I need summarize to this Researchmethodology

1.1Study1

Study 1 testsH1and the basic assumption of a direct effect ofcustomer incivility on service levels through a eld study. Fortheeldstudy,wetrainedconfederatestomakeicecreampurchasesat19differenticecreamshopsinGermanyforatotalof 221 unique service interactions with different FSEs at theseshops. Ice cream shops were chosen to test the rst hypothesisfor several reasons. First,interactions in this setting are guidedbynorms;employeesshouldgenerallyexpectcustomerstoexhibitcivilbehavior.Therefore,uncivilbehaviorwouldconstituteanormviolationthatcouldmotivatetheFSEtoreciprocate the uncivil behavior. Second, FSEs can offer lowerservice levels by underserving customers and giving out lessgenerousscoopsoficecream.Therefore,inthiscase,measuring the amount of ice cream received would provide anobjective measure of retaliation.This approach was used giventhat an explicit measure of retaliation, such as asking a surveyquestion,wouldlikelybeimpactedbyresponsebiases.

Each confederate-FSE interaction involved either an uncivil,neutral or polite customer interaction. The three-level researchdesign followed ethical codes guiding research in Germany,and we ensured that each FSE encountered a limited numberof uncivil interactions. As ice cream vending interactions aretypically very short in duration, the display of uncivil customerbehavior can only be transferred by a lack of"please"and"thank you,"a comment about waiting in line too long orthroughfacialexpression(Appendix1).Thus,uncivilencounters in an ice cream vending setting are likely to onlyhave a mild effect, if any, on FSE's overall well-being orcommitment.

EachFSEwasdebriefedafterthetransactionandwasgiventhe opportunity to opt out of the study. All FSEs consented,andtheirinteractionswereincludedinthedataset.

1.1.1Procedure

A total of 14 university students from a German universityserved as confederates in the eld study.Study recruitment wascarriedoutaspartofanelectivecourseinagraduateprogramforbusinessstudents,andstudentsweregivenanopportunityto drop out of the study. The confederates received actortraining and self-selected into the roles of polite, neutral oruncivil customers. It is important to note that we encouragedconfederates to only assume behavioral roles that they werecomfortable with. As a result, several confederates selectedpoliteorneutralrolesonly,whichwasnotedandimplementedaccordinglyduringtheeldstudy.However,themajorityof

confederateswerecomfortablecyclingthroughallthreebehavioralroles.

Inthetrainingsessions,wegavetheconfederatesinstructionson how to behave as uncivil, neutral or polite customers, andwetestedtheirbehaviortoensureroleconsistency(Appendix1).Giventheshortnatureofthetransaction,displays of civility were limited to facial expressions, briefcomments about the service encounter and politeness: i.e.using"please"and"thankyou."Confederatesthenparticipatedintheicecreampurchases.

Tomeasureemployeeretaliation,weusedtheweightoftheice cream served to each confederate. Measuring employeereciprocity objectively, i.e. with a manifest weight variable andin a disguised experimental setting, has the advantage of rulingout social desirability biases. Furthermore, prices per scoop arefully transparent such that no obscure charge could be added asameansofretaliation.Tomaketheweightvariablecomparable, we allowed only for two avors-vanilla orchocolate-and controlled for their corresponding densities.Each confederate ordered two scoops of ice cream during eachinteraction, and the weight varied between 3.60 and 6.28ounces. We identied 19 ice cream shops that fullled ourprerequisites: full price and service transparences, standardizedspoons for scooping and multiple customer service employees.In total, data consisted of 221 service interactions across 19 icecreamshops.Oftheinteractions,55involvedpoliteconfederatecustomers,92involvedneutralconfederatecustomersand74involveduncivilconfederatecustomers.

1.1.2Results

We performed an analysis of covariance of the ice creamportions related to the three manipulated customer behaviorconditionstotesttherobustnessandcausalityofthecustomerincivility-employeeretaliationrelationship.Wefoundasignicanteffectofcustomerbehavioronportionsize(Eta²=0.032,p<0.05,Figure 1). The mean portion sizes conrmedH1,whichstatedthatemployeesreciprocateuncivilbehavior.Inotherwords,confederatesdisplayingpolitebehaviorreceivedthebiggestportions(mean=5.57 oz;standarddeviation [SD] = 1.03), and confederates displaying uncivilbehavior received the smallest portions (mean = 5.13 oz; SD =0.97,p<0.05). Confederates displaying neutral behavior(mean = 5.46 oz; SD = 1.01) yielded portions similar to thepolite customers (mean = 5.57 oz; SD = 1.03,p>0.10) andgreaterthanthosetheuncivilgroupreceived(p<0.05).

