Staff Word-of-Mouth (SWOM) and retail employee recruitment
Kathleen A. Keeling a,1, Peter J. McGoldrick a,∗, Henna Sadhu b,2
a Manchester Business School, The University of Manchester, Booth Street West, Manchester M15 6PB, United Kingdomb The Association of Corporate Treasurers, 51 Moorgate, London, EC2R 6BH, United Kingdom
bstract
Researchers and employers have largely neglected the wider influence of Staff Word-of-Mouth (SWOM), whereas employee referral programsre an established form of recruitment. This paper positions SWOM as a specific form of WOM, communicated by present and former employees,hich can influence potential applicants at the prerecruitment stage. Scenario-based Study 1, with retail employees/applicants, shows differential
ffects on organizational attractiveness of SWOM with positive versus negative messages and tangible versus intangible information, if obtainedrom strong versus weak social ties.
In Study 2, a survey of retail prehires demonstrates mediation and moderation effects on organizational attractiveness of job-seekers’ precom-
itment, fit perceptions with the retailer, and mentoring/aspirational ties. Retailers are urged to develop the potential of SWOM through improved
nderstanding, responding, motivating, and keeping employees informed. The study establishes a conceptual foundation to encourage furtheresearch into SWOM as a communication channel and a means to influence precommitment of prospective employees.
2013 New York University. Published by Elsevier Inc. All rights reserved.
omm
K‘tmp
S
eBare(
eywords: Staff Word-of-Mouth; Employee recruitment; Social networks; Prec
Introduction
Retail employees in the United States numbered 14.77 mil-ion in 2012 with 6.75 million new employees hired in 2011Bureau of Labor Statistics 2012). Replacing even a loweraid retail employee costs around $3637, due to many recruit-ent and indirect costs (Coca∼Cola Retail Research Council).urnover remains high, especially at store level, due in part
o low wages, long and variable hours, and some employeesaving no commitment to stay long-term (Rhoads et al. 2002).hus, there are calls to develop ways to improve cost effec-
ive recruitment communications (Cable and Judge 1996; Rynes991).
Retail employment seekers have a wide choice of formalnformation sources, including local media, employment agen-
ies, and retailer websites; however the importance of socialetworks as informal information sources is a recurring theme,specially for younger and less skilled workers (Finneran and
itment; Information valence; Organizational attractiveness
elly 2003). Granovetter (1995) estimates that 60–90% ofblue-collar’ jobs are found through social networks, abovehe 50–75% general workforce average. Thus, informal com-
unication through word-of-mouth (WOM) in the job seekingrocess is especially relevant for the retail sector.
ituating and defining Staff Word-of-Mouth
Employee referral programs (ERPs) are a well established,mployer initiated recruitment method (e.g., Breaugh 2008;reaugh and Starke 2000) in which current employees identifynd suggest potential candidates from their social networks. Theeferral is normally recorded on the application and the referringmployee receives an extrinsic reward. The Business Dictionary2012) defines an ERP as a “Recruitment method in which theurrent employees are encouraged and rewarded for introduc-ng suitable recruits from among the people they know”, whileambridge Dictionary (2012) defines an ERP as “a system inhich a company pays an employee if someone they know takesjob in the company because of their suggestion.”
Beyond employee referral programs, Van Hoye and Lievens
2009) also discuss a role for more general WOM in the recruit-ent process, which includes information from non-employees
s well as employees. Within this continuum of sources, Fig. 1ituates Staff Word of Mouth (SWOM) as a more specific
contact with ex-employees, providing further opportunitiesto influence SWOM.
Fig. 1. Conceptual and operational positioning of SWOM.
oncept than general WOM but much broader than employeeeferrals. While SWOM includes the social network element ofhe employee referral process, it can also influence and informpplications initiated through other recruitment processes. Thus,e define SWOM as follows:
Staff Word-of-Mouth (SWOM) is the process of staff andformer employees communicating information and opinionsabout the organization, both within and beyond their socialnetworks.
Our rational for considering specifically the SWOM processs that it recognizes the complexity of retail job seeker informa-ion search, where potential employees will themselves initiatenteractions with present and former retail staff. Stores are openo the public, so employee sources are plentiful and the inter-ction can be between people who do not know each other, forxample, when an employee is approached in store.
Fig. 1 depicts the overlaps between WOM, SWOM, andRPs, also the alternative information and process flowsetween employers and potential employees. Retailers clearlynfluence general WOM through their corporate communica-ions and through visible aspects of their performance; some ofhis general WOM may be of relevance to those at preapplica-ion stages. However, information regarding retailers’ actionsess visible to the public, such as staff benefits, fairness asn employer, and keeping promises is of most direct interesto potential applicants. Thus, SWOM is not only dissemi-ated throughout conversations within social networks, it islso specifically sought by many applicants, whether consid-
ring application through an ERP, making a direct approach, oresponding to a job advertisement.
tailing 89 (1, 2013) 88–104 89
Our motivation for focusing on SWOM rather than generalOM is the greater scope for organizations to develop and influ-
nce this job seeker information source. All of a retailer’s actionsnd policies affecting staff can become topics of SWOM conver-ations but there is scope for retailers to become more proactiven understanding and developing this SWOM channel of com-
unication. The wider dotted line in Fig. 1 represents the largelyntapped opportunities for retailers (or any other employers) tonfluence the flow, via staff members, of information likely toeach potential job applicants. This ability to influence prospec-ive employee perceptions directly through SWOM represents,n managerial terms, the point of potential development and anbjective of this research.
Regarding the potential of SWOM for cost-effective com-unication, there are analogies with general WOM, where
ompanies go beyond dependence on reputation and media com-unications, for example, to engage in viral marketing (Hinz
t al. 2011), targeting “mavens” who are most likely to spreadhe message (Higie, Price, and Feick 1987), or both. Thus, with-ut replicating the formal referrals and rewards of ERPs, retailersan be far more proactive in understanding, informing, and influ-ncing messages within the SWOM communication channel,hile also seeking to identify and, when possible, reduce levelsf employee dissatisfaction that lead to negative SWOM.
WOM and ERPs compared
While acknowledging overlaps between ERPs and SWOM,hese two processes differ in many conceptually and strategi-ally important respects. Fig. 1 lists eleven particular points ofifferentiation between applications made through ERPs andhose influenced only by the broader processes of SWOM; weummarize these differences very briefly below.
1. ERPs have formal administrative processes concerning pro-gram scope, vacancies targeted, scale and terms of referralrewards, usually after a given period of successful serviceby the referred employee (Rankin 2008; Sullivan 2011). Incontrast, SWOM comprises largely informal processes, notadministered but could be influenced more proactively bycompanies.
2. ERPs are initiated by employers (Rankin 2008; Sullivan2011), which define the scope of the ERP. SWOM is a pro-cess initiated by current or former staff, or by prospectiveapplicants. In the second of our studies, most prospectiveapplicants who had initiated SWOM went to stores to talkwith staff, or talked with social contacts directly, by tele-phone, by e-mail, or both.
3. ERPs involve mainly current employees, although Rankin(2008) and Sullivan (2011) now recommend using also“staff alumni”. Due to high levels of retail staff turnover,many former employees can provide valued SWOM, anal-ogous to ex-customer WOM. Some companies maintain
4. ERPs may not include the post(s) of interest to a poten-tial applicant, as ERPs normally prioritize certain posts at
9 of Re
1
1
E
wmm
o2ctro(t
amszcpp
F
iftUtcwjaasato
cbt(T“sweotwL
aspPm
0 K.A. Keeling et al. / Journal
any given time (Sullivan 2011). General SWOM is spreadconstantly and prospective applicants may seek SWOMopinions on job opportunities at any time.
5. Business Dictionary (2012) and Cambridge Dictionary(2012) mention ERP referrals of known people. Refer-rers lose financial rewards and possibly reputation if theirreferred hire is not successful. Potential applicants outsidestaff social networks often seek SWOM; many of our study 2respondents had visited stores to speak to (largely unknown)staff members.
6. ERPs create a specific context for interactions with a focuson recruitment. As jobs are a major part of many people’slives and conversations, general SWOM conversations willalso influence organizational attractiveness prior to deci-sions to apply. Researchers call for more research of thisstage (e.g., Breaugh and Starke 2000; Collins and Stevens,2002).
7. SWOM generally brings no extrinsic reward. Conversely,terms such as “bonuses”, and “bounty schemes” (PersonnelToday 2009) reflect the importance of financial motives inERPs, usually individual rewards (Rankin 2008), occasion-ally prize draws (Sullivan 2011).
8. Those benefitting personally from persuasion, as in ERPs,are less likely to be credible (Breaugh and Starke 2000);those representing organizations often show informationbias (Wanous and Colella 1989). SWOM (like WOM) usu-ally comprises personal, unpaid communications, hencemore credible (Arndt 1967; Grewel, Cline, and Davies2003).
9. Fernandez and Weinberg (1997) note ERP incentives ofbetween $200 and $1000, even for entry level bankappointees. The UK Industrial Relations Surveys showa median ERP reward of $1130 (Rankin 2008). SWOMinvolves no direct costs, although companies might in futureincur some modest costs to monitor and to enhance SWOMeffects.
0. Diversity issues with ERPs are a recurrent theme, as refer-rals tend to be within familiar social networks (Schneider1987), leading to “strong inbreeding biases between indi-viduals of the same race, religion, sex, age, and education”(Montgomery 1991, p. 1413). As noted above (point 5),SWOM is often sought by strangers, so has less diversityissues.
1. ERPs are well established systems in many companies(Rankin 2008; Sullivan 2011), attracting extensive atten-tion in the management (e.g., Breaugh and Starke 2000),behavioral (e.g., Fernandez and Weinberg 1997) andlabor economics (e.g., Kugler 2003) literatures. In con-trast, SWOM offers extensive development potential forresearchers and managers.
arly recruitment stages and impact of SWOM
WOM (including within the referral process) is identifiedith early-stage recruitment activities, such as generating andaintaining applicant interest, areas of ‘critical need’ for recruit-ent research (Breaugh 2008). Despite some work on the impact
atBi
tailing 89 (1, 2013) 88–104
f early-stage recruitment activities (e.g., Collins and Stevens002). Van Hoye and Lievens (2005) believe that effects ofompany-independent sources of job information on organiza-ional attractiveness or job pursuit constitute a gap in recruitmentesearch. Barber (1998) also identifies as crucial the initial phasef recruitment, when seeking to attract applicants. Breaugh et al.2003) argue “more attention needs to be given to applicants ashey move through the recruitment process”.
