega-tiveWoM intention decreases from−0.34 to−0.21. A Sobel test showsthat this decrease is statistically significant, Z=3.67, pb0.001. Thus,negative emotion mediates the relationship between the responsetiming–service separation interaction and negative WoM intention.Thus, the results support for H2.

4. Discussion and conclusion

How to make service recovery more effective is always an essen-tial concern in service literature. This paper focuses on the issue of re-sponse timing and seeks to determine whether the service providercan take advantage of timing to gain more favorable consumer re-sponses. This research conducts a pilot study and an experiment toverify the conceptualization and hypotheses of this paper.

Specifically, the pilot study demonstrates that a delaying resolutionproduces more favorable consumer responses than an immediate reso-lution. The subsequent experiment shows that in a non-separatedservice's failure, an immediate resolution produces better effects thana delaying resolution. However, in a separated service's failure, adelaying resolution produces better effects than an immediate resolu-tion.Moreover, negative emotionmediates the relationship of responsetimingwith consumers' post-consumption intention. In the situation ofa non-separated service's failure, an immediate resolution produces alower level of negative emotion than a delaying resolution; in the situ-ation of a separated service's failure, a delaying resolution produces alower level of negative emotion than an immediate resolution.

4.1. Theoretical and managerial contribution

Service recovery researchers focus more on developing an overallframework for measuring and managing service providers' responseto service failures and complaints (i.e., Schoefer & Diamantopoulos,2008; Smith et al., 1999; Tax, Brown, & Chandrashekaran, 1998),but do not focus on one specific aspect of recovery strategy and itsspecific driving factors. Response timing as an important attribute ofa recovery strategy falls within the purview of this paper.

Extant research identifies that a delaying resolutionmay trigger twocompeting effects on consumers' post-complaint responses (Boshoff,1997; Chebat & Slusarczyk, 2005; Davidow, 2003; Dubé-Rioux et al.,1989; Karatepe & Ekiz, 2004; Mattila & Mount, 2003; Schoefer, 2008;Schoefer & Diamantopoulos, 2008; Taylor, 1994). Consistent withDavidow's (2003) statement that response timing is situation specific,this research distinguishes conditions under which a delay can lead tomore favorable or less favorable consumer responses. An interaction

Table 1Mediated moderation analyses for re-patronage intention.

Re-patronage intention Negative emotion Re-patronage intention

Predictors b t b t b t

X: response timing −0.03 −0.49 0.07 0.95 −0.02 −0.30MO: service separation −0.12 −1.91 −0.03 −0.49 −0.01 −0.06XMO: response timing×service separation 0.40 6.13⁎⁎⁎ −0.28 −4.14⁎⁎⁎ 0.30 4.71⁎⁎⁎

Me: negative emotion −0.37 −5.91⁎⁎⁎

Me MO: negative emotion×service separation −0.13 −0.61

⁎ pb0.1.⁎⁎ pb0.05.⁎⁎⁎ pb0.001.

164 Y. Zhou et al. / Journal of Business Research 67 (2014) 159–166

effect between service separation (separated, non-separated) and re-sponse timing (immediate, delaying) suggests that service separationis a moderator and guides us to a better understanding of how adelaying resolution influences consumer responses.

The emotional perspective helps understand more about con-sumers' psychological mechanisms. Negative emotion is central to un-derstand consumers' reaction in service failure and recovery (Berry &Parasuraman, 1991; Oliver, 1997; Smith & Bolton, 2002; Zeithaml etal., 1985). A delay itself can result in negative emotion and also can cre-ate a quenching effect. This research indicates that in the situation of anon-separated service's failure, an immediate resolution produces alower level of negative emotion than a delaying resolution; in the situ-ation of a separated service's failure, a delaying resolution produces alower level of negative emotion than an immediate resolution.