CustomerBehavior

To control for spurious effects, we tested for gender effects butdid not nd signicantdifferences in the netweightof icecream (Eta²= 0.03,p= 0.42), indicating that neither males norfemales received systematically bigger portions of ice cream.We also tested forthe individual studentactor bias(Eta²=0.05,p= 0.61), but there were no systematic differences in theamount of ice cream a confederate received. We also tested thedyadsforsex-relatedeffects,i.e.ifopposite-sexcustomerswould receive more ice cream than same-sex customers, but nosignicanteffectswerefound.Finally,wetestedfordifferencesamong the ice-cream shops and found a signicant difference(Eta²= 0.63,p<0.01). However, these differences are mostlikely because of the different models of scooping spoons, cupsizes and corporate philosophies; yet, these differences do notcompromise the validity of our ndings on customer behavior.These results support the argument that FSE are more likely tooffer lower service levels to customers that treat them in anuncivilmanner,whichsupportsH1.

1.1.3Discussion

InStudy1,wetestedthebasicrelationshipbetweencustomerincivility and employee service level outcomes. With ice creamweight as a proxy variable for service levels, we tested thedifferences in ice cream weight between polite, neutral anduncivil customer behavior in a eld-experiment across 221distinct service interactions staged by 14 different confederates.Theeffectofcustomerbehaviorwassignicantandprovidesinitial support for employees matching service levels withbehaviorreceivedfromcustomers.Onelimitationoftheeldstudy is that we are unable to examine the process throughwhichthisdifferentialtreatmentoccurs.Therefore,theobjectiveofStudy2istoreplicateStudy1usingalaboratorysettingandtestingformediatedandmoderatedeffectstoshedlight on the reasons and conditions under which this effectoccurs. This allows us to test bothH1andH2in a controlledsetting.

1.2Study2

Theobjectiveofthislaboratoryexperimentistoreplicatetheeld experiment's nding that customers who behave in anuncivilmannerreceivelowerservicelevelsthancustomerswhobehave in a neutral or polite manner. Moreover, we seek toaddressegothreatandinteractionalinjusticeasprocessesthatexplain the differences in employee reactions. We also use adifferentservicecontext:airlinetravel.Wechosethiscontexttonot only increase the generalizability of our results but alsobecauseairtravelisafamiliarsettinginwhichFSE-customerinteractions play an important role in value creation (BabbarandKoufteros,2008).Furthermore,thissettingallowedustouse an indirectmeasureofretaliation because anydirectmeasurecouldbeimpactedbyresponsebias.

1.2.1Participants

A total of 90 English-speaking respondents (53% female,median age = 31 years) participated in this study on Amazon'sMechanical Turk in exchange for monetary compensation.Respondents were randomly assigned to either an uncivil,neutral or polite customer-employee interaction. Therefore,thisstudyusedathree-condition(customerbehavior:uncivil,neutralandpolite),between-subjectsdesign.

1.2.2Procedure

Respondents were asked to put themselves in the shoes of anairline check-in clerk and to respond to a scenario involving aninteractionwithanairlinecustomer.Theyweretoldthattheywere approached by the customer about changing seats. Wemanipulated uncivil, neutral or polite behavior by adjusting thetone the customer used when approaching the airline employee(Appendix 2). Respondents were then shown a seating chart ofthe plane and asked to make a decision about the customer'sseatassignment.Respondentshadtheoptiontoleavetheseatas is, upgrade the customer to economy plus or business class-both with an additional upcharge, or offer the customer a bettereconomy seat on a ight leaving 4 h later. After indicatingwhichoptiontheywouldpresenttothecustomer,respondentscompletedamanipulationcheckregardingtheperceivedbehavioroftherespondent(seven-pointscale,anchors"Veryfriendly-Veryunfriendly"and"Polite-Rude,"r=0.92)andadditionalquestionsaboutthecustomerinteraction,includingego threat (a= 0.90) and interactional injustice scales (a=0.95) (Appendix 3). In line withSkarlickiet al.(2008), wedeveloped and added ameasureof interactional injusticerelevant to our research context. Thead hocego threat measuretaps into the feelings of being devalued and having strongnegative emotions regarding the uncivil behavior as thesenegative emotions are strongly related to ego threat (Stucke andSporer,2002).