Thus, understanding the effects of SWOM on maintainingpplicant interest is important, as perceptions in early recruit-ent stages serve as “anchors for evaluating organizations”,
trongly predicting applicants’ eventual attraction to an organi-ation as an employer (Cable and Turban 2001). Many possibleandidates reject a company at this stage, without any com-any awareness, underlining the need for companies to be moreroactive in understanding, informing and targeting SWOM.
actors affecting applicant interestTalking to current and past members of staff is an important
nformal source of difficult-to-obtain information about the jobor potential employees (Fernandez and Weinberg 1997), buthis source is not under the direct control of the organization.ncontrolled sources can undermine or reinforce company con-
rolled advertising (Van Hoye and Lievens 2005) as they willontain both positive and negative content (message valence),hich may have encouraging or detrimental effects. Further,
ob seekers may desire several distinct types of informationbout employment attributes and weight these differently, suchs reward mix and working conditions, or company attributes,uch as their performance on core values or ethical issues (Judgend Bretz 1992). Job seekers may perceive that some informa-ion is best obtained from current or former employees of therganization, that is, through SWOM (Cable and Turban 2001).
In addition to message valence, Breaugh et al. (2003) call forloser examination of other factors changing the relationshipetween messages from employees and organizational attrac-iveness, such as the relationship between source and receivertie strength) (Brown and Reingen 1987; Levin and Cross 2004).he tie strength between the recipient and the WOM sourceis a combination of the amount of time, the emotional inten-ity, the intimacy (mutual confiding), and the reciprocal serviceshich characterize the tie” (Granovetter 1973). These conditions
ncourage trust building, so the job seeker can gauge the valuef information from the employee. Previous research supportshese suppositions, linking tie strength and source credibilityith effective knowledge transfer (Brown and Reingen 1987;evin and Cross 2004).
Recognizing the recruitment process complexity (Breaughnd Starke 2000), a further factor capable of changing the mes-age – receiver outcome relationship is the level of job seekerrecommitment to the job, organization, or information source.recommitment factors include desire to enter the sector, assess-ent of the fit of the job and organization with own needs
nd values, or assessment of the specific benefits of the rela-ionship (e.g., aspirational/mentoring ties) (Breaugh et al. 2003;retz and Judge 1998). This is important as motivated reason-
ng theories (Jain and Maheswaran 2000; Kunda 1990) predict
of Re
tc2ip
O
o
1
2
3
4
tawa(stjatt(sah1K
epmaoiTj1i
oc
o2(itwrOt1
jmWc(cadrIspcvtim
T
fL(tHetwtLotapM
K.A. Keeling et al. / Journal
hat positively motivated or precommitted people will discountounter-attitudinal information (Khare, Labrecque, and Asare011), so are less likely to adjust attitudes accurately after receiv-ng pertinent information (Breaugh 2008) than those with lessrecommitment.
bjectives of the studies
Based on the arguments advanced above, the main objectivesf this investigation are:
. To develop a solid theoretical grounding, recognizing thecomplexity of the recruitment process (Breaugh and Starke2000) and integrating previous research to conceptualizepotential modifiers of SWOM message effects.
. To assess the influence of different SWOM information typeson retail organizational attractiveness (OA), based on a sce-nario study and survey evidence, to understand better how jobseekers process information from different sources (Barber1998; Rynes 1991).
. To investigate effects of precommitment as a modifier ofSWOM message effects on perceived OA based on a surveyof job seekers in the retail sector.
. To develop an awareness of the potential and processes ofSWOM, providing a basis for retailers and others to developthis channel of communication to prospective employees.
This paper reports on two studies. Study 1, an experimen-al study informed by literature and the results of a prestudy,ssesses how variations in SWOM message content interactith the message valence and source to alter perceptions of
n employer, hence employment choices in the retail sectorobjectives 1, 2, and 4); social exchange expectancies and theearch/experience/credence paradigm are employed to explainhe effects. Study 2 is a cross-sectional field study of retailob seekers during the early phase of gathering informationbout a job opportunity and deciding to apply, including howhree precommitment factors might help protect against nega-ive information, affecting source credibility influence on OAobjectives 1–4). Theories of biased processing, such as deci-ion precommitment, outcome dependence, issue involvement,nd the diagnostic value of negative information are relevantere (Ahluwalia, Burnkrant, and Unnava 2000; Berscheid et al.976; Dholakia and Sternthal 1977; Johnson and Eagly 1989;hare, Labrecque, and Asare 2011; Kunda 1990).Due to the high likelihood of job seekers talking to
mployees, employers may be able to strengthen job seekerrecommitment factors and diminish or amplify the effects ofessage type and message valence on assessments of OA. This
nswers calls to establish components of organizational imagef interest to retail job seekers that can be “cost-effectively mod-fied or communicated” (Cable and Judge 1996; Rynes 1991).
here is also interest in gaining better understanding of how
ob seekers process information from different sources (Barber998; Rynes 1991), especially for employers wishing to exam-ne the potential of SWOM to influence employee acquisition,
pbce
tailing 89 (1, 2013) 88–104 91
r consider actions that may increase prospective employee pre-ommitment to the job or organization.
Conceptualization
Prior research establishes the potent effects of general WOMn perceived organizational attractiveness (Collins and Stevens,002) over and above influence from other recruitment sourcesVan Hoye and Lievens 2009). Negative WOM produces greatermpact than positive and can interfere with recruitment adver-ising effectiveness (Van Hoye and Lievens 2005). This greatereighting for negative as opposed to positive information
eflects a general negativity effect documented in other domains.ne reason is the informational or diagnostic advantage of nega-
ive compared to positive information (Skowronski and Carlston987).
Nevertheless, prospective applicants do not always acceptob information at face value, assessing the credibility of both
essage and source before accepting employment information.hen the source is a strong-tie, information is considered more
redible and useful, thus more influential, than if from a weak tieBreaugh and Starke 2000; Van Hoye and Lievens 2007). Sourceredibility is a function of expertise and trustworthiness (Cablend Turban 2001; Pornpitakpan 2004), so credibility perceptionsepend on job seeker assessments of the knowledge and expe-ience the source has of the organization (Bone 1992). Fisher,lgen, and Hoyer (1979) find an interaction between messageource and message valence. Generally, more trust accrues toroviders of negative as opposed to positive information butlose ties are considered knowledgeable, regardless of messagealence. To summarize, information valence, relationship withhe source (tie strength), and source credibility influence thempact of information on OA; negative information and infor-
ation from (credible) close ties is generally most influential.
angible and intangible attribute information
However, a consideration of the type of knowledge trans-erred requires further interpretation of these associations.ievens, van Hoye, and Schreurs (2005) find intangible attributes
trait inferences) explain more variance (41%) in OA than moreangible job and organizational attributes (35%). Lievens andighhouse (2003) differentiate between two types of retail
mployer attributes. The first, variously characterized as objec-ive, tangible and concrete, concerns pay, benefits, type ofork and promotion prospects, all well documented as impor-
ant to employee job choice and retention (Arndt, Arnold, andandry 2006; Bettencourt and Brown 1997). Beliefs about levelf reward for effort are positively related to retail employeeurnover (Rhoads et al. 2002). Tangible attribute informationlso includes terms and conditions of employment, mental andhysical job demands, and promotion of employee well being.uch of this information is available and verifiable from com-
any and other formal sources, as well as informal sources,ut some information on hidden practices regarding workingonditions can come only from employees with direct personalxperience, such as compliance with cleanliness or food safety
9 of Re
rthev2
aamro
cophtw
tb1opvcac
sds
Im
e(ctoatwws
cmad(uatj
swoc
jeopeeo(bmfoawrfmesfsmrp
olsfhcte
dakttts
tTieia
2 K.A. Keeling et al. / Journal
egulations, working time norms (e.g., ‘voluntary’ unpaid over-ime, pressures to work split shifts) and protection from physicalazards or unreasonable customer behavior (McColl-Kennedyt al. 2009). Employees are an important source of this lessisible information on company practices (Cable and Turban001).
The second type of attribute is characterized as subjectivend intangible, such as the reputation of the organization asn employer or in general, often conveying symbolic infor-ation (Lievens and Highhouse, 2003). Such information on
Concerns about fitting in with colleagues, training, andhallenging work situations drive the value of information onrganizational climate (Parker et al. 2003). Front-line retail stafferceptions of supervisor support reduce the stress of unsocialours, heavy workloads, and difficult customers, and are linkedo positive appraisal of the environment (Babin and Boles 1996),hich is strongly related to OA (Chapman et al. 2005).Benefits of organizational climate and reputational informa-
ion to the employee are the ability to judge perceived similarityetween personal needs and organizational standards (Schneider987), or to reflect prestige, social identification and signalingf particular skills and values through association with the com-any (Carmeli, Gilat, and Waldman 2007). Employees are aaluable source of information on reputation and organizationallimate, as they have the experience of whether internal normsnd behaviors are congruent with the values externally publi-ized.
Study 1 therefore looks beyond the established effects of mes-age valence and tie strength in an initial examination of whetheriverse types of message content interact with valence and tietrength, when job seekers assess OA.
nteractions between message content, tie strength, andessage valenceEmployer attractiveness is multi-dimensional and employ-
es look for a range of benefits when considering future jobsCollins and Stevens, 2002). The distinction between objective,oncrete attributes and subjective, intangible attributes is impor-ant, because information on pay, benefits, and type of work isbservable and relatively easy to find, compared with reputations an employer. Less observable information, such as how wellhe company actually lives up to its external reputation or socialork ‘climate’, can come only from people who have actuallyorked for the company, that is, staff, past or present, and is
ubsequently difficult to verify.This distinction has parallels in the search, experience and
redence product attribute differentiation and consumer judg-ents about these information types. Lim and Chung (2011)
rgue that, for consumers, the credibility of the source of cre-ence (intangible) attributes is more important than for searchtangible) attribute communication, as search attributes can
sually be verified or acquired from other sources, includingdvertising. For credence attributes, the only source of informa-ion may be the reports of other people on performance. In theob search context, less observable types of information, e.g.,
urma
tailing 89 (1, 2013) 88–104
ocial relations at the workplace, can come only from peopleho have worked for the company, so are parallel to experiencer credence attributes. More tangible aspects, such as pay andonditions, are similar to search attributes.