This research has several managerial implications. Conventionalwisdom suggests that service providers should immediately recoverservice failures and complaints. Do immediate responses always pro-duce better effects? This paper suggests that under some conditionsdelaying a response can also maintain favorable effects. This paper in-troduces service separation as a consideration context. Some servicesare inseparable by nature, such as hairstyling, and some are separable,such as freight transport. Also, there are some services that providerscan choose whether to deliver in a separated mode or a non-separatedmode. For example, you can either go to the training service companyto attend your course or you can watch the uploaded webcast of thetraining program from the company's website. An immediate responseto service failure is a must in non-separated services' failures. Con-sumers will be more sensitive about waste of time in waiting for aresolution and this is apt to magnify consumers' negative emotion.However, for separated services' failures, managers and employeescan deliver their resolution to service failure using a delaying response;this will make their resolution more effective.

4.2. Recommendations for future research

Future research should extend the scope of this research and resolvesomeof its limitations. First, some factors related to response timingneedfurther consideration: (1) the levels of response timing. This paperdichotomizes response timing into only two categories (immediate re-sponse, delaying response). If a delay produces a favorable response in

Table 2Mediated moderation analyses for negative WoM intention.

Negative WoM intention

Predictors b t

X: response timing −0.04 −0.62MO: service separation 0.05 0.75XMO: response timing×service separation −0.34 −5.13⁎⁎⁎

Me: negative emotionMe MO: negative emotion×service separation

⁎ pb0.1.⁎⁎ pb0.05.

⁎⁎⁎ pb0.001.

separated services, does this mean that a firm can delay its resolutionfor an extended period of time? Is there the same effect for a slightlydelaying response and a severely delaying response? Determining theoptimal response time can enrich our understanding of the effects ofdifferent response speeds levels; (2) the nature of response timing.Davidow (2003) classifies three responses: an immediate response, anecessary delay to solve the problem, and an unnecessary delay. Wheth-er a failure allows a delay should be a critical factor. Investigating thesefactors should provide valuable insights into developing a comprehen-sive understanding of this phenomenon.

Second, this paper stresses economic recovery. The positive effectsof a delay may vary according to different loss results from the servicefailure. Economic recovery and social recovery differ in their recoveryattempts. Economic recovery commonly involves utilitarian resourcessuch as discounts and refunds, while social recovery involves psycho-logical or symbolic resources such as apologies and explanations(Smith et al., 1999). The effects of a delay on social recovery needmore consideration to further our understanding of the responsetiming issue in service recovery.

Third, further research should extend and investigate whether andhow cultural values impact the effect of a delay. Collectivists showmore tolerance than individualists for firms' responses and differ-ences in time orientation (Chen, Ng, & Rao, 2005). The data of this re-search are from China and provide supportive evidence. Furtherresearch should cover other cultures to enhance generalizability.

4.3. Conclusion

The literature has identified two competing effects in delayingservice-failure resolutions. This research introduces service separa-tion as a moderator and sheds light on the conditions under whicha delaying response will produce favorable consumer responses interms of re-patronage and negative WoM intentions. Specifically, animmediate resolution produces more favorable responses for anon-separated service's failure and a delaying resolution producesmore favorable responses for a separated services' failure. Negativeemotion helps understand the underlying mechanism. Comparedwith an immediate resolution, the level of negative emotion is higherfor a non-separated service's failure and lower for a separatedservice's failure with a delaying resolution.

Negative emotion Negative WoM intention

b t b t

0.07 0.95 −0.06 −1.05−0.03 −0.49 −0.04 −0.19−0.28 −4.14⁎⁎⁎ −0.21 −3.40⁎⁎

0.49 7.97⁎⁎⁎

0.11 0.54

165Y. Zhou et al. / Journal of Business Research 67 (2014) 159–166

Appendix A

Scenarios for service separation.

Separated service Non-separated service

You will be taking an important English exam this term. You would like to attend atraining program for this exam and thus consult ABC company, which is famous forproviding this kind of English training program.After seeing some types of English training programs in this company, you choose anonline program. The following is some information about your program:

(1) Online English training program for XX examThis program is a long-distance English training service. You can take yourcourses according to your own schedule and do not need to go to the company.You have the right to choose the teacher who is right for this program.