Respondentsthenratedeachoftheoptionstheyhad

available for the customer's new seating assignment, includingthe one they decided to assign (1 ="Very Bad"to 5 ="VeryGood").Weusedtheratingforchosenoptionasourdependent variable because it reects the service level thecustomer would receive. As we focus on a context where weexamine whether the participant would retaliate against thecustomerbyofferingalowerservicelevel,weusedthisindirectmeasure of retaliation to reduce the likelihood of response biaswhenusingadirectmeasure.

1.2.3Results

-

The manipulation check indicated that our manipulation ofuncivil behavior was successful [F(2,87) = 103.94,p<0.01].Respondents in the uncivil condition perceived the behavior ofthecustomertobemorerudeanduncivil(M_uncivil=6.24,SD=1.33)thanintheneutral(M_neutral=2.48,SD=1.36)andinthepolite condition (M_polite= 1.93, SD = 1.02, both differencessignicantatp<0.01).Astherewasnodifferencebetweenthepoliteandneutralconditions(p=0.20),weassessedwhetherthese conditions predicted ego threat or interactional justicedifferently before combining these levelsof the behaviorcondition.Wedidnotndthattheseconditionsdifferinhowthey predict ego threat (b=0.04, standard error (S.E.) =0.28,p=0.87)orinteractionaljustice(b=0.06,S.E.=0.29,p= 0.84), and we therefore combined the neutral and politelevelsintoasinglelevel,whichwerefertoasthepolitelevel.

-

We analyzed the data using Hayes PROCESS macro forSPSSusingModel4formediationanalysisusingseatoptionratingasthedependentvariableandegothreatandinteractionaljusticeasmediatingvariables.Wefoundthatthepolitelevelrelativetotheuncivilconditionreducedperceivedego threat of the employee (b=1.91, S.E. = 0.27,p<0.01)andthategothreat,inturn,reducedthequalityoftheseat

assigned tothe customeras reected inthe rating of the seat(b=0.32, S.E. = 0.11,p<0.01). The direct effect of thepoliteconditiononthedependentvariablewasinsignicant(p>0.10),indicatingthategothreatfullymediatedthecustomerbehavior-seatratingrelationship(effect=0.60,bootstrappedS.E.=0.25,bootstrapped95%condenceinterval(CI)[0.17,0.1.17])[1],supportingH2.

-

-

Wefurtherndthatpolitetreatmentbythecustomerreducesperceptionsofinteractionalinjustice(b=3.02,S.E.=0.25,p<0.01)butinteractionaljusticedoesnotimpact

-

-

the quality of the assigned seat (b= 0.08, S.E. = 0.11,p>0.10).Thebootstrappedindirecteffectofcustomertreatmentoftheemployeethroughinteractionaljusticeisnotsignicant(effect =0.23, bootstrapped S.E. = 0.47, bootstrapped 95%CI[1.15, 0.69]). Therefore, interactional justice does notmediatebetweenrudeorpoliteemployeetreatmentandretaliatory behavior. We therefore rejectH3. The mediationmodelisdepictedinFigure2.1.2.4Discussion

The ndings from Study 2 provide further support that FSEretaliationintheformoflowerservicelevelsisdependentonhowcustomerstreattheemployee-uncivillyascomparedtopolitely-replicating the ndings from Study 1. Moreover, thisstudyaddressedtheprocessthroughwhichthisbehavioroccurs. We tested two plausible processes, perceived ego threatand perceived interactional injustice. The ndings show thatego threat is a more compelling process in this context, and wend full mediation of customer behavior on FSE retaliationthroughthisconstruct.Interactionaljustice,however,didnotmediatetheeffect.