Information from strong ties is generally more credible, butob seekers also bring into play their perceptions of speakerxpertise and trustworthiness and social expectations of the typef information associated with weak or strong ties. Clark (1984)osits that the strength of ties linking exchange participants influ-nces perceptions and motivations, therefore behavior, in socialxchange. Strong ties carry more responsibility for the welfaref others; network members keep track of each other’s needsClark, Mills, and Powell 1986). Speaker motivation is towardeing helpful and accurate, so information from close ties isore likely to be context specific, procedural and tacit than if
rom weak ties. Tacit knowledge is difficult to express and sharedften through storytelling, involving high levels of interactionnd mutual understanding (Zack 1999), and most likely to occurith close ties. Hence, information from close ties will be more
elevant and social, exposing aspects of the job typically notound through formal sources, such as the organizational cli-ate or how well reputational promises are kept (intangible,
xperience/credence attributes). In contrast, exchange relation-hips are typical of weak tie communications, where individualseel no special responsibility, so information content may beubsidiary to self-interest (Frenzen and Nakamoto 1993). Infor-ation from weak ties is more likely to be general and explicit
ather than tacit, we would argue, and so more likely to relate toay, benefits, and type of work (tangible, search attributes).
Past experiences shape expectations of the content and naturef interactions. Hence, in weak tie interactions, receivers haveow expectations for context related, personal experience mes-ages (e.g., about individual relationships with supervisors orellow employees); unlike strong tie interactions, there is noistory of mutual confiding, emotional intensity, and recipro-al services (Granovetter 1973). Burgoon and Hale (1988) posithat disconfirmation of social interaction expectancies activatesvaluation and attribution of intention and motive.
Hence, negative personal experience information about cre-ence/experience job attributes emanating from a weak tierouses difficult to resolve questions of motive, as a lack ofnowledge about the weak tie makes for difficult assessment ofhe credibility of information about personal experiences. Onhe other hand, listeners can check negative, explicit informa-ion, such as, pay packages and conditions of work (tangible,earch attributes), against other sources.
Conversely, people expect and value from close ties informa-ion with tailored, social and detailed content (Clark et al., 1986).hus, we might posit that receivers find SWOM content on more
ntangible experience/credence attributes more unusual and notxpected from weak ties, and vice versa for tangible attributenformation from strong ties, where people expect more tailorednd socially orientated information. Unexpected content is less
seful to the receiver than expected SWOM content, as it mayaise reservations regarding motive and credibility. In this case,essage content effect offers a refinement to the Fisher, Ilgen,
nd Hoyer (1979) findings regarding message valence and tie
of Re
sm
jd(ae
Hjto
Hrwi
S
ttdoeft
psfeasemcbar
T
droo1rwaztt
Hced
S
wBwcccg(
Hwmi
T
iip2oIiataachc2updoOp
He
tdhc
K.A. Keeling et al. / Journal
trength: weak tie, negative search attribute SWOM could beore influential than from strong ties.The following hypotheses summarize these arguments that
ob seekers’ perceptions of job information are interactive andiffer according to combinations of tie strength of the sourcestrong/weak), the valence of the message (positive/negative),nd the content of the message (tangible search or intangiblexperience/credence attribute information), such that:
1a. When the SWOM message contains more intangibleob experience content (experience/credence attributes), nega-ive information from a strong tie source has a stronger influencen OA than the same information from a weak tie source.
1b. When the SWOM message contains more tangible jobelated content (search attributes), negative information from aeak tie source has a stronger influence on OA than the same
nformation from a strong tie source.
tudy 2 hypotheses
Study 2, a field survey of actual job seekers, continueshe exploration of difference in impact between informationypes, considering first the relative importance of pay and con-itions (tangible, search attributes) information compared torganizational climate and reputational information (intangible,xperience/credence attributes) for OA, also assessing any dif-erence in source credibility effects in conjunction with thesewo information types.
The exploration then broadens by considering how job seekerrecommitment factors might affect relationships between mes-age content and source credibility, and perceived OA. Theeatures chosen to operationalize precommitment (to retailmployment, an organization, or a message source) providen illustrative range and potential applications to employertrategies. All can be influenced by organizations as potentialmployers and so meet calls to improve cost effective recruit-ent communications (Cable and Judge 1996; Rynes 1991), in
ontrast to other employee motivations to seek work promptedy macro-economic factors, such as financial recession and jobvailability. Further, they cover both outcome based and valueelevant involvement (Johnson and Eagly 1989).
angible versus intangible attribute informationPay packages and working conditions are important in job
ecisions but, in the long term, retail work is a social envi-onment, where the support of fellow employees is vital to jobutcomes and even well-being. Employees also seek work inrganizations that fit their personal values (Judge and Cable997). Thus, employees value information on workplace socialelations as signals of how well they could integrate with felloworkers and job expectations, using reputation information as
heuristic for the fit between their personal values and organi-
ational values. Lievens et al. (2005) support this view, findinghat intangible attributes explain more variance (41%) in OAhan more tangible, concrete job attributes (35%). Hence:
fthi
tailing 89 (1, 2013) 88–104 93
2. The impact of information valence on OA is stronger whenonsidering organizational climate and reputational (intangible,xperience/credence) information, compared with pay and con-itions (tangible, search attributes) information.
ource credibility and information typeThe persuasion benefits of highly credible sources are
ell established (Jain and Posavac 2001; Pornpitakpan 2004).reaugh and Starke (2000, p. 410) note that communicationshere the source benefits from misleading the receiver will lack
redibility, whereas most SWOM is not associated with finan-ial benefit for the employee. Evidence also suggests that sourceredibility has distinctive effects on acceptance of search (tan-ible, concrete) or experience attribute (intangible) informationJain and Posavac 2001), such that:
3. The relationship of source credibility to OA is strongerhen considering pay and conditions (tangible attribute) infor-ation, compared with organizational climate and reputational
nformation (intangible attribute).
he role of precommitment factorsResearch indicates that goal directed (motivated) reason-
ng biases cognitive outcomes (Kunda 1990). Precommitments a powerful motivation in determining resistance to counter-reference information (Ahluwalia, Burnkrant, and Unnava000); hence, motivations to maintain pre-existing preferencesr valued goals may modify the effects of negative information.n consumer research, decision precommitment is implicatedn consumer responses to WOM messages (Khare, Labrecque,nd Asare 2011). Stronger preferences result in a recall advan-age for positive over negative information as the former is moreccessible, prevalent and better organized, so aiding the recalldvantage for positive over negative information, which aidsounter-arguing. Thus, low-commitment consumers will showigher weighting of negative information compared to high-ommitment consumers (Ahluwalia, Burnkrant, and Unnava000; Khare, Labrecque, and Asare 2011). We expect individ-als to hold greater or lesser precommitment to a retail role,articular retailer or message source and that this will lead toiscounting (or not) of counter-preference information from anyne individual and moderate the outcome of this information onA in a similar way, that is, show interaction effects betweenrecommitment factors and information valence on OA.
4a. Precommitment factors moderate information valenceffects on OA.
Nevertheless, we must consider an alternative explanation forhe effects of precommitment, based on the premise that outcomeependence enhances liking (Berscheid et al. 1976). A different,igher baseline for OA for those with stronger precommitmentompared to those with lower precommitment, could account
or differences in the effects of information valence. If this ishe case, we would expect no moderating effects, as there is noypothesis of variation in treatment of preference consistent ornconsistent information, rather we can expect mediation effects.
9 of Re
He
aaKpto
ilgpnfMop
ciaieovhaoal
Hiii
M
P
rciarfl4
fiv
iae4
S
ofdfnfviipoeafic(oio
P
ai(aosapoatq
M
fam
4 K.A. Keeling et al. / Journal
4b. Precommitment factors mediate information valenceffects on OA.
Of course, it is possible that both processes are in operationnd will be observable, such a model is a case of moder-ted mediation, termed a conditional indirect effect (Judd,enny, and McClennand 2001).3 That is, individuals with higherrecommitment may tend toward a higher baseline for organiza-ional commitment, but this effect is conditional on the valencef the information.
Precommitment may also mediate the effects of source cred-bility. When people are highly involved with an issue, they areess sensitive to cues regarding source credibility, leading to aeneral expectation that involvement should mediate credibilityerceptions (Dholakia and Sternthal 1977). For some time, orga-izational commitment has been regarded as an attitude resultingrom identification and involvement with the organization (e.g.,
owday, Steers, and Porter 1979), hence precommitment to therganization may also mediate employee information credibilityerceptions.
Moreover, Johnson and Eagly (1989) argue for differentonsequences for credibility effects between outcome relevantnvolvement (salient to currently important goals and outcomes)nd value relevant involvement (reflects on important and endur-ng aspects of the self). The latter typically inhibits credibilityffects (Johnson and Eagly 1989), hence the highest influencef precommitment is expected in factors connected to personalalues. The notion of person-organization fit is most relevantere, as measures of fit often include reference to values (Judgend Cable 1997). On the other hand, an interest in the retail roler perceptions of the benefits of mentoring or aspirational tiesre more akin to important goals and outcomes, hence we expectower mediation of credibility.
4c. Precommitment factors mediate the effects of credibil-ty of information from the staff member on OA; this effects strongest when precommitment is based on value relevantnvolvement.