(2) Instruments for this program1. You will get an account with password which is assigned by our company

through which to receive your materials for this program;2. You can log on software to learn using this account and password according to

your own schedule;3. After logging on, you can click the button “video of courses” and “notes of

courses” to obtain your materials. These are the uploaded webcast and notesof the English training program, which are assigned by the teacher you choose.You can leave a message for the teacher if you have questions about yourcourses. The teacher will solve your problems patiently throughout thisprogram.Today, you are preparing for a course as usual. However, you cannot downloadyour material for this lesson because of an unknown login failure. You ask thecompany the reason for this problem, but no one there knows. You feel veryangry. You think that it is not reasonable that an unknown login failure deniesyou access to today's course. Thus, you complain to the manager of thecompany.

You will be taking an important English exam this term. You would like to attend atraining program for this exam and thus consult ABC company, which is famous forproviding this kind of English training program.After seeing some types of English training programs in this company, you choose anin-class program. The following is some information about your program:

(1) In-class English training program for XX examThis program is a one-by-one English training service. You can go to thecompany and attend the courses there per your appointment schedule. Youhave the right to choose the teacher who is right for this program.

(2) Instruments for this program1. You will get a classroom which is assigned by our company to attend your

courses;2. You can take courses using this room per your appointment schedule;3. In this period of time, the teacher you choose will provide you with courses

and hand you the notes of the English training program. You can ask questiondirectly to the teacher if you have questions about your courses. The teacherwill solve your problems patiently throughout this program.Today, you are preparing for a course as usual. However, you cannot have classfor this lesson because other person is occupying the classroom. You ask thecompany the reason for this problem, but no one there knows. You feel veryangry. You think that it is not reasonable that an unknown occupy failuredenies you access to today's course. Thus, you complain to the manager ofthe company.

Appendix B

Scenarios for resolution.

Immediate resolution Delaying resolution

Separatedservice

Now, the manager of the company attempts to solve the problem asfollows: offers a free makeup course for this lesson with an additional freeonline course.

(This experiment manipulates 30 min as a delaying response. During the 30 min,participants complete a learning task, the content of which is unrelated to ourresearch.) Imagine the time you have waited for a resolution after your complaint.Now, the manager of the company attempts to solve the problem as follows: offersa free makeup course for this lesson with an additional free online course.

Non-separatedservice

Now, the manager of the company attempts to solve the problem asfollows: offers a free makeup course for this lesson with an additional freein-class course.

(This experiment manipulates 30 min as a delaying response. During the 30 min,participants complete a learning task, the content of which is unrelated to ourresearch.) Imagine the time you have waited in the company for a resolution afteryour complaint. Now, the manager of the company attempts to solve the problemas follows: offers a free makeup course for this lesson with an additional freein-class course.

References

Andreassen, T. W. (1999). What drives customer loyalty with complaint resolution?Journal of Service Research, 1, 324–332.

Berry, L. L., & Parasuraman, A. (1991). Marketing services: Competing through quality.New York: The Free Press.

Berry, L. L., Seiders, K., & Grewal, D. (2002). Understanding service convenience. Journalof Marketing, 66, 1–17.

Bitner, M. J., Booms, B. H., & Tetreault, M. S. (1990). The service encounter: Diagnosingfavorable and unfavorable incidents. Journal of Marketing, 54, 71–84.

Bless, H., Bohner, G., Schwarz, N., & Strack, F. (1990). Mood and persuasion: A cognitiveresponse analysis. Personality and Social Psychology Bulletin, 16, 331–345.

Bonifield, C., & Cole, C. (2007). Affective responses to service failure: Anger, regret, andretaliatory versus conciliatory responses. Marketing Letters, 18, 85–99.

Boshoff, C. (1997). An experimental study of service recovery options. InternationalJournal of Service Industry Management, 8, 110–130.

Brehm, J. W. (1999). The intensity of emotion. Personality and Social Psychology Review,3, 2–22.

Carman, J. M., & Langeard, E. (1980). Growth strategies for service firms. StrategicManagement Journal, 1, 7–22.

Chang, C. C. (2006). When service fails: The role of the salesperson and the customer.Psychology and Marketing, 23, 203–224.