The question remains whether all employees are likely toexperience ego threat because of uncivil customer behavior.Notallemployeestargetedbycustomerincivilityarelikelytoengageinanemotional-dysfunctionalcopingbehavior,manifestingasretaliatoryservicebehavior.Alternatively,employeescouldengageinmorefunctionalemotion-focusedcopingbehavior,suchasignoringuncivilbehavior,givingthecustomer"the benetofthe doubt"ordeemingthe rudecustomerunworthyoffurtherattention(BiesandTripp,1995).Importantly,researchshowsthatsomeindividualsarebetter at managing emotions, including negative ones, thanothers (Grandey, 2000;Grandeyet al., 2005;Totterdell andHolman,2003),whichtypicallyisrelatedtothedegreeofEI.

EI is"the ability to acquire and apply knowledge from one'semotions and those of others to produce benecial outcomes"(Kidwellet al., 2011, p. 78). This construct has been studiedextensivelyintheorganizationalbehaviorliterature(JosephandNewman, 2010;O'Boyleet al., 2011;Mayeret al., 2000). Thisresearch proposes that EI is a skill that can be trained ratherthan a stable trait making it extremely powerful during serviceencounters where it can enhance customer relationships andperformanceasitcanincreasedeliberation,adaptationandevenrapport(McFarlandetal.,2016;Delcourtetal.,2013).

Given the benets of EI during interactions as mentionedabove, we suggest that EI allows employees to use theirknowledge of their own and others'emotions, it should also bebenecialwhenfacinguncivilcustomerbehavior.Whilepriorresearch has shown that some individuals are unable to self-regulateafteranegothreat(Baumeisteretal.,1993;Lambird

Figure2Study2:egothreatmediatesbetweencustomerbehaviorandseatrating

and Mann, 2006), we propose that EI allows for this self-regulationprocesstooccur:

H4.EIwillattenuatetherelationshipbetweenperceivedcustomerincivilityandofferedservicelevel.Specically,EIwillreducetheimpactof egothreatonoffered servicelevel,suchthatthisrelationshipisweakerformoreemotionallyintelligentindividuals.

1.3Study3

The objective of Study 3 is to testH4and expand on thendings from Studies 1 and 2. In this study, we use the hotelindustryasaservicesettingwhereFSE-customerinteractionsare important in inuencing the customer experience (Kim andCha, 2002). Additionally, FSE have a signicant impact onperceived service quality and value (Hartline and Jones, 1996).Similar to Study 2, using this context allows us to use anindirectmeasureofemployeeretaliationintheformoflowerservicequalityoffered.

1.3.1Participants

x

Atotalof115English-speakingrespondents(80%female,medianage=47years)participatedinthisstudyusinganonlinepanelinexchangeformonetarycompensation.Respondents wererandomly assignedtoanuncivil,aneutralorpolitecustomer-employeeinteractionlevelwithinthecustomerbehaviorcondition.Therefore,thisstudyusedathree-condition(customer behavior:uncivil,neutralandpolite)measured(EI)between-subjectsdesign.1.3.2Procedure

The respondents were directed to a rst unrelated task thatasked them to answer several questions about themselves. Thissection included the EI scale (Lawet al., 2004), which includessubscales on self-emotion appraisal (a= 0.91), others'emotionappraisal(a=0.90),useofemotion(a=0.87)andregulationofemotion(ROE)(a=0.91)(Appendix3).EIwasmeasuredbeforemanipulatingcustomerbehaviortoensurethatEIwasnot impacted by the manipulation. After this section, therespondents weredirectedtoasecondtask,whichincludedour

manipulation of the customer-employee interaction. Similar tothe airline study, we asked respondents to imagine working atthe front desk of a hotel during a four-day conference. In thisscenario,theywereapproachedbyacustomer,andthecustomerrequestedadifferentroom.Thecustomerhadaneconomy-priced room and wanted a room with a view of thebay instead of the current assigned room; the current room'sviewwasblockedbyaneighboringhotel.Thetoneoftherequest was manipulated to be uncivil, neutral or polite, similartoStudy2(Appendix2).

After reading about the customer interaction, respondentscompletedamanipulationcheckregardingtheperceivedbehavior of the respondent. Then, respondents chose what theywouldofferthecustomer:leavetheroomasis,upgradetoaneconomy plus room with a view (costing $50 more a night),upgradetoajuniorsuite(costing$100moreanight)oraskingthecustomertowaitafewhourstoseewhetheroneoftheothereconomyroomswithapartialbayviewwouldopenup.Aftermaking this choice, respondents completed questions about theinteraction with the customer, the ego threat scale (a= 0.89)and their rating for each of the different room options on aseven-pointscale(1=VeryBadto7=VeryGood).SimilartoourprocedureinStudy2,weusedthissubjectiveratingoftheoption offered to the customer as our measure of employeeretaliation.