Study 1
ethods
articipantsA large US online market research panel, owned by a global
esearch agency, provided a quota sample of 362 respondents tooncur with the USA working age gender and age demograph-cs. Eligibility prescreening assured high self-relevance (that is,s working or seeking employment in the retail sector), all wereesponse integrity checked, including double opt-in recruitment,
raud and duplication prevention and monitoring for ‘straightine clicking’ and completion time thresholds. The sample was8% male, 23 percent were seeking retail work; those employed
3 Judd, Kenny, and McClennand (2001) note that it is not uncommon tond conditional indirect effects. Residual direct effects (in this case, messagealence) are also commonly reported.
ttsps
gc
tailing 89 (1, 2013) 88–104
n retail indicated their positions as shop-floor, clerical, or juniordministration (64%), as management (28%), or other (8%) cat-gories. Age ranges are 18–24 (13%) 25–34 (20%) 35–44 (25%)5–54 (22%) 55–65 (20%).
timuli developmentA prestudy (eight male and ten female retail staff, age 18–59)
f recruitment stories about when advice was received or soughtrom an employee of a prospective retail employer, followed byual moderated in-depth group discussions, formed the basisor scenarios concerning different categories of positive andegative SWOM. The development process was supported byeedback both from job seekers and industry experts. We createdery positive or very negative messages to prevent ambigu-ty about either personally experienced, intangible job relatednformation or more impersonal job performance information,rovided by a fictional employee stated to be well-known (A)r less well-known (B) to the respondent. ‘Intangible’ experi-nce/credence information included interpersonal relationshipsnd group cohesion at work, supervision, and company concernor worker comfort; the ‘tangible’ search attribute informationncluded promises kept on salaries, staff discounts, bonus poli-ies, flexible working hours, and other family friendly policiessee Appendix A). To complete the scenario, respondents readne of two profiles of the employee providing the SWOM,ncluding their relationship and how long they had known eachther.
rocedureThe Study 1 research instrument was an online questionnaire
nd the design a 2 (positive vs. negative message) × 2 (morentangible message vs. more tangible message content) × 2close vs. weak tie) manipulation. The agency randomlyssigned participants to one of these eight scenario combinationsf tie strength, information valence and information type, but alltarting with identical company information in the form of a jobdvertisement. To promote realism, actual job notices from com-anies in the retail sector provided a basis of a general descriptionf a fictitious retailer, including common claims about benefitsnd rewards found in job advertisements. Participants viewedhe information for one scenario and then completed the onlineuestionnaire.
easures
The Organizational Attraction measure adapts nine itemsrom Highhouse, Lievens, and Sinar (2003 p. 992) concerningssessment of the company as “a potential place for employ-ent”, covering reputation, popularity, status and intentions
owards the company (see Appendix B). Other questions cap-ured gender, age, country of origin, and occupation. Testamples of 50 people in the USA and UK, plus a small sam-le of retail professionals, provided feedback on question and
timuli clarity.
For OA, removal of one item (see Appendix B) resulted in aood fit of the data for the original three components of theonstruct (Highhouse et al. 2003) (see Appendix B): CMIN
K.A. Keeling et al. / Journal of Retailing 89 (1, 2013) 88–104 95
Table 1Three-way interaction: tie strength, information valence and information type.
Information type Intangible information(organizational climate)
Tangible information (payand conditions)
Test of 3-wayinteraction
Tie strength Weak Strong Weak Strong ANOVA
I
O
6(rnsttatTm.fpr
S
motpaad
T
sv
mstmthiSstOfer
O
rfUimstsa
nformation valence Neg Pos Neg Pos
rganizational attractiveness: means 5.12 7.53 4.66 7.69
3.77, df 17, p < .01 CMIN/DF 3.75 GFI .944 RMSEA .065.049–.083) p = .063. These three components were highly cor-elated with each other (.955, .903, .877), suggesting these areot truly separate constructs; however, a first order one factorolution produced a degradation of fit. The alternative is that thehree components contribute to a single, second order factor, inhis case, overall OA. This second order model is as constraineds the initial model; the advantage is a structure accounting forhe interrelationships among the factors (Brown 2006, pp. 320).he second order model produced no appreciable reduction inodel fit and, as all three factors load significantly (.97, .91 and
98) onto the second order factor (Brown 2006), it accounts wellor the first-order factor correlations. The three first order factorsrovide a composite variable of overall OA with AVE 98% andeliability .98; coefficient Alpha is .94, mean 6.19, SD 1.93.
tudy 1: analyses and results
Data screening confirmed there were no missing data, butultivariate outliers and consistency checks suggested removal
f 38 cases. Manipulation checks confirm appropriate percep-ions of each scenario: differentiating between negative andositive messages (t = 25.86, df 295.78, p < .01), pay/conditionsnd organizational climate information (t = 4.51, df 321, p < .01)nd perceived tie strength with the fictional employee (t = 7.03,f 321, p < .01).
esting Hypotheses 1a and 1bAn ANOVA with OA as the dependent variable and mes-
age valence, message content and tie strength as independentariables revealed significant main effects for valence and
aflo
Fig. 2. Interaction of tie strength, informa
Neg Pos Neg Pos F p
5.29 7.54 5.91 7.61 35.4 <.01
essage content. A test of the model required by the hypotheseshows a significant three-way interaction between valence, con-ent and tie strength (Table 1; Fig. 2). Supporting H1a, when the
essage contains intangible experience/credence attribute con-ent, negative information from a strong tie source (mean = 4.66)as a stronger negative influence (resulting in a lower OA rat-ng) than the same information from a weak tie source (5.12),upporting H1b, when the message contains more concreteearch attribute content, then negative information from a weakie source has a stronger negative influence resulting in lowerA (5.29) than the effect resulting from the same information
rom a strong tie source (5.91) (see Fig. 2). There is no mainffect of organizational level of the respondent (i.e., managerialole or not) (F = 036; p = .849).
verview: Study 1
In the controlled conditions of an experimental setting, theesults of Study 1 confirm that message content is an additionalactor in the impact of SWOM on job seeker assessments of OA.ntil now, messages from a strong tie source were assumed more
nfluential than from a weak tie source. Study 1 demonstrates thisay not always be the case, depending on the content of the mes-
age. The experimental study controls effects of familiarity withhe company and, through the planned manipulations, demon-trates the value of distinguishing between types of informations experience/credence job attribute information and search
ttribute related information about pay packages/conditions. Theormer is difficult to obtain and verify from other sources; theatter, while also based on personal experience, can be mostlybtained and verified from a wider variety of sources.
tion valence, and information type.
9 of Re
ieswcsdssma
M
P
tai(aaaotarhe
P
atmocpwaiOgcetfcjtml
tem(bnfiiitfftsucwamarSnamqKirb2
M
o00aaw2ts1
p(rv
6 K.A. Keeling et al. / Journal
Nevertheless, experimental work suffers from external valid-ty limitations, as the situation is artificial and cannot captureffects in the actual job search process or long term relation-hips. Further, there is difficulty in accounting easily for realorld job seeker motivational factors that might increase pre-
ommitment to a particular job or organization. Hence, this firsttudy lays the foundation for a second study amongst respon-ents who are actually at the information gathering stage of jobearch and explores further the effects of message content andource credibility, together with three possible precommitmentotivational aspects affecting the influence of information frommember of staff on OA.
Study 2
ethods
articipantsData were collected from a US national sample through invi-
ations sent to a random selection of 4837 panelists of workingge across the USA, identified within the database as workingn the retail sector but not in a managerial capacity. Respondentsn = 250) represent a wide range of ages (18–69; mean 36 years)nd retail experience (51% up to 5 years; 18% over 10 years),good gender split (female 59%), 61% had no college degree,
nd 66% were employed. Nearly 13% of respondents reportedn their discussion with a former rather than a current employee:his shows no systematic relationships with demographic vari-bles, the valence or type of information talked about, or theesults of the conversations. Former employee informants areowever given a higher rating for tie strength than currentmployees.
rocedureA series of prescreening questions ensured respondents were
ctively looking for work in the retail sector within a 30-dayime frame, had found a job opportunity, had talked to one or
ore members of staff about that job, but reported no interviewr receipt of a job offer at this stage. A number of distracterues within the screening process helped ensure the sample com-rised only those genuinely interested in retail, e.g., respondentsere asked if they were looking for work in any of three sectors in
ddition to retail, namely health care, retail banking and cater-ng, each specifying the type/level of jobs (non-managerial).nly those choosing retail were allowed to continue. Demo-raphic and human capital variables are relevant in job choiceontexts; however, Judge and Bretz (1992), using age, gender,ducation, job experience and job offer receipt, found the lat-er the only consistently significant control variable (accounting
or 24% of variance in OA). This aspect is, therefore, alreadyontrolled in the present study, as data are collected only fromob seekers reporting no interview or job offer from the prospec-ive employer. Thus, remaining control variables are age (years),
onths of retail role experience and education level (some col-ege or more versus no college).
tcjdas
tailing 89 (1, 2013) 88–104
At subsequent steps, sufficient choices were given thathe focus of the questionnaire was not self-evident, to helpnsure the validity of the final sample. To help avoid com-on method variance (CMV) and self-generated validity (SGV)
Feldman and Lynch 1988) respondents were screened toe within the job search process and the initial question-aire items prompted respondents to make accessible theirrst hand, firm-specific SWOM knowledge before respond-
ng to the scales (Kleine, Kleine, and Kernan 1993). Thesencluded: the age of the informant staff member comparedo the respondent, whether the staff member is a current orormer employee of the retail company, works in the sameunction and at the same/lower/higher organizational level thanhe job being considered, and a measure of their relation-hip with the informant before discussion of the job/companynder consideration. Procedures were implemented to restrictonnections between and access to prior answers. Measuresere mostly on separate pages, with no back button use
llowed; different scaling formats were used, with separation ofeasurement in the questionnaire where possible (Mackenzie
nd Podsakoff, 2012), all serving to interrupt respondentesponse style. Further, in this investigation the effects ofGV may be alleviated, as it concerns vivid and direct cog-itions on ‘central life interests’ (Lichtenstein, Netemeyer,nd Maxham 2010, p. 430). Such cognitions are consideredore accessible and “more diagnostic of how to answer a
uestion than is their answer to an earlier question” (Kleine,leine, and Kernan 1993). The marker variable approach
ndicates that CMV effects between individual variable cor-elations fall well below the ‘reasonable range’ of up to .15efore CMV becomes problematic (Malhotra, Kim, and Patil006).
easures
Valence of information assessments came from single itemmnibus measures with continuous rating scales (anchors–100). Technology assists by reading the exact point from–100, where a cursor is placed along the line, according torespondent’s judgment. Single item scales are considered
ppropriate for omnibus measures of psychological constructshere holistic assessments such as these are required (Rossiter002). Such measures reduce redundancy in measurement andhe respondent task, enhancing their attention; these mea-ures are often used in recruitment research (Judge and Bretz992).