Chebat, J., & Slusarczyk, W. (2005). How emotions mediate the effects of perceived jus-tice on loyalty in service recovery situations: An empirical study. Journal of BusinessResearch, 58, 664–673.

Chen, H., Ng, S., & Rao, A. R. (2005). Cultural differences in consumer impatience.Journal of Marketing Research, 42, 291–301.

Clark, G. L., Kaminski, P. F., & Rink, D. R. (1992). Consumer complaints: Advice on howcompanies should respondbased on an empirical study. Journal of ConsumerMarketing,6, 41–50.

Conlon, D. E., & Murray, N. M. (1996). Customer perceptions of corporate responses toproduct complaints: The role of explanations. Academy of Management Journal, 39,1040–1056.

Davidow, M. (2003). Organizational response to customer complaints: What worksand what doesn't. Journal of Service Research, 5, 225–250.

Dubé-Rioux, L., Schmitt, B. H., & Leclerc, F. (1989). Consumers' reactions to waiting:When delays affect the perception of service quality. Advances in Consumer Research,16. (pp. 59–63).

Edvardsson, B., Gustafsson, A., & Roos, I. (2005). Service portraits in service research: Acritical review. International Journal of Service Industry Management, 16, 107–121.

166 Y. Zhou et al. / Journal of Business Research 67 (2014) 159–166

Folkes, V. S. (1984). Consumer reactions to product failure: An attributional approach.Journal of Consumer Research, 11, 398–409.

Gilly, M. C., & Gelb, B. D. (1982). Post-purchase consumer processes and thecomplaining consumer. Journal of Consumer Research, 9, 323–328.

Grégoire, Y., & Fisher, R. J. (2006). The effects of relationship quality on customer retal-iation. Marketing Letters, 17, 31–46.

Grégoire, Y., Tripp, T. M., & Legoux, R. (2009). When customer love turns into lastinghate: the effects of relationship strength and time on customer revenge and avoid-ance. Journal of Marketing, 73, 18–32.

Grove, S. J., Fish, R. P., & John, J. (2003). The future of service marketing: Forecasts fromten services experts. Journal of Service Marketing, 17, 106–119.

Gustafsson, A. (2009). Customer satisfaction with service recovery. Journal of BusinessResearch, 62, 1220–1222.

Hui, M. K., Thakor, M. V., & Gill, R. (1998). The effect of delay type and service stage onconsumers' reactions to waiting. Journal of Consumer Research, 24, 469–479.

Johnston, R. (1995). The determinants of service quality: Satisfiers and dissatisfiers.International Journal of Service Industry Management, 6, 53–71.

Kalamas, M., Laroche, M., & Makdessian, L. (2008). Reaching the boiling point:Consumers' negative affective reactions to firm-attributed service failures. Journalof Business Research, 61, 813–824.

Karatepe, O., & Ekiz, E. H. (2004). The effects of organizational response to complaintson satisfaction and loyalty: A study of hotel guests in northern Cyprus. ManagingService Quality, 14, 476–486.

Keh, H. T., & Pang, J. (2010). Customer reactions to service separation. Journal ofMarketing, 74, 55–70.

Lovelock, C. (2000). Christopher lovelock: Lovelock associates. In R. P. Fisk, S. F. Grove,& J. John (Eds.), Services marketing self-portraits: Introspections, reflections, andglimpses from the experts (pp. 133–152). Chicago: American Marketing Association.

Lovelock, C., & Gummesson, E. (2004). Whither services marketing? In search of a newparadigm and fresh perspectives. Journal of Service Research, 7, 20–41.

Mattila, A. S., & Mount, D. J. (2003). The impact of selected customer characteristics andresponse time on e-complaint satisfaction and return intent. Hospitality Management,22, 135–145.

Maxham, J. G., III (2001). Service recovery's influence on consumer satisfaction, posi-tive word-of-mouth, and purchase intentions. Journal of Business Research, 54,11–24.

McColl-Kennedy, J. R., & Sparks, B. A. (2003). Application of service failures and servicerecovery. Journal of Business Research, 5, 251–266.