1.3.3Results

The manipulation check indicated that our manipulation ofuncivil behavior was successful [F(2,112) = 47.00,p<0.01].Respondentsintheuncivilconditionperceivethebehaviorofthe customer to be more rude and unfriendly (M_uncivil= 5.15,SD=2.20)thanintheneutral(M_neutral=2.03,SD=1.55)andinthepolitecondition(M_polite=1.82,SD=1.17,bothdifferencesweresignicantatp<0.01).Respondentsperceived the neutral and polite condition as similar in terms offriendliness (p>0.10). As in Study 2, these conditions did notdiffer in how they predict ego threat (b= 0.16, S.E. = 0.37,p>0.10).Therefore,wecollapsedthesetwolevelsintoasinglepolitecondition,whichwewillrefertoasthepolitelevel.

WetestedformediatedmoderationusingHayesPROCESSmacro for model 14 with room rating as the dependent variable,customer behavior as the independent variable, ego threat asthe mediator and EI as moderator of the ego threat and roomrating relationship. The results indicated that mediation [F(1,113) = 25.90,p<0.01] and the main effect were statisticallysignicant [F(4,110) = 4.81,p<0.01]. In the ego threatmodel,wendthat,asexpected,thepoliteconditionisassociated with lower perceived ego threat than the uncivilcondition (b=1.59, S.E. = 0.31,p<0.01), indicatingsupport forH2. In the room rating model, we found furtherevidence that EI moderates the relationship between ego threatandroomrating(b=0.07,S.E.=0.04,p=0.08).Therefore,wendmarginalsupportthatEIbuffersthenegativeeffectofegothreat,providingsupportforH4. ToinvestigatethemarginalinteractionbetweenEIandegothreat more closely, we split up the EI measure into its foursubscalesandrananalyseswhicheachofthedimensions separately to test which dimensions are driving themarginally signicant interaction. Only the dimension ROEwas signicant [F(4,110) = 5.66,p<0.01], and this dimensionhadasignicantinteractionwithegothreat(b=0.07,S.E.=0.02,p<0.01).

-

Theindirecteffectofpolitecustomerbehavioronthedependent variable is signicant when ROE is below 1 S.D.fromthemean(effect =0.21,bootstrappedS.E.=0.12,bootstrapped95%CI[0.01,0.48]).WhenROEislow(1S.

-

D.belowthemean),theslopeofegothreatisnegative(b=0.14,S.E.=0.06,p<0.05)butitisnotsignicantforaverageorhighlevelsofROE(bothp>0.10).ThisndingsupportsthehypothesisthatEI,particularlytheemotionalregulationinoneself,bufferstheeffectofegothreatonretaliatorybehavior,butnotforindividualswhoscorelowonemotionalregulation,whichsupportsH3.

1.3.4Additionalanalyses

We further address the question whether EI also moderates thecustomer behavior-ego threat relationship (i.e. the rst part ofthe mediated effect). To test this relationship, we run Model 58of the Hayes PROCESS macro. The results from the modelwith ego threat as the dependent variable [F(3,111) = 11.74,p<0.01] revealed that neither the main effect of EI (p= 0.91)nor the interaction with polite behavior (p= 0.75) is signicant.ThisndingisreplicatedforallEIsubscales.Consequently,EI,specically the ROE, acts as a buffer of whether employeesrespondtoegothreateningbehaviorwithretaliatorybehavior,but it does not impact the extent to which employees areexperiencingegothreatintherstplace.

1.3.5Discussion

This study replicates the nding from Study 2 that perceivedegothreatmediatestherelationshipbetweencustomerincivilityandemployeeretaliationandfurtherextendsthemtoa different service context. Moreover, this study shows that theimpact of perceived ego threat on retaliation is weaker foremployeesthatcanregulatetheiremotions.Therefore,notallemployees are engaging in counterproductive behaviors towardcustomers even when faced with rude behavior. We discuss thetheoreticalandpracticalimplicationsofourresultsnext.