The precommitment features operationalizing this factorrovide a range of outcome and value relevant involvementJohnson and Eagly 1989). Outcome relevant goals are rep-esented firstly, by an interest in the retail role providingarying levels of precommitment based on general expecta-ions of retail careers (Hurst and Good 2009) formed duringareer exploration independent of information about a particular
ob opportunity. As discussed above, the measure for respon-ent interest in the retail role also employed a single item withnchors ‘not true of me at all’ to ‘very true of me’ (7-pointcale).
of Re
(manipmrpcaa(
rtiiktjfipB
iCeie
f((.Ald
pCAt(
S
alt(ciat
difb(sea
T
poaor5
hmOpc
T
si(isv(2efmem
f2vcupois, moderated mediation, or conditional indirect effects (Judd,Kenny, and McClennand 2001). That is, the precommitment
4 Consistent with previous studies (Judge and Bretz 1992); the contributionfrom the control variables is small, in the interests of parsimony, these are not
K.A. Keeling et al. / Journal
Secondly, personal goals often drive communication choicesGraham, Argyle, and Furnhan 1980). A prospective applicantay seek relationships with employees to provide support in
chieving employment and through early career, but they areot necessarily friends (Shah 1998). The investment in build-ng such mentoring or aspirational tie relationships enhancesrecommitment to the decision object and so goal-directedotivated reasoning (Khare, Labrecque, and Asare 2011). To
epresent dimensions of network contact benefits in the preem-loyment stage for a mentoring/aspiration measure, four itemsover perceptions of the employee informant as provider ofccess to information and resources, career sponsorship and asrole model (Dreher and Ash 1990; Ragins and McFarlin 1990)
Appendix B).The notion of person-organization fit is germane to value
elevant involvement, as measures of fit often include referenceo values (Judge and Cable 1997) and play a substantial rolen job pursuit intentions (Chapman et al. 2005). Before receiv-ng SWOM information, job seekers may use prior companynowledge as a proxy to form judgments about the fit betweenhemselves and the organization (Judge and Bretz 1992). Sub-ective fit is assessed through the measure of demands-abilitiest from Cable and Judge (1996) and their measure of subjectiveerson-organization fit (Judge and Cable 1997) (see Appendix).
Credibility covers the elements from initial source credibil-ty research (Hovland and Weiss 1951), and directly linked byable and Turban (2001) to job information sources, namelyxpertise and trustworthiness (e.g., Pornpitakpan 2004); fourtems relate to trustworthiness, information reliability, being anxpert source, and being a credible source.
A CFA produced a satisfactory three factor fit with the dataor the aspirational or mentor tie, fit, and credibility constructsCMIN 92.58 df 51 CMIN/DF 1.82, GFI .899, RMSEA .058.038–.076) p = .239). The AVE for credibility is .57, reliability80; AVE for mentoring or aspirational tie is .57, reliability .80;VE for fit perceptions is .56, reliability .86. The squared corre-
ations (.20, .44 and .45) did not raise significant concerns aboutiscriminant validity.
For OA, the same measurement model as in Study 1roduced a satisfactory three factor fit (CMIN 30.00 df 17MIN/DF = 1.77, RMSEA .056 (.02 –.09) p = .351, GFI .934).gain, a second order model showed no degradation of fit: all
hree primary constructs load well on the second order factor.97; .91; .98).
tudy 2: Analyses and results
Appendix C reports correlations, means and standard devi-tions of the composite study constructs. Regression is lessimiting than SEM when considering a variety of cross-producterms, therefore we deployed Moderated Regression AnalysisMRA) (Sharma, Durand, and Gur-Arie 1981), after mean-
entering of independent variables. Complete mediation isndicated when tie strength or information valence no longerffects OA, after addition of the precommitment variable. Par-ial mediation is when the coefficient reduces in size but is still
r
aa
tailing 89 (1, 2013) 88–104 97
ifferent from zero when the mediator is controlled. A signif-cant contribution from the interaction term without the samerom the moderating variable signifies a pure moderator variable;ut a quasi-moderator if both produce significant contributionsSharma, Durand, and Gur-Arie 1981). Spotlight analyses (onetandard deviation above and below the mean) were used toxamine the nature of the interaction (Cohen et al. 2003; Irwinnd McClelland 2001).
esting Hypotheses 2 and 3Model 1 (see Table 2) reports the results before the addition of
recommitment variables.4 As expected, in all cases, the valencef the information is a strong predictor; together, credibility, paynd conditions information valence explain nearly 22 percentf OA variance in this study, while organizational climate andeputation information and credibility combined explain nearly2 percent.
Supporting H2, climate and reputational information valenceas greater influence on OA (32%) than pay and conditions infor-ation (5%) (Z = 7.45 p < .01). There is a higher contribution toA variance explanation by source credibility perceptions in theay and conditions case (14%) than in the climate and reputationase (7%) (Z = 2.45, p = .014). Thus, H3 is supported.
he effects of precommitmentHypothesis H4a is mainly supported, as the results show a
ignificant two-way interaction between precommitment andnformation valence to affect OA for all except one conditioninterest in the retail role and intangible climate and reputationalnformation valence) (see Table 2, models 2–4). Spotlight analy-es were used to consider the effects of each factor (informationalence and precommitment) at high and low levels of the otherone standard deviation above and below the mean) (Cohen et al.003; Preacher, Rucker, and Hayes 2007).5 The results are in thexpected direction (Table 3); for negative information, the meansor OA are higher when an individual has stronger precommit-ent compared to lower precommitment. This does suggest the
ffects of higher precommitment in mitigating negative infor-ation.Hypothesis 4b is also supported, as all three precommitment
actors show a direct influence on OA (see Table 2, models–4) adding between 5 and 49 percent to the explanation ofariance and resulting in reduction in the information valenceoefficients, though the remaining significant direct contrib-tions of information valance indicate precommitment is aartial mediator (Sharma, Durand, and Gur-Arie 1981). Thus,verall, the results illustrate both moderation and mediation, that
eported here.5 Using an SPSS macro (Preacher, Rucker, and Hayes 2007) updatedt http://www.afhayes.com/spss-sas-and-mplus-macros-and-code.html (lastccessed 20th March 2012).
ediation of the message effect is conditional on the valencef the information.
H4c has some support, with partial mediation of credibilityerceptions by the precommitment factors. The degree of medi-
tion seems strongest for the fit perceptions factor (see Table 2)s the residual credibility coefficients in the fit condition are sig-ificantly lower than in the other two conditions (all p < .5),6
6 See http://psych.unl.edu/psycrs/statpage/regression.html for comparingultiple regression models (last accessed 20th March 2012).
t
i
5.03, p = .026
upporting the distinction between value and outcome rele-ant involvement (Johnson and Eagly 1989). For benefits of aentor/aspirational tie, credibility mediation is weak and source
redibility direct effects remain strong, as expected.Appendix D summarizes the support for the hypotheses in
his study.
Conclusions and implications
Previous research notes a difference in impact on OA betweennformation about company performance and other information
etetoostcple, using real members of staff in advertising and offering
K.A. Keeling et al. / Journal
oncerning pay and conditions (Lievens, Van Hoye, and Anseel007). A contribution of this paper is to confirm this effect (see2), but then develop a theoretical grounding to investigate the
ffects of different modifiers of SWOM on OA as an employer,dding to the current knowledge on how SWOM can alter theiews of prospective employees. A further contribution to theorys the suggestion that both moderation and mediation effects areresent in this complex processing of information; mediationan be explained by the ‘attraction increases liking’ theory, whileiased processing explains the effects of moderation.
The data from both studies confirm previous findings thatositive and negative information have distinct effects on OACollins and Stevens, 2002; Van Hoye and Lievens 2005) andhat relationship strength moderates these effects. The contribu-ion from Study 1 offers a nuanced account, considering effectsredicted from theories of social exchange expectancies and theearch/experience/credence paradigm. For OA, negative orga-izational climate (experience/credence attribute) content fromstrong tie source has a stronger effect (results in a lowerA rating) than the same information from a weak tie source.onversely, when SWOM from a weak tie source contains neg-tive pay and conditions (search attribute) content, it results intronger negative influence on OA than the same informationrom a strong tie. This finding modifies the Brown and Reingen1987) proposition that strong-ties are more influential on deci-ion making than weak-ties, as these influences are moderated bynformation type. In the retail employment context, the subjects well as the valence of the message clearly matter.
The contributions from Study 2 are how job seeker precom-itment may modify the impact of negative information and
redibility for pay and conditions messages (tangible, concretenformation) or organizational climate and reputation messagesintangible information). To summarize, for the baseline modelefore the addition of precommitment factors, the valence ofWOM regarding reputation and organizational climate hasost effect on OA (H2). The effect of adding precommitment
actors shows at least some mediation of the direct effectsf information valence (H4b), consistent with the ‘attractionncreases liking’ premise that precommitment raises the base-ine for OA. Nonetheless, information valence still maintains
significant direct role in prediction and this is the case foroth information types. This demonstrates that staff members areegarded as a valid source of information, especially on hiddenractices in working conditions and employment practices.