McCullough, M. E., Fincham, F. D., & Tsang, J. (2003). Forgiveness, forbearance, andtime: The temporal unfolding of transgression-related interpersonal motivations.Journal of Personality and Social Psychology, 84, 540–557.

Miller, E. G., Kahn, B. E., & Luce, M. F. (2008). Consumer wait management strategies fornegative service events: A coping approach. Journal of Consumer Research, 34,635–648.

Muller, D., Judd, C. M., & Yzerbyt, V. Y. (2005). When moderation is mediated and me-diation is moderated. Journal of Personality and Social Psychology, 89, 852–863.

Nie, W. (2000). Waiting: integrating social and psychological perspectives in opera-tions management. Omega: The International Journal of Management Science, 28,611–629.

Nowlis, S. M., Mandel, N., & McCabe, D. B. (2004). The effect of a delay between choiceand consumption on consumption enjoyment. Journal of Consumer Research, 31,502–510.

Oliver, R. L. (1997). Customer satisfaction. A behavioral perspective on the consumer.New York: McGraw-Hill.

Reisenzein, R. (1994). Pleasure-arousal theory and the intensity of emotions. Journal ofPersonality and Social Psychology, 67, 525–539.

Schoefer, K. (2008). The role of emotions in translating perceptions of (in)justice intopostcomplaint behavioral responses. Journal of Service Research, 11, 91–103.

Schoefer, K., & Diamantopoulos, A. (2008). Measuring experienced emotions duringservice recovery encounters: Construction and assessment of the ESRE scale.Service Business: An International Journal, 2, 65–81.

Schwarz, N. (2002). Situated cognition and the wisdom of feelings: Cognitive tuning. InL. F. Barrett, & P. Salovey (Eds.), The wisdom in feelings: Psychological processes inemotional intelligence (pp. 144–166). New York: Guilford.

Sierra, F. F., & McQuitty, S. (2005). Service providers and customers: Social exchangetheory and service loyalty. Journal of Service Marketing, 6, 392–400.

Smart, D. T., & Martin, C. L. (1992). Manufacturer responsiveness to consumercorrespondence: An empirical investigation of consumer perceptions. Journal ofConsumer Affairs, 26, 104–128.

Smith, A. K., & Bolton, R. N. (2002). The effect of customers' emotional response to ser-vice failure on their recovery effort evaluations and satisfaction judgments. Journalof the Academy of Marketing Science, 30, 5–23.

Smith, A. K., Bolton, R. N., & Wagner, J. (1999). A model of customer satisfaction withservice encounters involving failure and recovery. Journal of Marketing Research,6, 356–372.

Tax, S. S., Brown, S. W., & Chandrashekaran, M. (1998). Customer evaluations of servicecomplaint experiences: Implications for relationship marketing. Journal of Marketing,62, 60–67.

Taylor, S. (1994). Waiting for service: The relationship between delays and evaluationsof service. Journal of Marketing, 58, 56–69.

Vargo, S. L., & Lusch, R. F. (2004). The four service marketing myths: Remnants of agoods-based, manufacturing model. Journal of Service Research, 6, 324–335.

Vázquez-Casielles, R., Suárez Álvarez, L., & Díaz Martín, A. M. (2010). Perceived justiceof service recovery strategies: Impact on customer satisfaction and quality rela-tionship. Psychology and Marketing, 27, 487–509.

Wilson, T. D. (2011). Redirect: The surprising new science of psychological change.New York: Little, Brown and Company.

Wirtz, J., & Mattila, A. S. (2004). Consumer responses to compensation speed of recov-ery and apology after a service failure. International Journal of Service IndustryManagement, 15, 150–166.

Zeithaml, V. A., Berry, L. L., & Parasuraman, A. (1985). The nature and determinants ofcustomer expectations of service. Journal of the Academy of Marketing Science, 21,1–12.

Zeithaml, V., Parasuraman, A., & Berry, L. (1993). Problems and strategies in servicesmarketing. Journal of Marketing, 49, 33–46.

Zhou, Y., Huang, M., Tsang, S. L., & Zhou, N. (forthcoming). Recovery strategy for groupservice failures: the interaction effects between recovery modes and recoverydimensions. European Journal of Marketing.