Evidence also supports the explanation of motivated reason-ng to maintain pre-existing preferences, even in the presence ofegative information (Ahluwalia, Burnkrant, and Unnava 2000)H4a). Spotlight analyses of interactions show a consistent pat-ern of negative information accompanying reduced OA in lowerrecommitment compared to higher precommitment (Table 3).igher precommitment seems to protect against negative infor-ation effects, as predicted.The mediation of credibility by all precommitment factors, as
ypothesized (Table 2: models 2–4) supports the Dholakia and
ternthal (1977) suggestion that highly involved subjects are
ess sensitive to non-message factors such as source expertise.he strong mediation of credibility when fit perceptions are
cea
tailing 89 (1, 2013) 88–104 99
dded is also consistent with our argument that precommitmentased on fit perceptions can be conceptualized as value relevantnvolvement (Johnson and Eagly 1989) and underlines the valuef understanding reactions to communications in this context.
mplications for recruitment practice
While our empirical work necessarily focused on very spe-ific hypotheses, it is important to consider wider implicationsnd opportunities of SWOM. In extrapolating beyond our empir-cal studies, we were advised by a panel of senior industryxperts from 19 major companies, who commented on thisesearch and debated its wider implications during seven full-ay meetings. Due to the large scale and cost of their recruitmentctivities, major retailers especially seek innovative ways tonhance the size, quality, and diversity of their applicant pool,hile reducing costs and errors in hiring. As retailers are often
mong the largest employers in their regions, their scale offersreat scope to develop the potential of SWOM.
As job seekers regularly approach staff members for companynformation, retailers must ensure a flow of accurate informationo staff about the organization, to improve the company’s appeals a place of work, particularly information regarding the orga-izational climate and reputation. For example, Lichtenstein,etemeyer, and Maxham (2010) note that employees as well as
ustomers see retailers engaging in socially responsible behav-ors as more attractive, and are drawn to see themselves as moreike that retailer. Such considerations could lead to better tar-eting of company messages both to and via different groups,ncreasing chances that prospective employees will perceive aelevant fit between their values and the organization.
Given the findings regarding the influence of mentoring orspirational ties within the linkages between SWOM and OA,mployers have more reasons to target their SWOM promo-ion efforts toward employees representing good role models.hese employees’ positive influence on the OA to those peopleill likely be the strongest. Retailers should also consult and
nclude staff, as practiced and knowledgeable communicatorsbout work issues within their networks, during the develop-ent of communication messages to prospective employees.uch consultations will generate ideas about what job seek-rs find relevant, interesting and thought-provoking for use inraditional or SWOM channels.
Information gained from such staff insight research will ben-fit training; organizations could educate their employees in theypes of information of particular interest to prospective employ-es, such as staff benefits or the social aspects of working forhe company. Helping existing staff be more aware of availablepportunities, such as training, career progression and devel-pment, should lead to increased employee commitment andpontaneous communication of job advantages to people withinheir social networks. Traditional recruitment communicationhannels could also both harness and promote SWOM, for exam-
ontact with staff who can act as mentors to new employ-es. The latter strategy could particularly help in enlarging thepplicant pool, attracting also those job seekers whose more
100 K.A. Keeling et al. / Journal of Retailing 89 (1, 2013) 88–104
for e
re
tmtmctsabnp
temScpiSm
towmitsSgepo
L
otfd
itSfSrSpt
roeproso
trapdoptthr
ttoslra
Fig. 3. Actions suggested
estricted social networks may offer little access to companymployees.
In summary, Fig. 3 suggests steps toward developing an effec-ive SWOM strategy. This needs to extend beyond the realms of
arketing and training to identify and address the root causes inhe organization of positive/negative SWOM, and whether staff
embers are disposed toward enthusiastic brand advocacy. Forompanies to be aware of their SWOM reputation requires theype of research endeavors currently devoted to assessing con-umer WOM, possibly through greater coordination of the staffnd marketing insight functions. Failure to diagnose and if possi-le rectify major sources of employee dissatisfaction will largelyeutralize even the most sophisticated measures to encourageositive SWOM.
Most retail organizations already have motivational practiceso enhance staff morale and productivity, normally requiringffective communications and sometimes extensive internalarketing. The additional objective to encourage positiveWOM and to mitigate potential negative SWOM is a logi-al and relatively inexpensive extension of these processes. Inarticular, it is important to equip present and former staff withnformation that must be sufficiently interesting to encourageWOM conversations, which unlike formal referrals are nototivated by financial gain or specific employment targets.In these endeavors, companies may deploy some of the
echniques of viral marketing (e.g., Hinz et al. 2011), devel-ping messages and probably online content that people willant to share. Analogous to the practice of deploying marketavens (Higie et al. 1987), companies may also target the opin-
on leaders most likely to spread SWOM within and beyondhe organization, based on knowledge of staff members’ socialkills, staff surveys extended to include SWOM activity, or both.WOM effectiveness should also be assessed and the strate-ies further refined. Again, many retailers could integrate thevaluation of SWOM effects at modest cost into their existingrocedures, such as application forms, induction questionnaires,r both.
imitations and future research
This was exploratory research and, as such, could include
nly certain criteria. Not included were differences in theype of information desired between seekers of part-time orull-time work. Furthermore, message order effects need moreetailed research (Van Hoye and Lievens 2005), as does message
Ratw
ffective SWOM strategy.
ntensity and amount, as these aspects are important in other con-exts (Khare, Labrecque, and Asare 2011). Distinctive features ofWOM summarized in Fig. 1 also point to areas of worthwhileurther investigation, including the credibility and valences ofWOM information from current versus former employees, theelative credibility of ERP and SWOM sources, the impact ofWOM compared with ERPs on the diversity of the applicantool, and strategically oriented work on the development poten-ial of SWOM.
These empirical studies considered one crucial stage in theecruitment process. Future research could examine the rolef moderators at other stages in the recruitment process, forxample, after interview, where the job seeker perceptions ofrocedural fairness or perceptions of the interviewer may play aole. Researchers could also extend the fairly constrained rangef jobs used in this study, to examine the nature of the relation-hips between information valence and type of information inther job roles, conditions, and sectors.
SWOM is a powerful, direct influence on job seeker attrac-ion to organizations, which retailers may employ as a valuableecruitment tool with potential to enhance the diversity, sizend quality of the application pool. However, developing theotential of SWOM requires not only spanning different aca-emic disciplines but also crossing functional boundaries withinrganizations, which may involve changes in strategy, personnelolicies, operations practices, internal marketing and informa-ion systems to address salient issues. While full exploration andesting of this potential is beyond the scope of this paper, it isoped that SWOM will attract further attention from researchers,etailers and other marketers.
Acknowledgements
The authors are grateful to the Journal of Retailing editors forheir comments on earlier versions of this paper. We also thankhe three anonymous reviewers for their extensive commitmentf time in helping us to develop this paper through the reviewtages. Their many insightful questions and suggestions haveed to a significant extension of this investigation and extensiveeshaping of this paper. For their funding and insights, we arelso grateful to the directors and senior executives within the
etail Research Forum, who commented on industry relevancend debated implications at several stages in the work. We alsohank the many retail staff members and potential retail recruitsho participated in the two main studies.
T
w
wT
etw
tbp
eaus
ibte
O
ST
K.A. Keeling et al. / Journal of Retailing 89 (1, 2013) 88–104 101
Appendix A. Study 1 scenario options
IE STRENGTHa) CLOSE TIE: This employee is someone you know very well and have extensive contact with, such as a close friend or family member. He/she has been
orking at Company X for 3 years. This person and you share similar values and preferences about work.b) WEAK TIE: This employee is someone you don’t know particularly well and have limited contact with, such as a passing acquaintance. He/she has been
orking at Company X for 3 years. This person and you share similar values and preferences about work.HE INFORMATION GIVEN BY THE EMPLOYEE:SOCIAL CONDITIONS (ORGANIZATIONAL CLIMATE)
NEGATIVE: “Working here can be very boring; we don’t have much to look forward to. They always promise us that they are going to arrange socialvents and staff outings, but I’ve yet to be invited to one! Also, you often find yourself doing more work than you bargained for, there’s just no sense ofeamwork in this place. I barely get to see my manager compared to other staff here. I confronted him recently about this. It ended up as a heated discussion,ith him claiming that I was being oversensitive! Sometimes I feel like I just don’t matter and it puts me off going in to work.”
POSITIVE: “One of the main reasons I love working here is because of the people I work with. The staff here really pull together to get things done andhere is always someone ready to help out. Working here can be quite fun. We organize staff outings and buy each other gifts on special occasions likeirthdays. My manager includes everyone in regular meetings and always takes time out to make sure I’m OK and shows appreciation of hard work. I feel likeart of a team and I actually look forward to going in to work!”
PAY AND CONDITIONSNEGATIVE “The salary here is on the lower end of current market rates. As well, although they say you get a 10% staff discount, you are not actually
ntitled to it for nine months. The same applies for the bonus scheme. Even if you have met all your targets within those first nine months, you don’t get paidnything for it. It’s also difficult to see opportunities to work your way up in the company. They are not very flexible about the hours you work; these aresually planned to fit around their needs rather than yours. If you need to change your hours, they would make you use your holiday allowance rather than findomeone to cover or swap shifts. They always seem to be looking out for their own interests!”
POSITIVE “Compared to other retailers, the pay rates here are actually above the average. Then, of course, there are the added perks of working here,ncluding generous staff discounts and performance based bonuses. The discount is 10% off the marked prices for you and your family. In terms of theonuses, if you meet your monthly targets then you can expect to receive approximately 5% extra. The opportunities are excellent for those looking to workheir way up in the company. Also they are really understanding about flexible working. If you need to change your hours they find someone to cover orxchange shifts. They always seem to be looking out for the interests of their employees.”
Appendix B. Measurement items and descriptive statistics
Mean SD
rganizational Attractiveness (OA) (Highhouse et al. 2003)Your overall impressions of the company now, from what this person has told you (1–7 scale: stronglydisagree–strongly agree)
General attractiveness of the organization as an employerFor me, this company would be a good place to work 6.05 .96The company is attractive to me as a place to work 5.96 .03I would only be interested in this company as a last resort (rev)(removed)
Intentions to pursue a job at the organizationIf this company invited me for a job interview, I would go 6.30 .99I would exert a great deal of effort to get a job with this company 5.98 .51I would recommend this company to a friend looking for a job 5.94 .13
Perceived level of prestige of the organizationEmployees must be proud to say they work for this company 5.79 .21This company must have a reputation as an excellent employer 5.99 .06There are probably many people who would like to work for this company 5.91 .21
ubjective Person-Organization Fit (Cable and Judge 1996; Judge and Cable 1997)hinking about the match between your personal skills and abilities and what you have been told about this job, please
consider the following:My personal abilities and education provide a good match with the demands of this job (1–7 scale; not at
all–completely)6.12 .91
The match is very good between the demands of this job and my personal skills 6.07 .99My abilities and training are a good fit with the requirements of this job 6.06 .84My values, goals, and personality ‘match’ or fit this organization and the current employees in this organization 5.76 .11
The values and ‘personality’ of this organization reflect my ownMy values and personality will prevent me from fitting in this organization because
other employees (removed)
5.78 .19they are different from most of the
1
B
S
C
1234567
H
H
H
H
H
HH
A
A
A
B
B
02 K.A. Keeling et al. / Journal of Retailing 89 (1, 2013) 88–104
Mean SD
enefit of Employee as Aspirational Role Model and Mentor (Dreher and Ash 1990; Ragins and McFarlin 1990)This person can help me understand what the organization is trying to achieve so I can decide whether the job is right
for me5.63 .20
He/she can give me access to resources to help apply for the job 5.55 .38This person is in a position to help me understand how best to apply for the job 5.74 .13This person is a role model for me 5.16 .62
ource Credibility (Cable and Turban 2001; Hovland and Weiss 1951)S/he is an expert in the area they work within 5.70 .24The information is credible since he/she works for the company 5.86 .25What he/she tells me is more reliable than other sources 5.72 .16I can trust what this person tells me about the company and the job 5.92 .12
Appendix C. Correlations, means and standard deviations of Study 2 constructs
onstructs Mean (SD) 1 2 3 4 5 6
Organizational attraction 6.03 (0.78)Valence of pay and conditions information 144.05 (36.44) .184**
Valence of climate and reputation information 157.06 (28.99) .645** .230**
Perception of source credibility 5.84 (0.98) .510** .113 .296**
Perception of person-organization fit 5.98 (0.75) .792** .132* .599** .408**
Perception of benefits of source as mentor/aspirational tie 5.35 (1.27) .449** .055 .228** .500** .338**
Interest in a retail job 5.68 (1.41) .574** .113 .402** .260** .506** .441**
* Significant at <.05.** Significant at <.01.
Appendix D. Support for hypotheses
1a. When SWOM contains more intangible job experience content (experience/credence attributes), negative information from a strong tie source hasa stronger influence on organizational attractiveness (OA) than if from a weak tie source.
1b. When SWOM contains more tangible job related content (search attributes), negative information from a weak tie source has a stronger influenceon OA than if from a strong tie source.
2. The impact of information valence on OA is stronger when considering organizational climate and reputational (intangible, experience/credence)information, compared with pay and conditions (tangible, search attributes) information.
3. The relationship of source credibility to OA is stronger when considering pay and conditions (tangible attribute) information, compared withorganizational climate and reputational information (intangible attribute).
4a. Precommitment factors moderate the effects of information valence on OA (except for the desire for retail role and intangible information valencecase).
4b. Precommitment factors mediate the effects of information valence on OA.4c. Precommitment factors mediate the effects of credibility of information from the staff member on OA; this effect is strongest when
precommitment is based on value relevant involvement.
References
hluwalia, Rohini, Robert E. Burnkrant and Rao H. Unnava (2000), “ConsumerResponse to Negative Publicity: The Moderating Role of Commitment,”Journal of Marketing Research, 37 (2), 203–14.
rndt, Aaron, Todd J. Arnold and Timothy D. Landry (2006), “The Effects ofPolychromic-Orientation upon Retail Employee Satisfaction and Turnover,”Journal of Retailing, 82 (4), 319–30.
rndt, Johan A. (1967), Word of Mouth Advertising, New York: AdvertisingResearch Foundation.
abin, Barry and James S. Boles (1996), “Effects of Perceived Co-worker
Berscheid, Ellen, William Graziano, Thomas Monson and Marshall Dermer(1976), “Outcome Dependency: Attention, Attribution, and Attraction,”Journal of Personality and Social Psychology, 34 (5), 978–89.
Bettencourt, Lance A. and Stephen W. Brown (1997), “Contact Employees:Relationships Among Work-place Fairness, Job Satisfaction and ProsocialService Behaviors,” Journal of Retailing, 73 (l), 39–61.
Bone, Paula F. (1992), “Determinants of Word-of-Mouth Communications dur-ing Product Consumption,” Advances in Consumer Research, 19, 579–83.
Breaugh, James A. (2008), “Employee Recruitment: Current Knowledge andImportant Areas,” Human Resource Management Review, 18 (3), 103–18.
Breaugh, James and Mary Starke (2000), “Research on Employee Recruitment:
Involvement and Supervisor Support on Service Provider Role Stress, Per-formance and Job Satisfaction,” Journal of Retailing, 72 (1), 57–75.
arber, Alison E. (1998), Recruiting Employees: Individual and OrganizationalPerspectives, Thousand Oaks, CA: Sage.
B
so Many Studies, so Many Remaining Questions,” Journal of Management,
26 (3), 405–34.
reaugh, James A., Leslie A. Greising, James W. Taggart and Helen Chen(2003), “The Relationship of Recruiting Sources and Pre-Hire Outcomes:
of Re
B
B
B
B
B
B
C
C
C
C
C
C
C
C
C
C
D
D
F
F
F
F
F
G
G
G
H
H
H
H
H
I
J
J
J
J
J
J
K
K
K
K
L
L
K.A. Keeling et al. / Journal
Examination of Yield Ratios and Applicant Quality,” Journal of AppliedSocial Psychology, 33 (11), 2267–87.
retz, Robert D. Jr. and Timothy A. Judge (1998), “Realistic Job Previews: ATest of the Adverse Self-Selection Hypothesis,” Journal of Applied Psychol-ogy, 83 (2), 330–7.
rown, Timothy A. (2006), Confirmatory Factor Analysis, 1st ed. New York:Guilford Press.
rown, Jacqueline J. and Peter H. Reingen (1987), “Social Ties and WOMReferral Behavior,” Journal of Consumer Research, 14 Dec, 350–62.
ureau of Labor Statistics (2012), “Retail Trade: NAICS 44–45,”(last accessed October 4, 2012), [available at http://www.bls.gov/iag/tgs/iag44-45.htm#workforce].
urgoon, Judee K. and Jerold L. Hale (1988), “Nonverbal Expectancy Vio-lations: Model Elaboration and Application to Immediacy Behaviors,”Communication Monographs, 55 (1), 58–79.
usiness Dictionary (2012), (last accessed September 25, 2012), [available athttp://www.businessdictionary.com/definition/employee-referral-program.html].
able, Daniel M. and Timothy A. Judge (1996), “Person–Organization Fit, JobChoice Decisions, and Organizational Entry,” Organizational Behavior andHuman Decision Processes, 67 (3), 294–311.
able, Daniel M. and Daniel B. Turban (2001), “Establishing the Dimensions,Sources and Value of Job Seekers’ Employer Knowledge during Recruit-ment,” Research in Personnel and Human Resources Management, 20,115–63.
ambridge Dictionary (2012), (last accessed September 25, 2012), [availableat http://dictionary.cambridge.org/dictionary/business-english/employee-referral-programme].
armeli, Abraham, Gershon Gilat and David A. Waldman (2007), “The Roleof Perceived Organizational Performance in Organizational Identification,Adjustment and Job Performance,” Journal of Management Studies, 44 (6),972–92.
hapman, Derek, Krista Uggerslev, Sarah Carroll, Kelly Piasentin and DavidJones (2005), “Applicant Attraction to Organizations and Job Choice: AMeta-Analytic Review of the Correlates of Recruiting Outcomes,” Journalof Applied Psychology, 90 (5), 928–44.
lark, Margaret (1984), “Record Keeping in Two Types of Relationships,” Jour-nal of Personality and Social Psychology, 47 (September), 549–57.
lark, Margaret, Judson Mills and Martha C. Powell (1986), “Keeping Track ofNeeds in Communal and Exchange Relationships,” Journal of Personalityand Social Psychology, 51 (2), 333–8.
oca∼Cola Retail Research Council, New Ideas for Retaining Store-Level Employees, (last accessed October 28, 2012), [available at http://www.employeeretentionconsulting.com/coca-cola-retention-study.pdf].
ohen, Jacob, Patricia Cohen, Stephen Aiken and Leona West (2003), AppliedMultiple Regression/Correlation Analysis for the Behavioral Sciences, 3rded. Mahwah, NJ: Erlbaum.
ollins, Christopher J. and Cynthia K. Stevens (2002), “The Relationshipbetween Early Recruitment Activities and Application Decisions of NewLabor Market Entrants: a Brand Equity Approach to Recruitment,” Journalof Applied Psychology, 87 (6), 1121–33.
holakia, Ruby Roy and Brian Sternthal (1977), “Highly Credible Sources:Persuasive Facilitator or Persuasive Liabilities?,” Journal of ConsumerResearch, 3 (March), 223–32.
reher, George and Ronald A. Ash (1990), “A Comparative Study of Men-toring Among Men and Women in Managerial, Professional and TechnicalPositions,” Journal of Applied Psychology, 75 (5), 539–46.
eldman, Jack M. and John G. Lynch Jr. (1988), “Self-generated Validity andother Effects of Measurement on Belief, Attitude, Intention, and Behavior,”Journal of Applied Psychology, 73 (3), 421–35.
ernandez, Roberto M. and Nancy Weinberg (1997), “Sifting and Sorting: Per-sonal Contacts and Hiring in a Retail Bank,” American Sociological Review,62 (6), 883–902.
inneran, Lisa and Morgan Kelly (2003), “Social Networks and Inequality,”
Journal of Urban Economics, 53 (2), 282–99.
isher, Cynthia D., Daniel R. Ilgen and Wayne D. Hoyer (1979), “Source Cred-ibility, Information Favorability and Job Offer Acceptance,” Academy ofManagement Journal, 22 (1), 94–103.
L
tailing 89 (1, 2013) 88–104 103
renzen, Jonathan K. and Kent Nakamoto (1993), “Structure, Cooperation andthe Flow of Marketing Information,” Journal of Consumer Research, 20 (3),360–75.
raham, Jean, Michael Argyle and Adrian Furnhan (1980), “The Goal Structureof Situations,” European Journal of Social Psychology, 10 (4), 345–66.
ranovetter, Mark (1973), “The Strength of Weak Ties,” American Journal ofSociology, 78 (3), 1360–80.
(1995), Getting a Job: A Study of Contacts and Careers,2nd ed. Cambridge, MA: Harvard University Press.
rewal, Rajdeep, Thomas W. Cline and Anthony Davies (2003), “Early-EntrantAdvantage, Word-of-Mouth Communication, Brand Similarity, and the Con-sumer Decision-Making Process,” Journal of Consumer Psychology, 13 (3),187–97.
ighhouse, Scott, Filip Lievens and Evan F. Sinar (2003), “Measuring Attrac-tion to Organizations,” Educational and Psychological Measurement, 63 (6),986–1001.
igie, Robin A., Linda L. Price and Lawrence F. Feick (1987), “Types andAmount of Word-of-Mouth Communications about Retailers,” Journal ofRetailing, 63 (3), 260–78.
inz, Oliver, Bernd Skiera, Christian Barrot and Jan U. Becker (2011), “Seed-ing Strategies for Viral Marketing: An Empirical Comparison,” Journal ofMarketing, 75 (6), 55–71.
ovland, Carl I. and Walter Weiss (1951), “The Influence of Source Credi-bility on Communication Effectiveness,” Public Opinion Quarterly, 15 (1),635–50.
urst, Jessica and Linda Good (2009), “Generation Y and Career Choice: TheImpact of Retail Career Perceptions, Expectations and Entitlement Percep-tions,” Career Development International, 14 (6), 570–93.
rwin, Julie R. and Gary H. McClelland (2001), “Misleading Heuristics andModerated Multiple Regression Models,” Journal of Marketing Research,38 (1), 100–9.
ain, Shailendra P. and Durairaj Maheswaran (2000), “Motivated Reasoning: ADepth of Processing Perspective,” Journal of Consumer Research, 26 (4),358–71.
ain, Shailendra P. and Steven S. Posavac (2001), “Pre-purchase AttributeVerifiability, Source Credibility, and Persuasion,” Journal of Consumer Psy-chology, 11 (3), 169–80.
ohnson, Blair T. and Alice H. Eagly (1989), “Effects of Involvement on Per-suasion: A Meta-analysis,” Psychological Bulletin, 106 (2), 290–314.
udd, Charles M., David A. Kenny and Gary H. McClennand (2001), “Esti-mating and Testing Mediation and Moderation in Within-Subject Designs,”Psychological Methods, 6 (2), 115–34.
udge, Timothy A. and Robert D. Bretz (1992), “Effects of Work Values on JobChoice Decisions,” Journal of Applied Psychology, 77 (3), 261–71.
udge, Timothy A. and D.M. Cable (1997), “Applicant Personality, Organiza-tional Culture, and Organizational Attraction,” Personnel Psychology, 50 (2),359–94.
hare, Adwait, Lauren I. Labrecque and Anthony K. Asare (2011), “The Assim-ilative and Contrastive Effects of Word of Mouth Volume: ExperimentalExamination of Online Consumer Ratings,” Journal of Retailing, 87 (1),111–26.
leine, Robert E., Susan S. Kleine and J.B. Kernan (1993), “Mundane Con-sumption and the Self: A Social-Identity Perspective,” Journal of ConsumerPsychology, 2 (3), 209–35.
unda, Ziva (1990), “The Case for Motivated Reasoning,” Psychological Bul-letin, 108 (3), 480–98.
evin, Daniel and Robert Cross (2004), “The Strength of Weak Ties You CanTrust: the Mediating Role of Trust in Effective Knowledge Transfer,” Man-agement Science, 50 (11), 1477–90.
ichtenstein, Donald R., Richard G. Netemeyer and James G. Max-ham III (2010), “The Relationships among Manager-, Employee-, andCustomer-Company Identification: Implications for Retail Store Financial
Performance,” Journal of Retailing, 86 (1), 85–93.
ievens, Filip, Greet Van Hoye and Frederik Anseel (2007), “OrganizationalIdentity and Employer Image: towards a Unifying Framework,” BritishJournal of Management, 18 (March), S45–59.
ievens, Filip, Greet Van Hoye and Bert Schreurs (2005), “Examining theRelationship between Employer Knowledge Dimensions and OrganizationalAttractiveness: An Application in a Military Context,” Journal of Occupa-tional and Organizational Psychology, 78 (4), 553–72.
ievens, Filip and Scott Highhouse (2003), “The Relation of Instrumental andSymbolic Attributes to a Company’s Attractiveness as an Employer,” Per-sonnel Psychology, 56 (1), 75–102.
im, Boon C. and Cindy M.Y. Chung (2011), “The Impact of Word-of-mouthCommunication on Attribute Evaluation,” Journal of Business Research, 64(1), 18–23.
ackenzie, Scott B. and Philip M. Podsakoff (2012), “Common Method Biasin Marketing: Causes, Mechanisms, and Procedural Remedies,” Journal ofRetailing, http://dx.doi.org/10.1016/j.jretai.2012.08.001
cColl-Kennedy, Janet, Paul Patterson, Amy Smith and Michael Brady (2009),“Customer Rage Episodes: Emotion, Expression and Behavior,” Journal ofRetailing, 85 (2), 222–37.
alhotra, Naresh, Sung Kim and Ashutosh Patil (2006), “Common MethodVariance in IS Research: A Comparison of Alternative Approaches and aReanalysis of Past Research,” Management Science, 52 (12), 1865–83.
ontgomery, James D. (1991), “Social Networks and Labor-Market Out-comes: Toward an Economic Analysis,” American Economic Review, 81(5), 1408–18.
owday, Richard T., Richard M. Steers and Lyman W. Porter (1979), “The Mea-surement of Organizational Commitment,” Journal of Vocational Behavior,14 (2), 224–47.
arker, Christopher P., Boris B. Baltes, Scott A. Young, et al. (2003), “Rela-tionships between Psychological Climate Perceptions and Work Outcomes:A Meta-Analytic Review,” Journal of Organizational Behavior, 24 (4),389–416.
ersonnel Today (2009), “Employee Referral Schemes Help Net TopTalent,” (last accessed September 13, 2012), [available at http://www.personneltoday.com/Articles/12/02/2009/49388/Employee-referral-schemes-help-net-top-talent.htm].
ornpitakpan, Chanthika (2004), “The Persuasiveness of Source Credibility:A Critical Review of Five Decades’ Evidence,” Journal of Applied SocialPsychology, 34 (2), 243–81.
reacher, Kristopher, Derek D. Rucker and Andrew F. Hayes (2007), “AssessingModerated Mediation Hypotheses: Theory, Methods, and Prescriptions,”Multivariate Behavioral Research, 42 (1), 185–227.
agins, Belle R. and Dean B. McFarlin (1990), “Perceptions of Mentor Rolesin Cross-Gender Mentoring Relationships,” Journal of Vocational Behavior,
hoads, Gary K., William R. Swinyard, Michael D. Geurts and William Price(2002), “Retailing as a Career: a Comparative Study of Marketers,” Journalof Retailing, 78 (1), 71–6.
ossiter, John R. (2002), “The C-OAR-SE Procedure for Scale Developmentin Marketing,” International Journal of Research in Marketing, 19 (4),305–35.
ynes, Sara L. (1991), “Recruitment, Job Choice, and Post-hire Consequences:A Call for New Research Directions,” In Handbook of Industrial andOrganizational Psychology, Vol. 2, Marvin D., Dunnette , Leaetta M. andHough eds. (2nd ed.). Palo Alto, CA: Consulting Psychologists Press, Inc,399–444.
chneider, Benjamin (1987), “The People Make the Place,” Personnel Psychol-ogy, 40 (3), 437–53.
hah, Priti P. (1998), “Who Are Employees’ Social Referents? Using a Net-work Perspective to Determine Referent Others,” Academy of ManagementJournal, 41 (3), 249–68.
kowronski, John J. and Donal E. Carlston (1987), “Social Judgment andSocial Memory: the Role of Cue Diagnosticity in Negativity, Positivity, andExtremity Biases,” Journal of Personality and Social Psychology, 52 (4),689–99.
harma, Subhash, Richard M. Durand and Oded Gur-Arie (1981), “Identificationand Analysis of Moderator Variables,” Journal of Marketing Research, 18(August), 291–300.
ullivan, John (2011), “The Complete List of Employee ReferralProgram Best Practices,” (last accessed September 24, 2012),[ere.net, available at http://www.ere.net/2011/08/15/the-complete-list-of-employee-referral-program-best-practices-part-1-of-2/(and)http://www.ere.net/2011/08/22/the-complete-list-of-employee-referral-program-best-practices-part-2-of-2/]
an Hoye, Greet and Filip Lievens (2005), “Recruitment-Related InformationSources and Organizational Attractiveness: Can Something Be Done AboutNegative Publicity?,” International Journal of Selection and Assessment, 13(3), 179–88.
and (2007), “Social Influenceson Organizational Attractiveness: Investigating If and When Word of MouthMatters,” Journal of Applied Social Psychology, 37 (9), 2024–47.
and (2009), “Tapping theGrapevine: A Closer Look at Word-of-Mouth as a Recruitment Source,”Journal of Applied Psychology, 94 (2), 341–52.
anous, John P. and Adrienne Colella (1989), “Organizational Entry Research:Current Status and Future Directions,” In Research in Personnel and Human
Resource Management, Vol. 7, Ferris Gerald R. and Rowland Kendrith M.eds. Greenwich, CT: JAI Press, 59–120.
ack, Michael H. (1999), “Managing Codified Knowledge,” Sloan ManagementReview, 40 (4), 45–58.
